Wayward Lab guides
Wayward Lab is a free companion for the game Wayward's image generation: custom ComfyUI workflows, model help, image packs for your own characters, and fixes for when pictures go wrong. These are its guides, the same ones that come inside Lab.
Written for Wayward Lab 1.7.6, Settings UI Mod 4.15.22 and the game of 2026-11-09. Settings UI Mod is a separate mod by its own author.
Start here
New since 1.7.1: make an image pack for your own character in Game tools > Custom Image Pack Creator (it was Share Plus). Step by step: CHARACTER-PACK-FOUNDATION.txt. The short editor for a workflow (reference image, character LoRA, steps): REFERENCE-WORKFLOW.txt. Pictures went wrong after a ComfyUI update: FIX-PICTURES-AFTER-AN-UPDATE.txt.
Lab is a companion and guide for Wayward's image generation, mainly for Settings UI Mod (custom workflows, guides, FAQ, packs, reports). Use one image server at a time: Official Gen, Settings UI Mod, or the Custom Image-Gen Server. Lab runs beside the one you choose.
Extract this ZIP into your Wayward game folder (beside index.html). Install Node.js, then double-click start-wayward-lab.bat (Linux and macOS: sh start-wayward-lab.sh). Official Gen also needs Bun; the current Settings UI Mod launcher uses Node. Lab opens at http://127.0.0.1:8799/ Lab keeps its files in wayward-lab-data.
The menu
Four groups: Workflows, Image generation, Help and Game tools. Pages that go together share one entry, with tabs at the top of the page (Guides holds the setup guides, Saves & files holds Backups, Lab storage and Character record).
The new controls in 1.7.1
- Game tools > Image variants & Helper > Image variants: read the Helper status line first. Choose How a scene's pictures are counted, then the scene prompt and image variant choices. Try ‹ ›, + and Next visit on the pretend scene. Off leaves those buttons to the game or mod. Refresh the game page (F5) after a change. The Helper offers Reload now; click its bookmark again after the reload if you use one.
- Each image source > Connection & folders: use Connections for who may use it, its port, and Where ComfyUI is. Press Check, then the card's Save button. Restart the server if asked. Settings UI Mod can offer Restart it now after a host or port change. See README.txt for all the steps.
- The Settings UI Mod shortcuts ask the Helper to open Pippa at that place. Press one, then switch to the game's tab. Nothing is edited or saved. If it does not open, use Image variants, step 3, to add the Helper.
- Game tools > Saves & files > Lab storage > Your settings in one file: Export my settings makes wayward-settings.json. Open a settings file… shows a preview. Tick parts, then Apply the ticked parts…. Folders, addresses, ports, start files, the Helper key, save ids and history are left out. Read the file before sharing it: prompts are your own words.
- Game tools > Saves & files > Lab storage: what Lab keeps and moves, where, and how big. Tidy up moves old leftovers to to-delete; Recently set aside lists every move with Put back. Lab never deletes your files: delete to-delete yourself.
- Custom characters and packs: with the Helper on both game pages (the file and the mod's address), each shows on both title screens. Saves stay separate; Backups > Both versions & room moves them. Pictures an image server drew stay with it: make them a pack to use them everywhere.
- What's new at the top of Lab lists the recent releases and opens their pages. Everything (What's new) opens the full changelog. Its Planned list is not a list of features you can use yet.
Keep the game up to date
Lab follows the latest game. Image packs (Title screen > Image packs) came with game version 2026-11-01; older versions don't have them. Update the game first (newest: 2026-11-09).
Adding an image pack in the game
- Use the downloaded game (the web version can't see your folders).
- Unzip the pack into the packs folder next to the game's index.html, so you have packs\<pack name>\pack.json. If Windows "Extract All" made a second folder of the same name inside, move the inner one up.
- On the title screen press Image packs (under Settings), then Add a pack, and choose that pack.json. A character from the pack is then on the New Game list.
Updating from an older Lab (the safe way)
- Close Lab's window. Wait for model downloads to finish first, or they stop (they carry on next time you press Download).
- Optional but wise: copy the wayward-lab-data folder somewhere safe. It holds Lab's settings, character records and backups. Your pictures, models, saves and the image servers are not in the Lab ZIP, and updating does not touch them.
- Extract the new ZIP into the same game folder, keeping its folders, and choose Replace the files in the destination. Do not delete the old Lab first, and do not extract into a new folder (wayward-lab-data would then be left behind).
- Start start-wayward-lab.bat. If files from an older Lab are still around (such as the old release-posts folder), Lab's page asks whether to move them to the to-delete folder next to Lab. Nothing moves unless you press Move. Then look through to-delete if you like, and delete it. Only folders holding nothing except Lab's own files are offered; anything you added is left where it is, and the Lab window says so.
- If a storage notice appears, choose Move existing Lab files, review any conflicts, then Reload Lab. The same move tool is always under Game tools > Lab storage. If the page looks old, press Ctrl+F5.
Coming from 1.4.9? The big changes since then: Help > Report a problem, the Wayward Lab Helper (Game tools > Image variants & Helper > Wayward Lab Helper), more pictures per scene in packs, downloads that resume, the Settings UI Mod and phone guides, numbered screenshots and languages. What's new (top right) and docs/CHANGELOG.txt list everything, version by version (1.5 to 1.7.1).
Use the section that owns your pictures:
Settings UI Mod - its models, image folder, and live Review screen.
The mod must run for Review and live game sync.
Official Gen - wayward-imagegen's folders, Review and redraw.
Custom Image-Gen Server - the tavern-bridge server's own settings.
Rebuild pack.json - a separate game-index pack.json builder.
Rebuild pack.json if it is missing or its picture mappings are outdated. If the game finds the manifest but not pictures, check the folder location first: avoid packs/packs and match the folder named in pack.json. If the pictures exist in the wrong location, Lab can move or copy them into the game's packs folder using the destination controls. See docs\README.txt for troubleshooting.
ComfyUI is needed for generation and workflow tests. Workflow editing, character records, game-index pack building, and browser save backups are separate tools. See docs\README.txt and docs\EXAMPLES.txt.
Wayward Lab downloads and updates (F95zone): https://f95zone.to/threads/wayward-ai-fanart-support-thread.318124/post-22121764
Setup, mods and resources on F95zone: Wayward Image Generation: Setup, Mods & Resources https://f95zone.to/threads/wayward-ai-fanart-support-thread.318124/post-21850407
This archive contains Lab program files and instructions. The game, image servers, models, pictures, user settings and saves are not included.
GIFs & videos has its own sidebar section with guides for every supported source. Full instructions: docs/GIFS-AND-VIDEOS.txt.
Help sections
Model guide is independent of Custom workflow (docs/MODEL-GUIDE.txt). GIFs & videos starts with worked examples (docs/GIFS-AND-VIDEOS.txt). Common troubleshooting has a searchable FAQ section (docs/FAQ.txt). Still stuck? Help > Report a problem makes a report you read, then copy or post for Lab's author or Settings UI Mod's author. Lab sends nothing by itself.
Backups groups game saves and kept file copies in one place. New automatic backups/replacement copies are opt-in. GIFs & videos now offers a provided workflow, requirements, picture test and result preview without leaving Lab. Lab warns about persistent Settings UI pipeline overrides; review Replace, Rename or Skip before installation. Closing Lab does not remove an installed file.
Per-source video steps with screenshots: Lab > Help > Guides > GIFs & videos The extra Custom workflows video source was removed; the main editor remains.
Find the file-transfer controls again under Game tools > Saves & files > Lab storage. It shows the current folder, checks for older Lab files, and reopens the previous move result even after the startup notice was dismissed. Nothing moves on a check.
Beginner screenshot guides
Help > Guides > Image generation: start here: find compatible files, install/select them and compare one change at a time. Text companion: docs/TRY-MODELS.txt. Help > Guides > Share a pack & workflow: export the actual graph, attach download pages and choose extras to share. Text companion: docs/SHARE-A-PACK.txt. Help > Play on your phone: play on a phone or tablet while your computer draws the pictures. Text companion: docs/PHONE-AND-COMPUTER.txt. Help > Settings UI Mod guide: start the mod, first setup, and where everything is in its menu. Text companion: docs/SETTINGS-UI-MOD-GUIDE.txt.
Troubleshooting & FAQ
In Lab: Help > Troubleshooting & FAQ. Search by symptom or filter by topic.
Buttons under an answer: "See in Lab" opens that page in Lab. "Find in Settings UI Mod" opens the mod's menu (Pippa) in the game's tab and searches for those words (it needs the Wayward Lab Helper on the game page); you can also type the same words in Pippa's Search box yourself.
Getting started
Something broke right after an update. What should I try first?
Updating normally means following each part's own update steps (Lab's is "How do I update Lab safely?"). When something breaks right after an update, leftovers are a common cause: old files, old settings or a second copy get in the way, and removing the new part often doesn't undo it. This goes for the game, Settings UI Mod, Official Gen, Lab, ComfyUI and its node packs alike.
- Find what you updated or installed just before it broke, and change nothing else for now.
- Read that part's window or log for the first error: it often names the cause.
- Back up your saves, custom characters, image packs, workflows and settings (Lab: Backups). Leave the models folder where it is.
- Reinstall only that part fresh, into a new folder, from its official download, not over the old copy, and bring back your data, not its old program files (for ComfyUI: not the old custom_nodes folder or its Python).
- Test with something that worked before. Still broken? Reinstall the next likely part, one at a time.
- Lab: a recovery reinstall keeps wayward-lab-data, which holds all your Lab work; copy it into the new folder.
See in Lab: Backups; Fix pictures after an update > 4. Clean reinstall; Report a problem
What do I need to make pictures, and do I need Lab?
Three parts work together: ComfyUI draws the pictures, an image server passes the game's requests to it, and the game shows them. Lab is optional: a workshop and guide beside them.
- Easiest: Official Gen, the image server that comes with the game, set up with the installer by Anongatari (install-imagegen.bat > Install). It installs ComfyUI, the node packs and about 11 GB of models, and makes start-wayward.bat, which starts everything. Windows 10/11, an NVIDIA RTX 20 or newer with 8 GB, about 20 GB free (install-imagegen-amd.bat for AMD is experimental). https://f95zone.to/threads/wayward-v2026-11-09-sintention.310923/post-22034259
- More control: Settings UI Mod as the image server, with Lab beside it for workflows, models, packs and guides.
- Run one image server at a time.
See in Lab: Image generation: start here
In what order do I start everything?
- ComfyUI first; wait until it has finished loading.
- Then one image server: Official Gen, Settings UI Mod or the Custom Image-Gen Server.
- Then the game: index.html for Official Gen, or the address the Settings UI Mod window shows (normally http://127.0.0.1:8788/).
Lab can start at any time; the game doesn't need it to play. Without ComfyUI the game still runs and shows the pictures you already have; it only can't draw new ones.
See in Lab: Image generation: start here
Pictures are very slow, fail at the very end, or ComfyUI won't start. What computer do I need?
A graphics card with enough memory: 8 GB or more is comfortable, 6 GB works but is slower. As a rough example, an RTX 3060 with 12 GB draws an Illustrious picture at the game's size (832 x 1216, about 25 steps) in roughly 10 to 20 seconds; your model, size and steps change that a lot. NVIDIA is the easiest; recent AMD and Intel Arc cards work with ComfyUI's builds for them. Built-in laptop graphics are usually not enough.
- Always read the last lines of the ComfyUI window first: the real error is there, not in the game.
- Fails at the end ("time remaining 1s", then an error): the picture is decoded and saved at the end, so it's usually the graphics card running out of memory, or a missing or wrong VAE. Try a smaller picture size, turn off hi-res fix or a second pass, and close other programs that use the card.
- "Timed out waiting for the render": first check whether ComfyUI is still drawing (its window, or Lab's Generation monitor): a slow card or a big workflow can simply take longer than the image server waits. If it shows an error, stopped, or was closed, restart it. The first picture after starting is slower because the model loads.
- Slow: a Lightning or Turbo LoRA with its own low steps and CFG helps a lot.
- ComfyUI won't start, or closes with 0xC0000005: update the graphics driver and restart the PC.
See in Lab: Generation monitor; Report a problem
Does Lab work on Linux or macOS?
Yes. Lab needs Node.js: start it with sh start-wayward-lab.sh in the game folder. The image servers' start files it makes there are .sh files too. Official Gen needs Bun and Settings UI Mod needs Node.js, as on Windows. The installer by Anongatari and the mods' .bat and .cmd files are Windows-only: on Linux install ComfyUI yourself and use Lab's guides; a mod's build.cmd only runs node mod/build.mjs, so run that.
See in Lab: Image generation: start here
Can I play, or use Lab, on a phone or tablet?
Playing works in any phone browser; Lab and drawing pictures need the computer. The usual way: the computer serves the game and draws, the phone plays over your home Wi-Fi. Lab > Help > Guides > Play on your phone has every way step by step (docs\PHONE-AND-COMPUTER.txt), including Android without a computer (JoiPlay) and iPhone.
- The phone keeps its own saves, separate from the computer's.
- Only open servers to every device (--host 0.0.0.0) on a home network you trust.
See in Lab: Play on your phone; Play on your phone > Which way to play
The game on my phone can't reach the image server.
- Start the server so other devices may use it: Official Gen with bun src/cli.ts --host 0.0.0.0, Settings UI Mod with --host 0.0.0.0.
- Open the game on the phone with http://, not https:// (an https page can't reach an http server).
- Allow the server (bun, or node for the mod) through the computer's firewall on private networks.
- Lab's Connections card for that server checks it and shows this computer's address to use.
See in Lab: Play on your phone
Settings UI Mod
Where do I do the basic things in Settings UI Mod?
Everything is in its menu: the Pippa button at the bottom right of the game (play at the mod's address, normally http://127.0.0.1:8788/). Its Search box at the top finds any setting by name, which is quicker than learning the menu; the buttons below search for you. Changes only count after Save & Apply to game at the bottom of the menu.
- Connect to ComfyUI and set the models folder: ComfyUI.
- Pick the model, LoRAs, steps and CFG for her, and test them: model family; Test this pipeline.
- How she looks and what she wears: Character prompts; Clothing.
- What a scene says: Scene prompts.
- Use an image pack, and in what order packs, saved pictures and new drawings are used: Where her pictures come from.
- Draw her pictures ahead of time, review them, and make her image pack: Pictures ahead of time.
- Pictures you kept with the thumbs up: Browse saved pictures.
- Remove a word from every prompt: Tag rules.
- Back up the mod or move it to another computer: Backup & move.
Find in Settings UI Mod: ComfyUI; model family; Character prompts; Scene prompts; Where her pictures come from; Pictures ahead of time; Browse saved pictures; Tag rules; Backup & move
See in Lab: Settings UI Mod guide > Where is it?
How do I start Settings UI Mod, and what are its buttons in the game?
Start ComfyUI first, then double-click Start-Wayward-Bridge.bat in the game folder (it needs Node.js 18 or newer) and play at the address its window shows. Keep that window open while you play. In a game:
- Pippa (bottom right): the settings menu: pictures, prompts, models, animation, packs, backups.
- Mara (above Pippa): the Workshop, for making new things for the game.
- Elena (bottom left): Daily Happenings, her mind and life; a switch at its top turns all of it off if you only want the pictures.
The names are fixed labels, whoever you play. The game's own Settings button (top right) is the game's, not the mod's.
See in Lab: Settings UI Mod guide > Start it; Settings UI Mod guide > The mod's three buttons
Settings UI Mod is running, but the game doesn't use it (or only Official Gen's pictures show).
- Open the game through the mod's address (http://127.0.0.1:8788/ unless it was started with --port), not by double-clicking index.html.
- Close Official Gen. Both answer the game on port 8189, and whichever started first wins. The mod's window says "port 8189 is in use" when Official Gen holds it.
- To keep Official Gen's pictures, make them a pack first (Lab: Rebuild pack.json) and add it in the mod.
Find in Settings UI Mod: Where her pictures come from
See in Lab: Settings UI Mod guide > Start it; Rebuild pack.json
Settings UI Mod keeps losing its connection, or won't connect, after an update.
- Do a fresh install of the mod (a new folder, not over the old one), and of the game if you updated it too; keep wayward-bridge-data so your settings and pictures stay.
- Use the newest Lab and reload the game page: older Helpers asked the mod for its whole picture list every few seconds while a picture was drawing, which slowed it down more the longer you played.
- Run one image server only, and check the mod's window for errors when it drops.
- Still happening? Make the mod's own bug report (next question): it shows its connection to ComfyUI and recent errors.
See in Lab: Report a problem
How do I report a bug in Settings UI Mod?
Update the mod first; it may already be fixed. Then in the game: Elena button > In this game > "Something wrong with the mod?" > Copy a bug report for the mod, and post it in the mod's support thread. It holds versions, ComfyUI's status and recent errors, without folder paths. Lab's Report a problem adds the same report by itself while the mod runs. Don't post screenshots of the mod's Diagnostics page: it shows your full folder paths.
See in Lab: Settings UI Mod guide > A bug report; Report a problem
I changed something in the mod's menu and nothing happened.
- Press Save & Apply to game at the bottom of the menu; "1 unsaved change" at the bottom left means you haven't. Tick "redraw current picture" next to it to see the change at once.
- A saved picture or an image pack may be shown instead of a new drawing: Where her pictures come from shows the order.
- You may be editing another heroine or another save: check HEROINE at the top of the menu and the save ID under it. Prompts are kept per save, so a new game starts from her latest settings.
Find in Settings UI Mod: Where her pictures come from; Save ID
Pictures fail with "Value not in list: unet_name" (or the Qwen CLIP), but I use another model. The mod keeps models per heroine, and a heroine that was never set up (often a character made in the game) uses its default family, Anima. Open Pippa, check HEROINE at the top is her, set her model family and checkpoint, press Test this pipeline, then Save & Apply to game. A settings profile makes it quick for the next one. Find in Settings UI Mod: model family; Test this pipeline; Settings profiles
Saving says "Not saved: Invalid value for ...". A value is over the mod's limits: as of 4.15.19, a scene holds at most 32 variants, one prompt at most 8,000 characters, a character's name at most 60. The name after "for" says which setting: saves.<save>... is that save's own settings, custom-... a custom character. Use fewer variants or a shorter prompt, then save again. If it's within the limits, send the mod's bug report with the exact message.
See in Lab: Settings UI Mod guide > A bug report
The game window looks rearranged since I updated the mod, or I can't find Screen layout. Newer mods arrange the game in panels. Click the ▦ Screen layout button on the game page and pick the game's own layout (or A, which keeps the game's columns), or drag the borders. It's kept for each window size, so resizing, fullscreen or browser zoom can show a different one.
See in Lab: Settings UI Mod guide > The buttons on the game's picture
A word keeps showing up in my pictures and I can't find where it comes from. Press inspect in the menu's "In game" strip: the Prompt inspector shows every part of that picture's prompt and where each came from (her looks, outfit, scene, mood, the game). Change it at its source, or remove it everywhere with a tag rule (Find: the word, Replace: empty). Find in Settings UI Mod: Tag rules
How do I keep her face the same in every picture?
Use her character LoRA, and/or the mod's face anchor (Her face: use the picture on screen, strength about 0.7). The face anchor needs ComfyUI's IPAdapter FaceID nodes; without them it's skipped. Find in Settings UI Mod: LoRA models; Her face
See in Lab: Settings UI Mod guide > Extras
Can I change how the men and patrons look?
Yes. Patron portraits holds the tags for each kind of patron's round portrait. Scene participants sets who is with her in scene pictures: start with how many besides her (1boy, 2boys, 1girl...) and describe them; your count replaces the game's. Named patrons in the tavern gives one man his own prompt and a face that stays the same. With several people in one picture, models sometimes mix up who is who: naming each person's traits helps, and expect to redraw now and then. Find in Settings UI Mod: patron portraits; Scene participants; Named patrons
See in Lab: Settings UI Mod guide > Can the mod do it?
Does the mod or Lab install the extras (face fix, face anchor, smoother GIFs)?
Neither installs ComfyUI node packs: use ComfyUI Manager (Custom Nodes Manager > search > Install), then restart ComfyUI. Lab's Extras checklist asks ComfyUI what it has and lists, for each feature, what is missing. Lab's GIFs & videos > Check requirements downloads the model files for animation.
See in Lab: Settings UI Mod guide > Extras
How do I back up the mod's settings, or move them to another computer?
Backup & move > Export everything makes one file with all settings, presets and add-ons (not the pictures); Import it on the other computer, and copy wayward-bridge-data too if you want the pictures. The mod also keeps copies of your game saves every 10 minutes, plus a weekly zip. Find in Settings UI Mod: Backup & move; Copies of your save
Which menu should I use, the old one or the new one?
Either: both change the same settings, and Try the new menu / Back to the old menu switches any time. Older guides use the old menu's names; the Search box finds a setting in both, so you don't need the names.
See in Lab: Settings UI Mod guide > Two menus
What is the Workshop for?
Making new things for the game and sharing them as projects: events, visual novels, shows, rooms and places, characters and kinds of patron, clothes, items and shops, fights and spells, even a theme. Open it with the Mara button. New to it? Projects & sharing > Make me a project... builds one for you, and the Glossary explains the words.
See in Lab: Settings UI Mod guide > The Workshop Workspace
Lab says the mod's settings changed since the panel loaded. Something saved the mod's settings after Lab read them (usually you, in the mod's menu). Lab saves nothing then, so it can't overwrite them. Reload the panel in Lab, check the values, and save again.
A "Find in Settings UI Mod" or "Show it" button does nothing. Check these, then press it again and read the message beside the button:
- The game is open at the mod's address (http://127.0.0.1:8788/), not as index.html in another tab.
- The Wayward Lab Helper is on that page (Lab: Image variants, step 3). Using the bookmark? Click it again after every reload of the game page.
- Close the mod's Workshop if it's open; the message says when that's needed.
- Switch to the game's tab after pressing: the menu opens there, not in Lab.
You can always type the same words in Pippa's Search box yourself.
See in Lab: Image variants > Put the Helper on the game page
What do the "Show it" and "Find in Settings UI Mod" buttons in Lab do?
They ask the Wayward Lab Helper on the game page to open Pippa: at a page, or with words typed in its Search box so the mod's own results show where a setting is. Press one, then switch to the game's tab (the game must be open at the mod's address with the Helper on it). Nothing is changed or saved. If it can't, the message beside the button says why; you can always type the same words in Pippa's Search box yourself.
See in Lab: Image variants > Put the Helper on the game page
Which Settings UI Mod version does Lab work with?
Lab 1.7.6 is checked with Settings UI Mod 4.15.22 and works with 4.15.2 and later. Lab only changes the settings it shows and leaves the rest as you set them. The mod's own changelog is in its menu.
Official Gen
Official Gen says connected, but no new pictures appear. In the wayward-imagegen folder, run bun src/cli.ts doctor. It checks that ComfyUI answers, that its node packs are installed (Impact Pack, Inspire Pack, KJ Nodes), that the model file names match, and draws one test picture; each failure says what to change.
- "ComfyUI is not answering at http://127.0.0.1:8188": ComfyUI isn't running, crashed, or uses another address (ComfyUI Desktop often uses 8000). Give Official Gen the address ComfyUI shows: Lab > Official Gen > Connection & limits, or bun src/cli.ts --comfy-url <address>.
- All passes? In the game, Settings > Image generation must be Automatic (or Specific address pointing at it).
See in Lab: Official Gen
Official Gen stops with "too many queued requests", "hourly image limit reached" or "GB ceiling". Those are its safety limits: 64 pictures waiting, 400 pictures an hour, and an optional size limit for its picture folder. They reset by themselves (the hourly one within an hour). Making pictures ahead of time on a fast card can pass 400 an hour. To change them: Lab > Official Gen > Connection & limits, save, restart Official Gen.
See in Lab: Official Gen
The game says "No pictures yet" for a character, but Official Gen has lots of her pictures. The game only counts pictures listed in Official Gen's index (images\.image-cache.json), not the files in the folder. If its window says "[cache] loaded N entries" with N far too low, or "unreadable index, starting fresh", the index was lost (often Official Gen closed while saving it). The pictures are still on disk:
- Make a pack from them: Lab's Rebuild pack.json on her folder (tick "review extra pictures of a scene to add as versions" to keep more than one per scene), then add it in the game.
- If Official Gen started after the game page, reload the game page.
- Back up .image-cache.json now and then; Official Gen keeps no copy.
See in Lab: Rebuild pack.json
After the thumbs down (Mark as bad and regenerate), the old picture comes back. A known game issue: the game makes its list of usable pictures once, when the page opens, so it still finds the old one. Save the game and reload the page (F5) after a thumbs down. If the old picture comes from an image pack you added, the thumbs down can't remove it: rebuild the pack without it, or remove the pack, under Title screen > Image packs.
I changed Official Gen's picture folder and my pictures disappeared. Only new pictures go to the new folder; the old ones stay where they were. Restart Official Gen after changing it. To use the old ones, change the folder back, or move the whole old folder (with .image-cache.json) to the new place.
Official Gen's Review page shows nothing or doesn't respond. A known issue with Official Gen's own review in this release. Use the game's thumbs down, or Lab's Pack review, until it's fixed. Review also needs Official Gen's index and saved review data, so a folder of loose pictures can't be reviewed there.
See in Lab: Pack review
Connections & servers
Can I run two image servers at once?
No, one at a time. Official Gen and Settings UI Mod both answer the game on port 8189, and Settings UI Mod and the Custom Image-Gen Server both use 8788. Stop one before starting the other; ComfyUI can keep running. "Port 8189 is already in use" usually means a second server is still open in a minimised or forgotten window. Lab finds every image server running on your computer, shows which one the game uses, and offers Stop this one... for the others.
See in Lab: Report a problem
ComfyUI is connected, but nothing draws. ComfyUI only draws; an image server (Official Gen, Settings UI Mod or the Custom Image-Gen Server) takes the game's requests and sends them to it. Start ComfyUI and one image server, then open the game the way that server needs. Lab doesn't draw for the game by itself.
See in Lab: Image generation: start here
Is ComfyUI stuck, or just slow?
Lab's Generation monitor (or click ComfyUI at the top of Lab) shows what ComfyUI is drawing, how many wait, the usual time per picture, graphics memory and recent errors, and says when it looks stuck. Then check ComfyUI's window for an error, press Cancel there, or restart it.
See in Lab: Generation monitor
Can ComfyUI run on another computer?
Yes. On that computer, start ComfyUI with --listen and allow port 8188 through its firewall; check by opening http://<its address>:8188 from the game computer (the address is in ipconfig there, IPv4 Address). Then give that whole address to your image server: Settings UI Mod's ComfyUI settings, Official Gen's Connection & limits in Lab. Scanning models and installing node packs need ComfyUI's own folder, so do those on that computer. Find in Settings UI Mod: ComfyUI
See in Lab: Generation monitor
Should I play from the file (index.html) or through Settings UI Mod's address?
It depends on your image server: Settings UI Mod only works when you play at its address (normally http://127.0.0.1:8788/); the plain game and Official Gen work from the file. Either way, each address is a separate place in your browser with its own saves, packs and custom characters, so switching looks like starting over. Pick one. To move between them:
- Saves: Lab > Backups > Both versions & room copies saves from one to the other (it keeps a copy first), or download one and use Load Game > Import from JSON.
- Packs and custom characters: with Lab running and the Wayward Lab Helper on both game pages, Lab shares them for you (reload when the page says so). Without it, add them again at the other address.
- Pictures an image server drew belong to that server: make them a pack to have them in both.
The same goes for localhost and 127.0.0.1: they are different addresses to the browser.
See in Lab: Backups; Image variants > Put the Helper on the game page
A server window says Node.js is missing, or shows a Node error. Install Node.js 18 or newer from nodejs.org and restart the PC. Start each server with its .bat file; double-clicking a .js file doesn't run it. Don't open and save .js files in Notepad, which can break them.
Saving settings fails with "EPERM: operation not permitted". Windows is blocking the server from writing to its folder. Check the folder isn't read-only (right-click > Properties), that your antivirus isn't blocking it, and keep the game out of protected places like Program Files. Running the .bat as administrator is a last resort.
Mods stopped working after a game update. Mods are made for specific game versions, so a game update can break them until the mod is updated too. Check the mod's post for a new version, or keep the previous game version until then; and install the new versions fresh (see the first question in Getting started).
The browser console shows a red CORS error about manifest.webmanifest. Harmless. When the game is opened as a file, it can't load the file used to install it as an app; pictures, saves and Lab aren't affected. Lab can take that line out of the game page (Wayward Lab Helper > Playing from a file > Take the line out; it keeps a copy). It doesn't fix anything else.
See in Lab: Wayward Lab Helper > Playing from a file
What do Check and Save on a Connections card do?
Check tests whether ComfyUI answers at that address; it saves and draws nothing. Save writes the choices into that image server. A changed port or "who may use it" takes effect when the server starts again (Lab offers Restart it now for the running mod), then reload the game at its address. Show this computer's address shows your home-network address only when you press it.
See in Lab: Settings UI Mod; Official Gen; Custom Image-Gen Server
Models & nodes
Where do I change the model, LoRAs, steps and CFG?
In your image server, not in Lab.
- Settings UI Mod: Pippa, per heroine (check HEROINE at the top of the menu): model family; Model / sampler presets (the model files, sampler, steps and CFG; blank uses the family's default); LoRA models; Turbo (steps / CFG override), which replaces steps and CFG while it's on. Then Test this pipeline and Save & Apply to game. This is the place to set them; Lab reads them from there.
- Official Gen: Lab > Official Gen, which edits Official Gen's settings file; restart Official Gen after saving.
- A custom workflow: its own settings, in Lab's Custom workflow (see Workflows).
Lab's Model guide explains which files go together; its Downloads lists put missing files in the right ComfyUI folder. Find in Settings UI Mod: model family; sampler presets; LoRA models; Turbo; Test this pipeline
See in Lab: Model guide; Official Gen
I changed steps or CFG in Settings UI Mod, but pictures still use different values.
- Check HEROINE at the top of the menu: settings are per heroine, so you may have changed another one.
- Press Save & Apply to game: changes only count after it.
- Turbo (steps / CFG override) on? Its values replace the others.
- Pictures already waiting to be drawn keep the values they were sent with; the next new picture shows the change.
- A custom workflow ticked in Add-ons? It draws with its own values (except those it has placeholders for): change them in Lab's Custom workflow, or untick it.
Find in Settings UI Mod: Turbo; sampler presets; Add-ons
See in Lab: Custom workflow
What do checkpoint, UNET, CLIP, VAE and LoRA mean, and which go together?
They are different parts of one model, even when they all end in .safetensors: the checkpoint (or UNET/diffusion model) draws, CLIP reads the prompt, the VAE turns the result into a picture, and a LoRA adds a style or a character on top. Parts must be made for the same family (Anima with Anima, Illustrious with Illustrious): a mix that doesn't fit gives errors or bad pictures, and adding more files can't fix it. Lab's Model guide shows which files each family needs.
See in Lab: Model guide
How do I install a missing custom node (node pack)?
- ComfyUI Manager (easiest): Manager > Custom Nodes Manager, search the pack's name, Install, restart ComfyUI. Not listed? Install via Git URL with its GitHub address.
- By hand: cd into ComfyUI\custom_nodes, git clone <its GitHub address>, then install its requirements with ComfyUI's own Python (portable ComfyUI: ..\..\python_embeded\python.exe -s -m pip install -r <folder>\requirements.txt; ComfyUI Desktop: in its Terminal panel). Lab shows these lines ready to copy for the pack you need.
- Then restart ComfyUI and read its window: IMPORT FAILED next to a pack means a requirement is missing. "requirements failed" often means Git for Windows isn't installed.
Lab finds which pack a missing node comes from, and shows How to install for each.
See in Lab: Custom workflow
I downloaded a model in my browser. Does Lab put it in the right folder?
Lab's own Download into ComfyUI saves straight into the right folder. For a file you downloaded yourself, Lab looks in your Downloads folder by its name and offers Move it into ComfyUI; it moves only when you press it. A file saved under another name: use Choose file & install on its row.
See in Lab: Custom workflow
A model download stopped, looks stuck, or left a .part file. A .part file is an unfinished download (the connection dropped, Lab's window closed, the PC slept or the disk filled). Press Download (or Try again) and Lab carries on where it stopped. Clicking inside Lab's black window can pause it (Windows text selection): press Esc. A .part file from another program can't be carried on: delete it and download again.
Pictures come out blurry, crunchy, burned or with odd colours. Usually the picture is "overcooked" or the parts don't fit together. Check, in this order:
- Steps and CFG as the model's page says. Lightning and Turbo LoRAs need their own low values; burned or tinted colours usually mean CFG is too high.
- Every LoRA made for the same family as the model.
- Nothing doing the same job twice: two adapters, the same LoRA added twice, too many LoRAs.
- Speed-up nodes (TeaCache and the like) and detail passes off, to compare.
- To find the cause: keep prompt, seed, size and model the same, turn extras off, then back on one at a time (Lab's A/B compare does this side by side).
- It started right after an update or install: see "Something broke right after an update" in Getting started.
- Official Gen changes steps and CFG per picture by itself: Lab > Official Gen has an override.
Find in Settings UI Mod: model family; Turbo
See in Lab: A/B compare; Fix pictures after an update > 1. Is it really the update
Pictures went bad after I installed or updated a node pack. Check these, in this order:
- ComfyUI's window when it starts: IMPORT FAILED next to a pack, or errors about a Python package, mean an install went wrong.
- Two packs doing the same job side by side (for example two IPAdapter packs after pressing the wrong Install): in ComfyUI Manager > Custom Nodes Manager, show Installed, uninstall the one you didn't mean to install, and restart.
- A pack updated to a version your workflow doesn't fit (nodes renamed or settings changed): its page says what changed; update the workflow or install the earlier version.
- An update that changed a shared Python package other packs need: the pack's page or issues usually say so.
If it's still wrong after that, uninstalling wasn't enough (packs leave Python packages behind): reinstall ComfyUI clean, as the guide below shows, and add back only the packs your workflow needs, one at a time.
See in Lab: Fix pictures after an update > 3. Look for two copies; Fix pictures after an update > 4. Clean reinstall
My workflow uses SDXL IPAdapter FaceID Plus v2. Which files go where?
First check what your workflow loads: the IPAdapter loader node names the file it wants. This list is only for the SDXL FaceID Plus v2 variant; other variants need other files (link below), and workflows without IPAdapter need none. Inside your ComfyUI models folder (or your shared models folder), create the folder if it's missing:
- clip_vision: CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors
- ipadapter: ip-adapter-faceid-plusv2_sdxl.bin
- loras: ip-adapter-faceid-plusv2_sdxl_lora.safetensors (only here; the loader finds it)
FaceID also needs InsightFace in ComfyUI's Python. Other variants and exact names: https://github.com/cubiq/ComfyUI_IPAdapter_plus#installation
See in Lab: Fix pictures after an update > 5. Put the model files; Model guide
What should I share when asking for help with bad pictures?
Enough for someone else to rebuild it; "default everything except the checkpoint" isn't enough:
- the workflow itself (exported JSON), a bad picture, and a good one from before if you have one
- every model file it loads (checkpoint or diffusion model, VAE, CLIP, adapters, LoRAs with strengths)
- sampler, scheduler, steps, CFG, size and seed; the errors in ComfyUI's window
- what you changed just before it went wrong
Lab's Report a problem collects most of it for you. Take out private prompts and folder paths with your name before posting.
See in Lab: Report a problem
Trigger words don't change my pictures. Typing a LoRA's trigger words does nothing unless that LoRA is also loaded: check it's on, its strength, and that it's made for your model's family. In Settings UI Mod, which LoRAs are used is set per heroine, but a LoRA's trigger words are shared by every heroine that uses it. Find in Settings UI Mod: LoRA models
See in Lab: Model guide
Lab says my model's family is "not known", or a LoRA is for another family. Lab reads a file's family (Anima, Illustrious, Pony, NoobAI) from its name or its data. A LoRA for another family does little or gives odd colours and faces; Illustrious, Pony and NoobAI are close cousins, so between them it's only a note. If Lab guessed wrong, pick the right family under the check; it's kept for good.
See in Lab: Model guide
Is there an easier way to manage my LoRAs?
LoRA Manager (by willmiao) is an optional ComfyUI add-on: LoRAs with preview pictures, downloads from Civitai, trigger words. Install it with ComfyUI Manager. Lab reads what it saves, so its trigger words and pictures show in Lab's lists. Lab and the image servers work without it.
Workflows & exports
When do I edit things in Lab instead of the mod?
For a custom workflow. Settings UI Mod has proper settings for its own pipeline (models, LoRAs, steps, sizes), so change those there. A workflow you bring yourself has no editor in the mod, and every workflow is different, so Lab's Custom workflow shows any workflow as a short, editable list: models and LoRAs, prompts, steps and CFG, size, reference pictures. Save it there, then switch it on in the mod.
See in Lab: Custom workflow
How do I use my own ComfyUI workflow in the game?
- In ComfyUI, export it with Workflow > Export (API). The normal Save format can't be used.
- In Lab: Custom workflow > open it. Lab shows its settings in groups.
- Open "Advanced tools · nodes, JSON & game connection" and press Templatize: it puts placeholders where the game's values go. Check that the prompt, the seed and the size went to the right nodes (Placeholders lists them), read the health notes, then Save to library.
- In the same section, Image server tab > In the game > Add it to the game's menu, and choose whether it draws scenes, portraits or moving pictures. Tick "the game draws with it now" to switch it on.
- In the game it's a Custom workflow tick under the mod's Add-ons. Lab, Settings UI Mod and ComfyUI all have to be running.
- Changed the workflow later? Save to library again updates the one the game uses. Untick it to go back to the mod's own pipeline.
Find in Settings UI Mod: Add-ons
See in Lab: Custom workflow; Fix pictures after an update > 7. Settings UI Mod
My custom workflow isn't used in the game.
- Is it ticked in the mod's Add-ons? Each imported version has its own tick: untick the old ones.
- No "In the game" section in Lab? Start Settings UI Mod, then Refresh active source in Lab's Image server tab.
- Not in the mod's Add-ons? Add it to the game's menu again from Lab, with the same mod running.
- Lab, the mod and ComfyUI must all be running while you play.
Find in Settings UI Mod: Add-ons
See in Lab: Custom workflow
What are placeholders like {{positive}} or {{seed}}?
Spots in your workflow where the image server writes the game's values for every picture: the prompt, the seed, the size, sampling, models, LoRAs, the file name, and for moving pictures the start picture, length and frame rate. Custom workflow > Placeholders lists them, shows where yours uses them, and Put here / Take out (which puts the old value back). Anything without a placeholder keeps the workflow's own value. For several LoRAs use {{loras}} on one LoRA node: it becomes the whole list, in whatever loader you use (including LoRA Manager's and rgthree's).
See in Lab: Custom workflow
My Lab test picture looks different from the game's. A test copies the game's prompt, not everything. Pictures only match when prompt, seed, size, workflow and models are all the same: pick a recent game prompt with its seed, and check the workflow uses the same models as the game.
See in Lab: Custom workflow
A workflow works in the mod but shows missing nodes in ComfyUI. The mod can swap in loaders it has (GGUF, plain LoRA loaders) while it draws; the saved workflow still names the original nodes. To run it in ComfyUI itself, install those nodes or replace them, and fill in every {{placeholder}} first. Export last-used API workflow (Lab's Settings UI Mod page) gives the exact workflow of the last picture, with real values.
See in Lab: Settings UI Mod
Importing a shared workflow ZIP says "damaged" or "No workflow found".
- Damaged: the download was cut short or changed on the way; download it again.
- No workflow found: the ZIP has no API workflow, usually because it was saved with ComfyUI's normal Save, or it holds only pictures. Ask for an Export (API) copy.
- "Set the ComfyUI folder first": Lab needs your ComfyUI folder before it can install files from the ZIP.
See in Lab: Custom workflow
F95 won't let me attach my workflow .json. Rename it to .txt before attaching and tell people to rename it back, or put it in a .zip with the pack.
GIFs & videos
How do I make moving pictures?
Settings UI Mod animates pictures in the game (its Animation settings); Official Gen does it with a video workflow. Lab's GIFs & videos guide checks what ComfyUI needs (node packs and the WAN model files), downloads the files, and can make a test clip from a picture first. Find in Settings UI Mod: What gets animated
See in Lab: Settings UI Mod guide > Extras
My GIF doesn't move, or a "GIF" came out as .webp. Settings UI Mod saves animated WebP by default (more colours, smaller), so a .webp is expected; choose GIF in its Animation settings if you need a .gif. Renaming a file doesn't convert it. If it doesn't move, open the file itself in a browser (some picture viewers and thumbnails show only the first frame). Still not moving there? It came out as one frame: check the mod's animation queue for an error. Find in Settings UI Mod: file and loop; Animation queue
The video button is missing, or MP4 isn't offered.
- Official Gen shows a video button only when started with a video workflow (--wan-workflow <file>, or wanWorkflowPath in its config); then press Check again in the game's image settings. A "missing Wan node" error means the workflow doesn't match what Official Gen expects: use the ready one from Lab's GIFs & videos guide.
- MP4 needs VideoHelperSuite and ffmpeg in ComfyUI.
- Check requirements only checks; it doesn't make anything.
Pictures in the game
The game keeps drawing new pictures instead of using my image pack. Is pack.json broken?
Almost never; the game is choosing to draw. The usual reasons, most common first:
- An image server is running. With one on, the game only uses a pack's picture that matches the scene exactly (same outfit, room, mood) and draws the rest. With none running it shows the closest picture in the pack.
- The pack was added at another address. Packs belong to the address you added them at (the file, or the mod's address): add it again where you play, or let the Wayward Lab Helper share it.
- With Settings UI Mod: the pack must be set to show its pictures in Where her pictures come from, and "a new picture each time a scene comes back" must be off.
- Official Gen started after the game page: reload the game page.
- Something that's part of a picture's name changed (her look, the outfit, the model, settings), so old pictures no longer match.
To use only your pictures: in the game set Image generation to "Only use the pictures that came with the game". Find in Settings UI Mod: Where her pictures come from
See in Lab: Image generation: start here
What are scene prompts and image variants (the ‹ › counter, +, the 1/1)?
A scene's pictures have two kinds of numbers:
- Scene prompts, __v-0 to __v-3: the same scene written a few ways. The game picks one per visit; Settings UI Mod can hold more or fewer.
- Image variants, __v-4 and up: more pictures of the scene. These are what + adds and ‹ › steps through.
The game and Settings UI Mod mix the two up (the picture flips, the counter is wrong, a pack's extra pictures never show). The Wayward Lab Helper on the game page keeps each scene's list in order and makes + draw a real new picture. Choose how + and ‹ › behave in Lab > Image variants (with a pretend scene to try it), then reload the game page.
- + adds one at a time; the next + waits until the last one is drawn. Shift+click + goes to the next scene prompt; Alt+click + is the game's or the mod's own +.
- A line on the picture says what's shown ("Scene prompt 3 · Image variant 1/2").
See in Lab: Image variants; Image variants > Put the Helper on the game page
A scene is stuck at 1/1, or + does nothing. A game issue the Helper works around. The usual causes:
- The scene only has one picture: an image pack holds one picture per name, and in group scenes (another man joins or watches) the game reuses the picture it has instead of drawing a new one. Put the Wayward Lab Helper on the game page: + then draws a real new version and it stays (1/2, 2/2...).
- Its list was lost: a picture was renamed, moved or deleted by hand, or the browser's storage was full. Settings UI Mod 4.15.21 or newer puts back what it still has by itself; otherwise the Helper's Image variants (1/1) tab finds scenes whose list doesn't match their pictures and fixes them (tick, Apply, reload).
- To keep more than one picture per scene in a pack: Rebuild pack.json with "review extra pictures of a scene to add as versions", tick the ones you want, then add the new pack in the game.
- + adds a version but no picture appears: press F12 > Console. "429" means an Official Gen limit (it resets within an hour).
- Next time, don't rename or delete pictures by hand: use the thumbs down, or a review page.
See in Lab: Wayward Lab Helper > Fix a scene stuck at 1/1; Wayward Lab Helper > Group scenes; Rebuild pack.json
The Helper's tiles say "drawing…" for a long time, then "didn't finish", though the picture is there. Fixed in Lab 1.7.3: newer games answer a draw in a way the older Helper didn't follow, so it never heard the picture was done, and + on that scene waited 3 minutes. Update Lab, restart it and reload the game page.
See in Lab: Image variants
I pressed + or the thumbs down and two pictures were drawn. Each image server can also draw by itself. Check, in this order:
- Settings UI Mod: Pictures ahead of time (Draw ahead while idle) and "a new picture each time a scene comes back" draw extra pictures; turn them off to compare.
- A portrait or a title-screen picture may be drawn alongside the scene.
- Pressing + again before the first is in: the Helper waits for it, the game's own + doesn't.
Still two? Lab's Report a problem with the Helper's Start tracking shows exactly which requests the game page sent. Find in Settings UI Mod: Pictures ahead of time; Where her pictures come from
See in Lab: Report a problem
How do I redraw one bad picture?
The thumbs down at the top right of the picture (Mark as bad and regenerate) draws a new one with a new seed and replaces it. It needs an image server running. Each version (2/2, 3/3...) is redrawn on its own. With Lab's Helper, "What's wrong?" first asks what's wrong (face, pose, wrong people...) and keeps that note with the picture for later. With Official Gen, save and reload the page (F5) afterwards.
See in Lab: Bad pictures
How do I go through many pictures and redraw the bad ones?
Use the review that belongs to whatever drew them; each redraws through its own image server.
- Settings UI Mod: its Review pictures (J keep, K redraw, or by scene). Find it under Pictures ahead of time.
- Official Gen, or any pack: Lab's Pack review: OK or reject scene by scene, quick review with the keyboard, the rejects redrawn in one batch with Official Gen, and rejects taken out of an installed pack.
- The game's thumbs down with Lab's "What's wrong?" works with all of them; Bad pictures counts the most common problems.
Find in Settings UI Mod: Pictures ahead of time
See in Lab: Pack review; Bad pictures
Which prompt drew this picture?
Only the image server that drew it knows, at the moment it draws. An image pack is names and files: pack.json doesn't keep prompts. Where Official Gen or Settings UI Mod recorded them, Lab can keep them beside the pack (prompts.json). Settings UI Mod's Prompt inspector (inspect, in its menu's "In game" strip) shows the prompt of the picture on screen and where each part came from.
See in Lab: Image variants > What a picture knows
I changed how my character looks, but the old pictures still show. Pictures already drawn keep her old look; only new ones use the new look. Redraw the ones you want (the thumbs down, or a review page), or start a fresh pack for her new look.
The game keeps showing its original pictures instead of mine. The game's own girls come from files like images-<name>-1.js beside index.html; while those are there, their pictures keep showing. Lab's Image Pack Tool can turn them into one pack and set the .js files aside, or move them out of the game folder yourself (keep a copy).
See in Lab: Image Pack Tool
Should I convert my pictures to WebP?
For a smaller pack, yes: usually a third to a fifth of PNG's size, and packs work the same. Settings UI Mod (4.15.18+) does it itself and keeps its records right (search "WebP" in its menu). Other pictures: convert a copy for sharing, never the originals (image servers keep records that point at those exact files), then Rebuild pack.json on the copy. Export your workflow first: a WebP no longer carries the workflow a PNG does.
See in Lab: Rebuild pack.json
How long does drawing a whole character ahead of time take?
Tens of thousands of pictures, days of graphics card time, so leaving it running overnight is the idea. The pictures you're most likely to see are drawn first, so stopping early still helps. Settings UI Mod lets you leave out outfits, rooms, moods and scenes you don't want, and only fill gaps; its confirm box shows how many pictures that is. The game's own "Make pictures ahead of time" always draws the full set. Find in Settings UI Mod: Pictures ahead of time
Settings UI Mod says the game's storage in this browser is nearly full. The browser gives the game limited room for saves, added packs and the picture lists (the 1/1 counter). Full, it can't save and forgets picture lists, so scenes drop back to 1/1. Press Tidy up in the mod's note (it only frees what rebuilds itself), and delete old save slots you don't need (back them up in Lab first). Don't clear the browser's data: that deletes your saves and packs too.
See in Lab: Backups
Image packs & sharing
How do I add an image pack in the game?
- Use the downloaded game (2026-11-01 or newer); the web version can only add a pack someone put online.
- Unzip the pack into a folder named packs next to the game's index.html, so you have packs\<pack name>\pack.json. If unzipping made a second folder of the same name inside, move the inner one up.
- Title screen > Image packs > Add a pack, and pick that pack.json. A character from the pack is now on the New Game list; your saves are untouched.
- Playing through Settings UI Mod, add it at the mod's address (packs belong to the address they were added at), or with the mod's own Add a pack..., which also takes the .zip.
Find in Settings UI Mod: Where her pictures come from
See in Lab: Image Pack Tool
How do I make an image pack for my own character, start to finish?
- Make her: in the game (Build a character) or in Lab's Character record, where you set how she looks; then put her in the game (Export her > Put her in the game) and start a game with her.
- Her default outfit: Lab's creator, Outfit tab, then Send outfit to Settings UI Mod.
- Her prompts, models and LoRAs: in Settings UI Mod, for her (check HEROINE at the top of its menu). A custom workflow, if you use one: Lab's Custom workflow.
- Test small first: play a few scenes, or draw a few pictures, and fix her look and prompts before drawing many.
- Draw the rest: Settings UI Mod's Pictures ahead of time (leave out what you don't want), then review and redraw the bad ones.
- Make the pack: Lab's creator, Pictures tab > Import pictures into this pack, then Export > Save as a pack; or the mod's Her pictures as an image pack.
- Add it in the game and check it, then share it (next questions).
Find in Settings UI Mod: Character prompts; Pictures ahead of time
See in Lab: Character record; New pictures; Save & share
How do I turn my pictures into one pack to share or keep?
It depends on what drew them:
- Settings UI Mod: Her pictures as an image pack (under Pictures ahead of time) downloads a .zip with pack.json: the newest picture the mod drew of each scene, outfit, room and mood, one heroine at a time. Redraw the bad ones first, because the newest of each is what goes in. Pictures you kept with the thumbs up (Saved pictures) are a separate collection and are not what it exports.
- Official Gen, or any folder of pictures named by scene: Lab's Rebuild pack.json writes the pack.json from the files (work on a copy in the packs folder). Lab's Custom Image Pack Creator > Save them all as one pack puts every picture Official Gen drew into one pack.
- A character you built in the game: the Share button next to her on the character picker.
Then zip the pack's folder; Lab's Share a pack guide adds your workflow and a README. Find in Settings UI Mod: Pictures ahead of time
See in Lab: Rebuild pack.json; Share a pack & its workflow
I added new pictures to a pack. How do I make it one pack again?
Lab's "Save as a pack…" makes an additions pack: a second folder you add next to the original, and the game uses both (the one higher in the Image packs list wins where both have a picture). To merge them into one: copy the additions folder's pictures (not its pack.json) into the original pack's folder, run Rebuild pack.json on that folder, then remove the old pack in the game and add the rebuilt one. Pictures kept with the mod's thumbs up can't be merged this way: the mod stores them by scene folder without the name the game needs.
See in Lab: Rebuild pack.json; Save & share
My pack doesn't work, or it was made wrong. Lab's Rebuild pack.json on the pack's folder writes a correct pack.json. It works out which picture is which from file names, an old pack.json, images-*.js files, Official Gen's index or the data inside the pictures, and keeps the old one as pack.json.bak. It can fix a missing pack.json, renamed or moved pictures, a folder one level too deep (packs\packs, or packs\name\name) and old pack formats, and puts the pack in the game's packs folder. It only places pictures it can recognise: with plain file names, no old pack.json and no data inside the pictures, some can't be placed, and pictures that are gone can't come back. Work on a copy, and before replacing anything, check in the new pack that the scenes you expect are there. Then remove the old pack in the game and add the new one: the game keeps its own copy from when it was added.
See in Lab: Rebuild pack.json
The game finds pack.json but not its pictures. The pictures must be in packs\<the folder named inside pack.json>. Usually the folder was renamed or moved, or is one level too deep (packs\packs). Move it back, or run Rebuild pack.json on a copy and check the result (see the question above). The pack.json file itself can be renamed or moved: the game only reads what's inside it.
See in Lab: Rebuild pack.json
I downloaded a character's pictures. How do I use them with Settings UI Mod?
Unzip them into the game folder, then open Lab's Image Pack Tool > Image packs you downloaded; it finds them and shows what to do:
- Already a pack (has pack.json): How to add it shows the steps.
- A folder of pictures without pack.json: Make it a pack writes one.
- The game's own girls as images-<name>-1.js, -2.js... files: Make it one pack combines the parts into one pack with one pack.json.
Then add it at the mod's address and set it to show its pictures in Where her pictures come from. Patron portraits can't go in a pack; they stay with the .js files. Find in Settings UI Mod: Where her pictures come from
See in Lab: Image Pack Tool
How do I export a character I made in Lab?
The game only takes a character inside an image pack. In Character record, open or make her, then Export her: Put her in the game writes packs\<her folder>\pack.json (her pictures there stay), Share her as a zip makes a small pack for someone else. Then Title screen > Image packs > Add a pack, and she's on the New Game list. A copied or saved record on its own is for editing her, not for the game. An AI can edit a copied record: paste its whole answer back with Paste record, and Lab checks every value and lists what it fixed.
See in Lab: Character record
Can the Custom Image Pack Creator use pictures from somewhere other than Official Gen?
Yes. Set New pictures from to a folder: another pack's folder, Settings UI Mod's exported pack (unzipped), or pictures named by scene (like custom_x__scene-...__v-0.webp). For a character's pictures the mod drew on this computer, Import pictures into this pack fills that folder in for you.
See in Lab: New pictures
Can I use an image pack on the website or on a phone without a computer?
Only a pack someone put online: Title screen > Image packs, paste the address of its pack.json, Add. The site must be https, have the pack unzipped and allow other sites to read it (GitHub Pages for a small pack, Cloudflare R2 for a big one; many file-sharing sites don't). On an Android phone with the game copied onto it, a pack in its packs folder works in JoiPlay.
See in Lab: Play on your phone > D. A pack put online
Sharing fails with "value is out of range ... 4294967295". The share zip can't go past 4 GB. Share fewer pictures at once (by outfit or scene), zip the pack's folder with 7-Zip instead, or share a WebP copy, which is many times smaller.
Where do I share an image pack, or ask about one?
On the game's Discord, image-packs (a forum) gives each pack its own post: share it as a new post, and questions and fixes for that pack go there. image-packs-discussion is for packs in general. On F95zone a pack can have its own thread. Lab's Save & share makes the zip to send.
See in Lab: Save & share
Phones, mods & cheats
I copied a character's pictures next to the game on Android, but they don't load. The files are probably fine; opening index.html from a phone's file manager is the problem. It opens through a content link, not as a file in its folder, so the browser can't load anything next to it (packs, .js files, mods). Ways that work: JoiPlay (add the game's folder there and put packs in as on a computer), a small web server app serving the game folder (then http://localhost:8000/index.html), playing from your computer over Wi-Fi, or sintention.com for the plain game.
See in Lab: Play on your phone > B. Game copied onto an Android phone
Can I use mods on a phone?
- Playing from your computer over Wi-Fi: yes, the computer serves the whole game folder, mods included; Settings UI Mod too (start it with --host 0.0.0.0).
- Game copied onto Android: in JoiPlay, mods load like packs do. Settings UI Mod can't run on a phone.
- Game copied onto an iPhone: doesn't work; play from your computer instead.
- sintention.com: the plain game, no mods.
See in Lab: Play on your phone > Mods on a phone; Play on your phone > C. iPhone and iPad
How do I change gold or her stats (wayward.help)? And on a phone?
On a computer: open the game, press F12 > Console and type wayward.help() for the list, for example wayward.set("economy.gold", 500) or wayward.get("wife.attributes"). Changes save as usual; back up your saves first. It changes the current game only.
- Phone: a bookmark whose address is javascript:wayward.set("economy.gold",500);void 0, opened while the game is loaded (some browsers block these).
- Anywhere: Load Game > Export current to JSON, change the number in a text editor, then Import from JSON. Keep a copy of the original.
See in Lab: Play on your phone > Cheats
How do I put a cheat mod on my phone?
Set it up on your computer first: copy the game folder (keep a clean one), install the mod in the copy as its post says (a mod that needs a build step must be built on the computer), and check it works there. Then play that folder on the phone: Android or iPhone with your computer serving it (npx http-server -p 8080 in the folder, then http://<computer's address>:8080/index.html), or Android without a computer by adding the folder in JoiPlay. Move saves with Export current to JSON / Import from JSON.
See in Lab: Play on your phone > Mods on a phone
Saves, storage & backups
My saves disappeared when I opened the game another way. They're still there. The browser keeps saves separately for every address: index.html opened as a file, the mod's address, and localhost versus 127.0.0.1 are all separate. Open the game the same way as before. To move saves between them: Lab > Backups > Both versions & room (it keeps a copy first, and Undo puts it back), or download one from Backups and use Load Game > Import from JSON.
See in Lab: Backups
Where are my save backups, and how do I restore one?
Lab > Backups shows where they are (by default wayward-lab-data\save-backups). To restore: download the save there, then in the game Load Game > Import from JSON, at the address you play at. Nothing being backed up? Lab must be running while you play, and Backups must point at the browser profile you play in. Character and settings files there are not playable saves.
See in Lab: Backups
I have more saves than the game's 12 slots. Lab > Backups > Both versions & room has a save room: park saves there to free a slot, and put one back when you want to play it.
See in Lab: Backups
Pictures an image server drew show in one version of the game but not the other. They belong to that image server and the address it was used at, so they can't be shared like packs. Make them a pack (see "How do I turn my pictures into one pack"); a pack works in both versions, without the server.
See in Lab: Image Pack Tool
Where does Lab keep its files, and what is the to-delete folder?
Lab > Lab storage shows everything in one place: Lab's own folder (wayward-lab-data: back it up to keep your Lab setup), what Lab adds to the game folder (packs, share zips), and to-delete. Lab never deletes your files: what it no longer needs goes to to-delete, which you can empty yourself whenever you like. Recently set aside > Put back undoes a move. Tidy up moves old copies, old logs and old share zips there, after you've seen the list.
See in Lab: Lab storage
Lab asks me to move existing files, or reports a conflict. Lab keeps its own files in wayward-lab-data. The storage notice moves older Lab files there from where older versions kept them; it checks every file and updates the paths. Where two different files have the same name, nothing is overwritten: both are kept and listed, so you can compare and choose. The mod's own files stay where they are.
See in Lab: Lab storage
How do I put my settings in one file, or apply someone's?
Lab storage > Your settings in one file > Export my settings downloads wayward-settings.json (Image variants choices, Lab's choices, model families, the mod's shared changes, other servers' model choices). It leaves out folders, addresses, ports, keys and saves, and has no models or pictures. To apply one: Open a settings file…, compare Now and From the file for each part, tick only what you want, apply, and read Applied and Not applied. Read a file before sharing it: prompts are your own words.
See in Lab: Lab storage
Why are there several copies of one character record?
The same character can have a Lab copy, a mod copy, one inside a pack and copies in backups. Open the character's entry to choose which to edit. Save to Lab updates Lab's copy only.
See in Lab: Character record
Lab itself
How do I report a problem so it can be fixed?
Lab > Help > Report a problem. Update everything first (Lab, Settings UI Mod, the game): the page checks your versions. Say what you were doing, what happened instead, and paste the error exactly. Lab adds what it can see (image servers, ComfyUI, the game page, the mod's own report) with likely causes at the top, shortens folder paths and removes your user name. Lab sends nothing itself: Save & open saves the report and opens the right place to post it (Lab's thread, the mod's thread, or the game's Discord for the game itself). With the Wayward Lab Helper's Start tracking, the report also shows what the game page did while the problem happened. Crop folder paths out of screenshots.
See in Lab: Report a problem; Wayward Lab Helper
What is the Wayward Lab Helper?
A small part of Lab that runs on the game page itself. It keeps + and ‹ › working, fixes scenes stuck at 1/1, opens places in Settings UI Mod for Lab's buttons, shares packs and custom characters between your two game addresses, and adds the game page's side to problem reports. Put it on with Lab > Image variants, step 3 (added to the game page, or as a bookmark); reload the game page after changing its settings. It touches no files and leaves saves alone.
See in Lab: Wayward Lab Helper; Image variants > Put the Helper on the game page
How do I update Lab safely?
Close Lab's window, then extract the new ZIP into the same game folder, keeping its folders, and choose Replace. Don't delete the old Lab first or extract into a new folder, or wayward-lab-data (your workflows, records, backups and settings) is left behind. Start start-wayward-lab.bat; if the page looks old, press Ctrl+F5. Copying wayward-lab-data somewhere safe first doesn't hurt.
How do I stop Lab?
Close its black window (or press Ctrl+C in it). Closing the browser tab leaves Lab running. Stopping Lab doesn't stop ComfyUI or the image servers.
Lab says it couldn't use port 8799. Lab is already running, usually because start-wayward-lab.bat was opened twice. Open http://127.0.0.1:8799/ to use it, or close its window first. A second copy can run on another port: start-wayward-lab.bat --lab-port 8800.
Windows Defender says Lab's zip has a trojan (Trojan:Script/Wacatac.H!ml). Is it safe?
The "!ml" in that name means Defender's machine-learning guess flagged it, not a match with known malware; since a Defender update on 11 November it has flagged Lab 1.6.0 and 1.7.1 zips from the official post. That makes a false positive likely, not certain for your copy:
- Check you downloaded it from the official post, and that the name matches exactly.
- Note the version, which file Defender names, and the zip's SHA-256 (PowerShell: Get-FileHash <the zip>); compare it with the one in the post, or post it in Lab's thread.
- Scan it at virustotal.com, and report it to Microsoft at microsoft.com/wdsi/filesubmission ("Incorrectly detected as malware").
Lab is plain JavaScript you can read: lab\lab-src has every source file, and build.js makes wayward-lab.js from them.
Does Lab change my game or Settings UI Mod?
Only when you press something that says so, such as Apply to Settings UI Mod, installing a model, or putting the Helper on the game page (which adds one line to index.html and keeps a copy). Lab keeps its own files in wayward-lab-data and never edits the mod's program files.
I can't find a page in Lab's menu. Pages that belong together share one menu entry, with tabs at the top. Workflow test, A/B compare and the Prompt builder are under Workflows > Custom workflows; the guides under Help > Guides; the Wayward Lab Helper under Game tools > Image variants & Helper; Rebuild pack.json under Game tools > Image packs; Backups, Lab storage and Character record under Game tools > Saves & files. Still missing? Show / hide pages may hide it.
See in Lab: Lab storage
The Help pages show "Screenshot guide missing". Only part of Lab was copied, or the ZIP was extracted without its folders. Extract the whole ZIP into the game folder again, keeping its folders, and restart Lab.
Can I use Lab in my language?
Yes: pick it in the Language list at the top right if someone has translated Lab, or use your browser's Translate. Model names, file names and commands are never translated, so what you see is what to download or pick. To help translate, Language > More languages... > Save translation template; docs\translations\README.txt explains the format.
Image generation: getting started
By ExTendZone. The F95 post with the screenshots: https://f95zone.to/threads/wayward-ai-fanart-support-thread.318124/post-21850407 Older versions: https://f95zone.to/threads/wayward-v2026-10-04-sintention.310923/post-21815880 and https://f95zone.to/threads/wayward-v2026-10-04-sintention.310923/post-21703538
Mod builds (wayward v2026-11-09)
The most asked-for part, first. Check each post for its newest version.
The game and its makers
- Main game thread and F95 downloads: https://f95zone.to/threads/wayward-v2026-11-01-sintention.310923/
- Official website: https://sintention.com/ It describes the game and links to everything else: play in the browser, the itch.io download (split into parts, not all in one like on F95), the Sintention Discord (community, feedback, troubleshooting, the latest mod builds), the developer's description of the game, and Patreon and Subscribestar to support the developer.
Step 0: a clean installation
This guide uses the files from itch.io: https://sintention.itch.io/wayward The F95 downloads may include all the girls' image packs and their scripts; on itch.io you can download only the base game.
This guide relies on image generation, not on ready-made image packs. So download only wayward-game.zip: the base game, without the girls' image packs, their scripts (images-pippa-1.js, images-pippa-2.js and so on) or the images folder. If you have those, remove them to follow this guide exactly: leftover packs and scripts can cause conflicts.
You only need index.html. The other Wayward .html files can be deleted.
What image generation needs
- ComfyUI. For beginners ComfyUI Desktop, the version used in this guide. The portable version works too: make sure it uses the right graphics card (search "ComfyUI change GPU"). Official installation guide: https://github.com/Comfy-Org/ComfyUI#installing
- The custom nodes (node packs) each pipeline needs. The guides below say which ones for Illustrious and Anima.
- Bun: https://bun.sh/
- Node: https://nodejs.org/en
- The model files (checkpoint, LoRA and so on): in their own guides.
Choosing your setup
- Play with the game's own image packs: then this guide isn't for you.
- Make your own pictures with wayward-imagegen: it's already in the files from step 0 (the wayward-imagegen folder). It's the official, unmodded way to generate pictures, uses the Illustrious pipeline and copies the developer's style.
- Use a mod build for custom image generation (or a mix of mod builds): modded, may use other pipelines too (Illustrious included), and can use a custom style. There may be bugs: share them clearly, with exact steps to make them happen again, and expect to deal with some problems.
Official image generation (wayward-imagegen, Illustrious)
The built-in Illustrious workflow, the game's official image generation companion, unmodded.
Easier ways to set it up:
The steps below are still here for reference.
Note: if you use a mod build like Settings UI Mod, it uses a different server for image generation. Wayward Lab now makes the difference clear, so you don't need to delete the wayward-imagegen folder any more.
Step 1: read the readmes
Read the README in wayward-game and the one in the wayward-imagegen folder if you're new to this; the steps below assume you have.
wayward-imagegen.config.example.json shows the checkpoint and LoRA it uses. They are for the Illustrious pipeline (base model) only and break with another base model like Anima.
To use another checkpoint or LoRA: copy wayward-imagegen.config.example.json to wayward-imagegen.config.json and change the file names in it. It still only works with Illustrious. If you get to that point, a mod build like Settings UI Mod makes changing the workflow settings much easier. The official wayward-imagegen and the mod builds start their own servers; try not to mix them up.
Step 2: the model files
Check the downloads are complete: compare file sizes (6.94 GB for the checkpoint). One player's download was cut short without him noticing. The LoRA is made for 8 steps, so pictures are drawn much faster than the usual 24 or so steps of an Illustrious workflow.
The game's default files for its Illustrious workflow:
Where the files go (to change the folder or add another one: https://docs.comfy.org/basic-concepts/models#adding-extra-model-paths):
- ComfyUI Portable: \ComfyUI_windows_portable\ComfyUI\models
- ComfyUI Desktop: AppData\Local\Comfy-Desktop\ComfyUI-Shared\models
Finding the Desktop models folder:
- Way 1: in ComfyUI Desktop, the menu in the top-left corner > Desktop Settings > Storage > the folder icon. Open ComfyUI-Shared, right-click the models folder and make a shortcut; put it somewhere handy (Downloads).
- Way 2: search for Comfy Desktop in the Start menu, open its file location (twice, if it shows the shortcut first). That's \AppData\Local\Programs\Comfy Desktop; go up to AppData and on to AppData\Local\Comfy-Desktop\ComfyUI-Shared\models. Pin it or make a shortcut.
In the models folder:
- Illustrious (SDXL) checkpoint: models\checkpoints
- Anima: the model (unet) goes in models\diffusion_models
- LoRAs: models\loras (use a LoRA only with its own pipeline)
Step 3: the nodes
Start ComfyUI. Top-left "C" logo > Extensions (the same in Desktop and Portable). In the Nodes Manager, set it to "Node Pack" (not "Nodes") and install the 3 node packs wayward-imagegen needs:
- ComfyUI-KJNodes (search "kj")
- ComfyUI Impact Pack and ComfyUI Inspire Pack (search "impact")
Step 4: start it
In the wayward-game folder, open the wayward-imagegen folder, right-click inside it, open a terminal and run:
bun src/cli.ts
(It's in the wayward-imagegen README too.)
Open index.html in the wayward-game folder. Settings > Image generation: set to "Automatic", it says "connected" when it finds wayward-imagegen.
Mod builds may generate pictures and use image packs in other ways. This section is the unmodded way, with the game's own style.
Mod builds and resources
Mod builds are for custom image generation and much more. Many are in the Sintention Discord; the ones posted on F95 are at the top of this guide.
Tools that help:
- ComfyUI LoRA Manager: helps with every kind of workflow and makes workflows easier to share: https://github.com/willmiao/ComfyUI-Lora-Manager
- Wayward Lab: starts the right image generator for you, so you don't mix up the official one and the modded ones.
Before using mod builds:
- Remove images-pippa-1.js (and similar scripts) and the images folder if you have them (from the full F95 download or a girl's image pack): they can conflict with the mods and bring back the original pictures. To be safe, also clear the browser's cached images and files (Delete browsing data, All time, Cached images and files). This may no longer be needed since custom girls arrived, but people still run into it.
- Settings UI Mod and Custom Image-Gen Server generate pictures their own way, not through wayward-imagegen. With Wayward Lab you don't have to worry about which one to start.
- After updating a mod like Settings UI Mod, check your settings (steps, CFG and the rest) even after importing your old config.json: they may have been reset or changed. Compare before and after every update.
- Leave the mod makers a like if you enjoy their work.
Reading
Why start with Illustrious, not Anima
From Anongatari's analysis in the Sintention Discord (summarized here): the game's prompts and workflow are made for SDXL / Illustrious, so Anima needs work on almost every layer:
- Its own workflow: three files (model, text encoder, VAE) with their own loader nodes, and no FaceID face reference (that's SDXL only).
- Converted prompts: the game writes tags with underscores (on_bed), Anima was trained on spaces; weights need to be stronger (1.2 becomes about 1.5); brackets like bar_(place) need escaping.
- Negatives: Turbo runs at CFG 1, where ComfyUI ignores the negative prompt, and the game relies on negatives (body shape, 2girls, its age guards). CFG about 1.5 brings them back at about twice the time.
- Speed: Turbo is about as fast as the default (a few seconds a picture on a 12 GB card); base Anima is many times slower, too slow for how many pictures the game draws.
- Look: Turbo doesn't match the game's art; style tags close part of the gap, style LoRAs the rest, and those work better with the slow base model.
So beginners are better off with Illustrious first.
His tips for a custom character's Additional tags:
- Don't add expressions, poses, gaze, clothes, temporary states (marks, fluids, drunk), room, lighting or quality tags: the game sets those per scene, and yours would fight it or show in every picture. Don't add body tags that clash with your Look picks, or tags that make her look younger.
- Do add what's true in every picture: body and face detail, permanent marks, accessories. Use real Danbooru tags (check the Danbooru wiki).
- Keep it short, 5 to 8 tags: the game picks the steps from the total tag count (up to 40 tags 8 steps, 41-60 12 steps, over 60 18 steps). Weights (tag:1.1) to (tag:1.2), no space after the colon.
- A named character base can replace your hair and eye picks; turn its strength down if your Look loses to it.
Image packs
Image packs are in the Sintention Discord: talk about them in #image-sets, find shared ones in the forum "[Image Packs] User-Shared Image Packs". The F95 post lists some shared packs too (their download links expire). If a pack's pack.json doesn't import, Wayward Lab can make a new one.
Asking for help
Say this so others can understand your setup:
- Guide: have you read this guide?
- Image generation: the official wayward-imagegen or a modded setup (such as Settings UI Mod)?
- Pipeline and workflow: Anima or Illustrious? A custom workflow?
- Settings: have you changed settings, prompts, checkpoints, LoRAs or config files?
- The problem: what you're trying to do, what happens instead, what you've tried, and any error messages or screenshots.
If nobody answers, they may not know or not be around. You can also ask in the Sintention Discord, or ask an AI with the details of your problem.
Modded setup and settings per pipeline
Based on ExTendZone's F95 posts. A plain, in-order guide for drawing the game's pictures with a mod build (Settings UI Mod here) and your own model, for any pipeline: Illustrious (SDXL), Pony, Anima and others. Then the usual settings for each pipeline, normal and fast (Lightning / DMD2 / Turbo). For the rest see Image generation: getting started, and the F95 post: https://f95zone.to/threads/wayward-v2026-10-04-sintention.310923/post-21815880
Wayward Lab can do much of this for you: it starts the right servers, finds your model files and sorts them per pipeline, and walks you through the settings.
First, pick a pipeline
A pipeline is the family of model a workflow is built for. Every file you use (model, LoRAs) must be made for the SAME pipeline: a LoRA for Illustrious does nothing (or breaks the picture) on Anima, and the other way round. The model site shows it as "Base model".
- Illustrious (SDXL): what the game ships with (WAI Illustrious plus the Lightning LoRA). The game's prompts are written for it. The easiest start, and the most LoRAs.
- NoobAI, Pony and other SDXL families: also SDXL, but each has its own prompt habits; use LoRAs made for that family.
- Anima: a newer anime model with its own workflow (three files: model, text encoder, VAE). Pretty, but needs prompt changes to fit the game; see "Why start with Illustrious" in Image generation: getting started.
The steps
- Install ComfyUI.
- Install Node.js 18 or newer.
- Install Bun.
- Install the nodes your workflow needs (the mod's or Lab's page says which).
- Download wayward-game.zip from itch.io (the latest game version only, not the girls' image packs).
- Install the mod build you want to use (here: Settings UI Mod).
- Get a model for your pipeline and put it in its folder (see "Where the files go"). Anima also needs its text encoder and VAE.
- Find LoRAs (a character, a style, more if you like) made for the same pipeline, and put them in the loras folder. Click pictures in their Gallery to see their prompt, trigger words, look, outfit and settings (not every picture shows them, but it's a good start).
- Make sure ComfyUI is running and can draw a picture in a workflow.
- Start the mod's server with its .bat file. A terminal opens; it also checks that ComfyUI is running.
- Open the game through the mod's shortcut ("modded wayward") or at http://127.0.0.1:8788/
- In the game there, click the Pippa button > Config, and set where your ComfyUI models folder is (ComfyUI\models, or the ComfyUI folder). The server looks through that folder and the ones in it on startup and when you click Scan.
- It should now find your files: check the Pipelines & Models tab.
- Pick your pipeline's workflow and your model (the checkpoint field for SDXL families, the unet field for Anima).
- Pick your LoRAs. Many need a trigger word in the imageBase to work (some work by themselves): check the LoRA's page.
- Set the steps, CFG, sampler and scheduler for your model (see "Typical settings" below, and the model's own page, which wins).
- In the Heroines tab (and Outfits), start her imageBase from the LoRA's trigger words and prompt: that brings back her look, the style and the character. An outfit trigger word goes in the outfit's prompt.
- IMPORTANT: save and apply after every step in Settings UI Mod, so nothing is lost, and check nothing was reset (after an update too).
- VERY IMPORTANT: refresh the game page so the settings are used for the next pictures.
- Restore all scenes in the Scenes & NPC tab: some may be blank to begin with.
- Turn off "Convert to GIF" in Pipelines & Models if you haven't set that up.
Where the files go
ComfyUI Desktop keeps its models in AppData\Local\Comfy-Desktop\ComfyUI-Shared\models (Portable: ComfyUI_windows_portable\ComfyUI\models).
- SDXL families (Illustrious, NoobAI, Pony) model: models\checkpoints
- Anima model: models\diffusion_models (the unet field). It's a checkpoint too, but for Anima it does NOT go in checkpoints. Its text encoder goes in models\text_encoders (or clip), its VAE in models\vae.
- LoRAs, Lightning / DMD2 speed LoRAs too: models\loras
A shortcut to these folders in your game folder makes them quick to reach.
Typical settings
Starting points; the model's own page always wins (it often names its sampler, steps and CFG). Keep steps and CFG close to a picture's if you want its style.
Illustrious / SDXL, normal
- Steps 20-30 (more for polish, fewer for drafts)
- CFG about 4-7 (lower: freer and softer; higher: follows the prompt more but can look harsh or oversaturated)
- Sampler Euler a (euler_ancestral) or DPM++ 2M; scheduler normal or karras
- CLIP skip 2 for most anime SDXL models
- Size around 1 megapixel: 1024x1024, 832x1216, 896x1152
- Negatives work fully
Illustrious / SDXL with a speed LoRA (Lightning or DMD2)
- The game's own setup: WAI Illustrious + SDXL Lightning 8-step LoRA at strength 1.0, 9 steps, CFG 1.5, Euler a, simple scheduler, CLIP skip 2.
- Lightning: 8 steps for the 8-step LoRA (4 for the 4-step one), CFG 1 to 1.5, Euler or Euler a, simple or sgm_uniform.
- DMD2: about 4-8 steps, CFG 1 to 1.5, often the LCM sampler.
- Several times faster, which matters: the game draws a lot of pictures.
Pony (SDXL)
- Like normal SDXL (20-30 steps, CFG about 6-7, CLIP skip 2), plus its score tags at the start of the prompt (score_9, score_8_up...) and its own LoRAs.
Anima, base or aesthetic
- Steps 30-50, CFG 4-5
- Samplers: er_sde (neutral, a good default), euler_a (softer lines), dpmpp_2m_sde, euler
- Size 512 to 1536 px a side
- Negatives work; much slower than Turbo
Anima Turbo
- Steps about 8-12, CFG 1
- Euler or er_sde; simple or sgm_uniform scheduler
- About as fast as the game's own Illustrious + Lightning setup
Fast (Lightning / DMD2 / Turbo) versus normal
- Speed: fast setups draw in a fraction of the time (8 steps instead of 30, one pass instead of two). For a game that draws hundreds of pictures, that's the big reason to use them.
- Negatives: at CFG 1 ComfyUI ignores the negative prompt completely. The game relies on negatives (body shape, 2girls, its age guards). Use CFG a little above 1 (1.1 to 1.5) to bring them back: slower, but negatives work again. The game itself uses 1.5 for this reason.
- Look: fast setups have a stronger default look and less variety. Style LoRAs work better on normal models; Anima Turbo and style LoRAs in particular don't mix well.
- Settings matter more: a fast setup with normal settings (30 steps, CFG 7) burns the picture; a normal model with fast settings (8 steps, CFG 1) comes out blurry and unfinished. Match the settings to the model.
- A speed LoRA belongs to its pipeline too: SDXL Lightning and DMD2 are for SDXL families only, not Anima.
Choosing your own model, LoRAs and workflow
Where to look
- Civitai (civitai.com, and civitai.red for adult content): the most models and LoRAs, with example pictures and their settings.
- Hugging Face (huggingface.co): the original model files (the game's default checkpoint and Lightning LoRA, Anima), usually with a model card that gives the settings.
- ComfyUI's own templates (Workflow > Browse Templates): ready workflows for each model family, a good start for a custom workflow.
- Other players: F95 and the Sintention Discord, where people share the model, LoRAs and settings behind their packs.
Filtering on civitai
It's not a needle in a haystack once you filter: sort by Most Downloaded (or Highest Rated), set Base model to your pipeline (Illustrious, Anima, Pony...), and the type: Checkpoint for a model, LoRA for a character, style or concept. Search the character's or style's name.
Reading a model or LoRA page
Before downloading, check on the right-hand side and in "About this version":
- Base model: must match your pipeline. A page can have versions for different base models (one for Illustrious, one for Anima): pick the version (tabs at the top) for yours.
- Trigger words: words to put in the prompt to switch the LoRA on (a character's name tag, a style word). Some LoRAs work without one. Copy them exactly.
- Recommended settings: the author's sampler, steps, CFG, CLIP skip, size, and for a LoRA its strength (weight), often 0.6-1.0. Start there.
- Files: a .safetensors file (avoid .ckpt / .pt from unknown sources). "pruned" or "fp16" is the smaller file you want for drawing. Note if the model needs its own VAE or says "VAE baked in".
- Its gallery and reviews: pictures other people made with it, which show what it's really like, flaws included.
Copying a showcased picture
The quickest way to learn a model or LoRA: copy a picture you like.
- On the model's page, open a picture from its gallery or showcase.
- Look for its generation data (the "i" or info panel): the prompt, the negative prompt, the resources used (model, LoRAs and their strengths), sampler, steps, CFG, seed and size. Use the copy button next to the prompt. Not every picture shares this; pick one that does.
- Get the same model and LoRAs, the same versions.
- Put the settings in: model and LoRA strengths in Settings UI Mod's Pipeline & models (or your workflow), steps, CFG, sampler and size the same, and the same seed to get close to the same picture.
- Split the prompt for the game: the parts about her (hair, eyes, face, body, the LoRA's trigger words) go in her imageBase; her clothes go in the outfit's prompt; the scene, pose, room and expression come from the game, so leave those out.
- Draw one picture and compare. It won't be identical: the site's own generator, upscalers, detailers (ADetailer, hires fix) and other versions change things. Close is the goal.
If the picture was made with ComfyUI and still has its data, download the original file and drag it into ComfyUI: it opens the whole workflow that made it.
Using trigger words and LoRAs properly
- Put a character LoRA's trigger words at the start of her imageBase, and a style LoRA's in every picture (imageBase too, unless your setup has a style field).
- Strength: start at the author's number. Too high burns the picture or copies the training pictures (same pose, same clothes); too low and the LoRA barely shows. Lower a character LoRA if her clothes stick to every outfit.
- Two or three LoRAs at once is plenty; lower each a little when stacking.
- A LoRA's outfit words go in the outfit's prompt, not the imageBase, or she wears that outfit in every scene.
- Speed LoRAs (Lightning, DMD2) need their own steps and CFG; see "Typical settings".
Your own custom workflow
- Settings UI Mod's Pipeline & models gives the built-in workflows (pick a pipeline, then the model, LoRAs, steps and CFG). Most players need nothing more.
- A custom workflow (an extra detailer, an upscaler, another loader): build or open it in ComfyUI (from a template, or from a picture's data), make sure it draws a picture there, then save it in the API format. Wayward Lab's My workflow and its workflow library take it from there: Lab fills in the game's prompt, seed and size, and checks which nodes and files it needs.
- Keep a custom workflow to one pipeline, and note its settings with it, so you (or someone you share it with) can redo it.
Links to start with
Models:
LoRAs (change the Base model filter for your pipeline):
An example: a character LoRA's page shows the model it was used with and its prompts. Its look words go into the imageBase in Settings UI Mod; its outfit words into the outfit (clothing) prompts: https://civitai.red/models/2647520/yor-briar-yor-forger-spy-x-family?modelVersionId=2992727
A style LoRA with versions for both pipelines (check its details for which base model each version is for; its v2 works with Illustrious): https://civitai.red/models/897828/theobrobine-illustriousanima-style-lora?modelVersionId=2858463
If you want to try the Anima workflow in ComfyUI first, and get the files it needs: https://f95zone.to/threads/wayward-v2026-10-04-sintention.310923/post-21867460
Older examples:
Which LoRA is right depends on the look you want. If you find good ones, share them for others to try.
Try models
In Lab: Help > Guides > Image generation: start here
Beginner outline
Image generation: start here
Pick one of the three ways below. Most people want way 1 or 2; Parts A-D are the in-depth steps for setting things up by hand.
Pick your way
- JUST PICTURES, SET UP ONCE (Windows, NVIDIA RTX 20 or newer, 8 GB+; AMD experimental) 1. Get the installer by Anongatari and put its files next to the game's index.html: https://f95zone.to/threads/wayward-v2026-11-09-sintention.310923/post-22034259 2. Double-click install-imagegen.bat and choose 1 (Install). It installs ComfyUI, Bun, the node packs and about 11 GB of models. 3. Double-click start-wayward.bat: it starts everything and opens the game. 4. In the game: Settings > Image generation > Automatic (once). You don't need Lab for this; it can still help with the FAQ, packs and the Helper.
- MORE CONTROL: SETTINGS UI MOD WITH LAB (models, LoRAs, your own workflows) 1. Install ComfyUI (ComfyUI Desktop or portable). Anongatari's installer can do it for you too. 2. Put Settings UI Mod next to the game's index.html and start it (Start-Wayward-Bridge.bat). 3. Start Lab (start-wayward-lab.bat). In Settings UI Mod > Downloads & requirements, download what's missing. 4. In the mod's menu: Pippa > Pipeline & models, pick your model. Open the game from the mod's address and play.
- BY HAND, OR ON LINUX AND MACOS: Parts A-D below. A. Prepare: ComfyUI, Node.js or Bun, the game folder. B. Files: pick Anima or Illustrious, get a main model and a LoRA, put each in its folder. C. Connect: node packs, start ComfyUI and your image server, pick the files. D. Test: trigger words in the prompt, save, make one picture. Linux and macOS: start Lab with sh start-wayward-lab.sh; Anongatari's installer is Windows-only.
Run one image server at a time.
Examples
Way 1: your game folder after the installer
D:\Games\Wayward\
index.html the game
wayward-imagegen\ Official Gen (comes with the game)
install-imagegen.bat Anongatari's installer
start-wayward.bat made by the installer: double-click to play
D:\ComfyUI\ installed by it (default: root of the game's drive)
Way 2: an Anima setup with a character LoRA
D:\Games\Wayward\
index.html, wayward-bridge.js, Start-Wayward-Bridge.bat game + Settings UI Mod
lab\, start-wayward-lab.bat Lab
ComfyUI\models\
diffusion_models\anima-base-v1.0.safetensors main model (UNET picker)
text_encoders\qwen_3_06b_base.safetensors its text encoder
vae\qwen_image_vae.safetensors its VAE
loras\yor-anima.safetensors character LoRA (strength 0.8)
Game: http://127.0.0.1:8788/ (the mod's address)
File names are examples: use the ones you downloaded.
Way 3: Official Gen by hand on Linux
cd ~/Games/Wayward
sh start-wayward-lab.sh # Lab, opens in your browser
cd wayward-imagegen && bun src/cli.ts # Official Gen (or Start it in Lab)
# ComfyUI running at http://127.0.0.1:8188
# Game: open index.html, Settings > Image generation > Automatic
In-depth steps (parts a-d)
Examples use Anima and Illustrious. When a menu says model family, it means the model type—for example, Anima or Illustrious. Pick files made for the type you are using.
Part A — Prepare the tools and game
- Install ComfyUI. Use ComfyUI Desktop (https://github.com/Comfy-Org/desktop) or a portable/manual installation. If it is already installed, keep it. Note where its models folder is; the application folder and model-storage folder can be different.
- Choose one image source. Official Gen is included with the game and uses the native Illustrious pipeline. Settings UI Mod lets you select a model type, including Anima or Illustrious. Custom Image-Gen Server uses Anima. A custom workflow uses the model type its graph was built for. Choose the source first so you download the right files.
- Install Node.js or Bun as needed. The bundled Official Gen uses Bun (https://bun.sh) (its package specifies 1.1+). Settings UI Mod and Wayward Lab use Node.js 18+ (https://nodejs.org). Other builds may differ: follow that build’s launcher instructions. Bun is not a requirement for every source.
- Download and extract the full game. Get the supported game build from the game’s itch.io page (https://sintention.itch.io/wayward), using wayward-game.zip rather than a girl/image-pack download. Check your chosen mod supports that game version. Find the extracted index.html ; this is the game folder. Put the chosen mod’s files there as its instructions say. For Lab, extract its ZIP beside that index.html and run start-wayward-lab.bat .
Part B — Choose Anima or Illustrious and get its files
- Choose Anima or Illustrious. For native Official Gen, use Illustrious-compatible files. For an Anima setup, use Anima files. In Settings UI Mod, open Pipeline & models and set the Pipeline — model family dropdown to Anima or Illustrious XL. That dropdown chooses which model type the mod uses. Do not combine an Anima LoRA with an Illustrious model just because both filenames end in .safetensors.
Example 1: Anima
- Use Settings UI Mod with Anima selected, or Custom Image-Gen Server.
- The basic Anima example uses these three files:
- Main image model · UNET picker anima-base-v1.0.safetensors Folder: models/diffusion_models
- Text encoder · CLIP picker qwen_3_06b_base.safetensors Folder: models/text_encoders
- VAE picker qwen_image_vae.safetensors Folder: models/vae
- Select those files in the UNET, CLIP/text encoder and VAE fields. The UNET field is the main Anima model picker.
- Add a LoRA whose selected version says Anima. Use its trigger words and strength. For a character, appearance words go in the character prompt; outfit words go with the outfit.
Example 2: Illustrious
- Use native Official Gen, or select Illustrious XL in Settings UI Mod.
- The bundled Official Gen example uses:
- Checkpoint picker waiIllustriousSDXL_v170.safetensors Folder: models/checkpoints
- Acceleration LoRA sdxl_lightning_8step_lora.safetensors Folder: models/loras
Select the actual filenames in its settings. Open Official Gen > Downloads & requirements for its download list and destination folders.
- Keep the checkpoint’s own VAE unless the selected model/workflow says to use another. Keep the Lightning LoRA with its intended sampling settings.
- Settings UI Mod or a compatible custom workflow can provide extra LoRA slots. Native Official Gen’s configured LoRA slot is commonly used by Lightning; additional character/style LoRAs may need a custom workflow.
- Find a main model. On Civitai/Civitai Red, filter by Base model, Model type → Checkpoint, then Most downloaded if you want a smaller starting list. Open a result and check the selected version’s base and instructions. For Anima, a file listed as “Checkpoint” on the website can be a diffusion-model file for the UNET picker; it does not automatically belong in ComfyUI’s checkpoints folder. Open the filtered searches and my Turbo v1.1 example here.
Filtered searches and examples from my post
Start with the filtered list for your pipeline rather than browsing every model. On the site, confirm the visible filters and choose Most downloaded. Search layouts can change.
My Anima Turbo v1.1 example: Anima, version 3263843 (https://civitai.red/models/2458426/anima?modelVersionId=3263843). Choose its actual downloaded filename in the UNET picker and put it in diffusion_models. The basic Anima model and Turbo are different choices; keep the selected version’s sampling instructions with it.
Example character page: Yor character LoRA, version 2992727 (https://civitai.red/models/2647520/yor-briar-yor-forger-spy-x-family?modelVersionId=2992727). Check the selected version’s base and trigger words; gallery details can show how its creator used it.
Example style page: Theobrobine style, version 2858463 (https://civitai.red/models/897828/theobrobine-illustriousanima-style-lora?modelVersionId=2858463). Versions on the same page can target different bases, so choose by the version’s details, not the page title alone.
These are examples from my original post. Check the selected version’s current base model and instructions before downloading. You can also search Hugging Face model cards (https://huggingface.co/models) for matching files.
- Open the download list for the source you chose. Each source has its own Downloads & requirements. Use its list to find the required models/nodes and their destination folders:
- Official Gen > Downloads & requirements
- Settings UI Mod > Downloads & requirements
- Custom Image-Gen Server > Downloads & requirements
- Custom workflow > Downloads (open a workflow first)
Custom workflow opens the Downloads tab for the workflow currently open in Lab. Source checklists use that source’s saved settings. Keep the selected version’s sampling instructions with its files, especially for Turbo models.
- Find one character or style LoRA. Filter by your same base model and Model type → LoRA, then search the character or style you want. Download the matching version. Read its trigger words and recommended strength. Click a showcase/gallery picture to inspect its prompt, model, LoRAs, sampler, steps and CFG when those details are available. Not every picture publishes all of them.
- Put each file in the right folder. Use this table. Lab can open the detected model location or install an already-downloaded file there. Refresh/rescan afterwards. Keep the exact filenames; selecting a different filename in the server does not rename the file.
File locations
Downloaded file | Folder under your ComfyUI models root | Illustrious/SDXL checkpoint | checkpoints | Anima main model for the UNET picker | diffusion_models | Character, style or acceleration LoRA | loras | Separate text encoder, such as Anima’s Qwen encoder | text_encoders | Separate VAE, such as Anima’s Qwen image VAE | vae |
Desktop example from my post: press Win+R and paste %LOCALAPPDATA%\Comfy-Desktop\ComfyUI-Shared\models . Use it only if that is the model root your installation actually uses. For example, an Anima model then goes in its diffusion_models subfolder. A portable example is C:\ComfyUI_windows_portable\ComfyUI\models . If the example folder does not exist, use Lab’s Open models folder or the model location configured in your ComfyUI installation instead of creating a guessed folder.
Part C — Connect the server and select the files
- Install the nodes needed by that workflow. Use your source’s Downloads & requirements from step 7 and ComfyUI’s node installer/Manager. Native Official Gen’s bundled graph requires ComfyUI Impact Pack, ComfyUI Inspire Pack and KJ Nodes for ComfyUI. Other workflows have their own requirements. Restart ComfyUI after installing nodes. You do not need the video dependencies just to start with still pictures.
- Start ComfyUI and keep it running. Note its actual address, commonly http://127.0.0.1:8188 . Check that its required models appear and that the intended pipeline can make one picture. A missing-model/node error means this step is not finished yet.
- Start your chosen image server and open the game. Use the matching row here. Keep its terminal open. ComfyUI renders the job; the image server passes game requests to it. Run only one game-facing image server at a time. A custom workflow uses the active source’s launcher; it does not require a second image server.
Which launcher and game page?
Image source | Start it | Open the game | Base game + Official Gen | Use its provided launcher, such as wayward-imagegen/start-image-generator.bat . If no launcher is provided, open a terminal in wayward-imagegen and run bun src/cli.ts . | Open the game’s index.html . Select Settings → Image generation → Automatic for the default local Official Gen port 8189 , then Check again. Use Specific address if you changed that port. | Settings UI Mod | Current bundled launcher: Start-Wayward-Bridge.bat , beside wayward-bridge.js and index.html. | Open its printed game address, normally http://127.0.0.1:8788/ , and use the Pippa button in-game. | Custom Image-Gen Server / another mod | Use the launcher shipped by that build. A file called server.bat is specific to builds that provide it. | Use that build’s printed address or its instructions for connecting the normal game. Its port may differ. |
- Connect the server to ComfyUI and scan the models. Set the source’s ComfyUI address to the address from step 11. In Settings UI Mod, open Pippa → Config, set its models-folder path and scan. In Official Gen, use its config or Lab’s Official Gen → Connection & folders; its available filenames come from ComfyUI. For another source, use its own equivalent settings. Use the actual model-storage folder, not the folder containing the desktop shortcut.
- Select your model files and save. Settings UI Mod: Pippa → Pipeline & models, select Anima or Illustrious XL, then the downloaded Anima model in UNET or an Illustrious/SDXL model in checkpoint, plus the matching encoder/VAE where required. In Settings UI Mod 4.15.13 and newer, each model family also has its own steps and CFG there (used when Turbo is off): set them to what your model's page recommends. Official Gen: choose the compatible checkpoint/LoRA through Lab’s Models & LoRAs or its config. Other sources: use their corresponding pickers. Keep a checkpoint’s own VAE when that is what the workflow/model expects. If the game has a per-character image-model choice, check that too: it can override the server’s default checkpoint.
Part D — Add the prompt, save and test
- Add the LoRA through a supported slot. Settings UI Mod and compatible custom graphs can have character/style LoRA slots. Select the file, enable it and use its recommended strength. Native Official Gen has its own configured LoRA slot, commonly occupied by its acceleration LoRA; adding extra character/style LoRAs can require a compatible custom workflow. Keep acceleration LoRAs and their prescribed steps/CFG together.
- Put the trigger words in the right prompt. In Settings UI Mod, open Heroines & outfits: character appearance/trigger words go in imageBase; clothing words go in Clothing / outfit prompts for the selected outfit. In the base game, use the custom-character builder’s appearance/additional tags where offered; advanced prompt inputs can live in your custom workflow. Other sources have their own character/outfit prompt fields. Only paste a LoRA’s trigger words when needed, and only if that LoRA is loaded. Avoid copying the example outfit into the permanent appearance prompt unless you want it on every picture.
- Apply the changes for your source. Settings UI Mod: check the selected heroine/save and press Save & Apply. Official Gen: save wayward-imagegen.config.json , restart its server for startup config changes, then use the game’s Check again. Other servers: use their save/reload procedure. Simply refreshing the browser is not a universal way to apply server settings.
- Make one new picture, then experiment. Use the source’s test/new-picture action, or a custom-character preview. Existing cached pictures may stay unchanged until explicitly regenerated. For a comparison, keep the same prompt, seed, size, steps, CFG and sampler and change one model/LoRA at a time. Copy settings from a gallery example only when you also understand its model and dependencies. Leave animation off until still pictures work; in current Settings UI Mod that control is Animation → Animate → nothing (only the GIF button). Do not restore every scene’s defaults as a routine setup step, because that can remove your custom prompt edits.
Screenshots: base game with Official GenLab → Official Gen → Connection & folders Use the Official Gen source’s connection and model controls. Save config changes and restart that server when requested.
Screenshot: docs/screenshots/official-connection.png
Normal game → Settings → Image generation Automatic finds the default local server. Check again after restarting it, or use Specific address for a different port.
Screenshot: docs/screenshots/native-settings.png
Screenshots: selecting models and LoRAsAnima: UNET, text encoder and VAE The basic setup uses three matching files. Select the actual file you downloaded in each field. This screenshot shows the base model, not Turbo.
Screenshot: docs/screenshots/beginner-anima-pickers.png
Illustrious: checkpoint and matching LoRAs Select an Illustrious checkpoint, then only LoRAs made for that model type. These quick edits are saved to the selected Settings UI Mod character/save.
Screenshot: docs/screenshots/beginner-model-pickers.png
Reference: the bundled wayward-imagegen/README.md and package/launcher describe Official Gen’s Bun runtime, native graph requirements and connection procedure. The bundled Settings UI Mod launcher runs Node. Folder roles follow ComfyUI’s model documentation (https://github.com/Comfy-Org/docs/blob/main/basic-concepts/models.mdx). A successful setup check is not proof that every scene or model combination works.
Official GenSettings UI ModCustom Image-Gen ServerModel-file meaningsShare your setup with a pack
Model guide
In Lab: Help > Guides > Model guide. This is independent of Custom workflow and works
without loading a graph. Each generation area also links to the same guide; actual selected-file descriptions/status remain beside its picker.
WHAT DO ALL THESE FILES MEAN? Workflow JSON: the recipe (nodes, connections and selected filenames). It does not contain the model weights or all generated pictures. Checkpoint: the main image model, often bundling text encoder and VAE outputs. Diffusion model / UNET: the part generating image/video latents. UNET can be a loader/folder label rather than the precise network architecture. Text encoder / CLIP / T5 / Qwen: processes prompt words into conditioning for that pipeline. A CLIPLoader node can load a non-CLIP encoder such as T5/Qwen. VAE: converts pixels to latents and latents to visible images/frames. The workflow can use a checkpoint's connected VAE output or load an appropriate separate file. LoRA: a smaller learned add-on for a character, style, effect or faster sampling. Load a compatible one and set strength; its trigger words alone do not load it. Vision encoder / IP-Adapter: processes reference pictures and guides generation. Embedding / textual inversion: a learned concept referenced in the prompt, with its own compatible base family. Detector / segmentation: locates faces, objects or regions for cropping/masking. Upscaler: enlarges/refines an image when a workflow includes that stage. Video/motion models: generate frames over time. Wan 2.2's high/low models cover different denoising stages; the default graph needs both. Custom nodes: ComfyUI extensions providing operations or loaders. Install into custom_nodes and restart ComfyUI; they are not model weights.
Managing LoRAs (optional)
LoRA Manager (by willmiao) is an optional ComfyUI add-on for managing LoRAs: it shows them with preview pictures, downloads LoRAs and checkpoints from Civitai, and fetches their trigger words and pages. Install it from ComfyUI Manager > Custom Nodes Manager (search "LoRA Manager"), then restart ComfyUI. Lab reads what it saves, so its trigger words and pictures show in Lab's LoRA lists. It is separate: Lab and the image servers work without it. https://github.com/willmiao/ComfyUI-Lora-Manager
Concrete example: basic Anima
anima-base-v1.0.safetensors -> main diffusion model. qwen_3_06b_base.safetensors -> text encoder supplied for Anima. qwen_image_vae.safetensors -> the Qwen-Image VAE also used by Anima. Anima-compatible LoRA -> optional character/style/effect add-on. These files have different jobs despite sharing an extension. The shared VAE's name does not make the entire graph a Qwen-Image pipeline.
Concrete example: checkpoint VAE
A checkpoint loader supplies MODEL, CLIP and VAE outputs. If decoding uses its VAE output, switching the checkpoint changes the VAE it uses too. You do not need a separate VAE file just to complete that graph. A workflow/source can instead load one separately when supported. Official Gen's native still pipeline uses its checkpoint VAE, while its separate Wan video graph has its own VAE setup.
Formats are not pipelines
.safetensors can contain a checkpoint, diffusion model, encoder, VAE or LoRA. .gguf is a weight format needing a matching loader; FP8 describes precision. Those choices affect loaders/memory, not the intended base family by themselves. Renaming .gguf to .safetensors does not convert it. pack.json maps existing game pictures; it is neither model weights nor a ComfyUI workflow. Media files are inputs/outputs. Recovering workflow instructions from one depends on retained metadata, not just its extension.
What Lab's labels mean
Installed -> found locally or listed by ComfyUI; not a compatibility guarantee. Declared base -> metadata/notes identify the intended model family. Detected architecture -> supported tensor-header signatures identify a family; SDXL structure alone cannot establish Illustrious/Pony/NoobAI tuning. Published component -> a documented component filename was recognized; the file may be shared by multiple pipelines. Unknown/conflicting -> not enough evidence, or sources disagree. Check the model's original download page/intended base rather than assuming the current selection is correct. Lab does not invent missing author descriptions.
Settings UI Mod has fuller model/LoRA management, trigger-word storage, previews and other features. Lab provides quick edits to some shared settings; Save to Settings UI Mod applies those changes. General help is here; file-specific details stay near each selected file. Common problems are in docs/FAQ.txt.
Reference workflow
In Lab: Workflows > Custom workflows > Custom workflow, with a workflow open.
For a one-pass workflow (one sampler, a positive prompt that ends with {{positive}}), Lab shows a short editor above the nodes: Make a picture. Other workflows show all their nodes in groups instead (Models & LoRAs, Prompts, Sampling & size, Images & references, Output, Other nodes), with every input still editable. Workflows with several passes stay in the node editor under Advanced tools.
The game still gives each picture its prompt, seed and size. These controls don't change the picture size and don't add an upscale or detail pass.
Image source
Generation mode:
- New image: the picture is drawn fresh. Best for different scenes, poses and clothes.
- Use a reference image: the picture starts from one you approved, so more of it stays (face, pose, clothes, background).
For a reference: Choose image... picks a PNG, JPEG or WebP (up to 20 MB) and puts it in ComfyUI's input folder for you. Browse uploaded images shows the pictures already there. Clear image goes back to New image. Reference influence: higher keeps more of the reference. Lower it when she should wear something else or stand differently. A high value can keep drawing the reference's clothes even when the game asks for another outfit; use New image for those.
Character & style
- Character LoRA and its strength: her identity and clothing detail.
- Permanent appearance / style tags: what never changes (her look, the style). Leave clothes out: the game adds the scene and her outfit by itself.
- Start with a different character clears the LoRA, the tags and the reference, ready for the next one. It doesn't change her record in the game.
Generation
Steps and CFG. Sampler & model files holds the diffusion model, text encoder, VAE, sampler and scheduler. Output & extra stages lists any extra nodes the workflow already has (a face fix, an upscale...). They are kept as they are; change them in the node editor.
Save and try
Save changes keeps the workflow in your library. Try a picture saves first, then opens the Workflow test with this workflow, so you can check the prompt before pressing Run test.
Her look and default outfit, in Settings UI Mod
Open "Character & default outfit" and press Read selected heroine. Check the name shown before you change anything. Her appearance, her negative prompt and her normal outfit each have their own Save button; each saves for that heroine only, in the same settings the game uses. Settings UI Mod has to be running (its menu doesn't need to be open). Reading changes nothing. Clothing-state prompts opens Lab's editor for the mod's clothing states.
Settings UI Mod guide
Written for Settings UI Mod 4.15.22 and game version 2026-11-09. (4.15.21 fixed a picture stuck at "1/1" although more versions were drawn: the mod now puts back every version it still has a picture of. 4.15.22 added Actions in the Workspace and a note when the browser's storage is nearly full.) Pictures: docs\screenshots\mod-new-menu.png, mod-config.png, mod-animation.png and beginner-native-pipeline.png show the screens named below.
Settings UI Mod is a separate mod by its own author. Lab works beside it but does not replace it: Lab helps you pick models, test workflows and build packs; the mod is what draws the pictures while you play. This guide covers the mod's own menu. For Lab's part, see LAB-START-HERE.txt.
Contents
1. What the mod does
2. Start it
3. The whole mod at a glance (its buttons and what each opens)
4. First-time setup (10 minutes)
5. How the settings menu is laid out
6. The two settings menus (classic and new)
7. Every settings tab, one by one
8. The game's picture: its buttons
9. Daily Happenings (the Elena button)
10. The Workshop (the Mara button) 11. Other things the mod adds to the game 12. Everyday jobs, step by step 13. What is saved where 14. Keyboard shortcuts 15. When something goes wrong 16. Using it together with Lab 17. Can the mod do it? Quick answers to "can I ...?" 18. Where is it? Find a feature by what you want to do 19. Old menu and new menu: what's different 20. Extras: what each optional feature needs in ComfyUI 21. Reporting a problem with the mod 22. Image packs with the mod: draw, check, share, add 23. Every button, place by place (Pippa, Elena, Mara and the Workshop)
1. WHAT THE MOD DOES
The mod does three big things:
- Pictures: the game asks for a picture whenever a scene changes. With the mod running, those pictures are drawn by ComfyUI on your computer, from prompts you control: her looks, her outfits, each scene, each room, her mood, the men in the scene. It also saves pictures to reuse, makes and adds image packs, animates pictures (GIF or animated WebP) and keeps copies of your saves.
- Daily Happenings: more life in the base game. She makes her own decisions, keeps secrets, has nights away, theme nights, busy hours and brawls, a debt that bites, Rowan's house, and more. It can be switched off.
- The Workshop: make new things for the game yourself (events, stories, people, places, clothes, fights, whole projects) and share them.
The mod never starts or stops ComfyUI. Start ComfyUI yourself first.
2. START IT
- Start ComfyUI and wait until it says it is ready (http://127.0.0.1:8188).
- Start the mod: double-click Start-Wayward-Bridge.bat in the game folder. It needs Node.js. Its window shows: Wayward Bridge 4.15.19 Play at http://127.0.0.1:8788/
- Open http://127.0.0.1:8788/ in your browser and play from there. Keep the mod's window open while you play; closing it stops the pictures.
If the window says "ComfyUI is not reachable yet", ComfyUI isn't running or uses another address. You can still open the game and fix the address in Config (section 4).
The first time a new version starts, a "What's new in 4.15" box lists the changes. Each line has a Go button that opens the place it talks about. Config shows that box again later.
3. THE WHOLE MOD AT A GLANCE
Start or load a game. The mod adds floating buttons at the bottom corners (you can drag them somewhere else). Their names are fixed labels, whoever you play:
Pippa (bottom right) the settings menu: pictures, prompts, models,
animation, packs, backups. Sections 4-7.
Mara (above Pippa) the Workshop: make new things for the game.
It opens the Workshop Workspace in a tab of its own.
Section 10.
Elena (bottom left) Daily Happenings: her mind and life in the base game.
Section 9.
Also:
- On the game's picture: buttons to draw again, keep, delete, flip between saved pictures, animate, and see the prompt. Section 8.
- Screen layout (top right of the game): move and resize the game's panels.
- What's new: shown once after each update, and again from Config.
- Under her card: her stance tonight, how sharp you are, Watch from the bar, Have a drink, and the scene with its buttons.
The game's own Settings button (top right) is the game's settings, not the mod's.
Nothing in the settings menu changes until you press Save & Apply to game.
4. FIRST-TIME SETUP (10 MINUTES)
Picture: mod-config.png (Config tab).
1. Open the menu and go to Config. 2. ComfyUI: check the ComfyUI address (normally http://127.0.0.1:8188) and
press Check ComfyUI. "offline" means ComfyUI isn't running or the address
is wrong.
3. Press Check my ComfyUI (small tests). It draws a few tiny test pictures
and says in plain words what is missing (a node, a model, a setting).
4. ComfyUI models folder: set Folder path to your ComfyUI\models folder, for
example F:\ComfyUI\models. Set Output folder to ComfyUI\output if you want
saved scenes and GIFs there; leave it blank to keep them in
wayward-bridge-data. Press Save path & scan. Now LoRAs are grouped and
their trigger words are found.
5. ffmpeg path (only for animations): the full path to ffmpeg.exe, for
example C:\ffmpeg\bin\ffmpeg.exe. Blank means "look for it on PATH".
6. Go to Pipeline & models (picture: beginner-native-pipeline.png). Pick the
model family you use: Anima / Z-Image, Illustrious XL, Pony or NoobAI.
Then pick its model files under Model / sampler presets.
7. Press Test pipeline (under Test this pipeline). One small picture is
drawn; if it fails it says why.
8. Press Save & Apply to game.
That's it. Play, and pictures appear as scenes change. Everything else in this guide is for when you want to change how they look.
Not sure which models to use? Lab's MODEL-GUIDE.txt and TRY-MODELS.txt help. Set the model you pick here in the mod (Pipeline & models, section 7); Lab is the editor only for a custom workflow (section 16).
5. HOW THE SETTINGS MENU IS LAID OUT
Top bar
- HEROINE: whose settings you are editing. "in game" marks the one being played. You can edit another heroine without switching games.
- Search prompts & settings: type a word (a tag, "ffmpeg", "WebP") to jump to where it is.
- Workshop Workspace: opens the Workshop in its own tab.
- Try the new menu / Back to the old menu: switches between the two menus (section 6). Both have the same settings.
- The line under it says where prompt changes are saved: for this heroine in the save being played (its save ID).
"In game" strip
- What the game is showing right now: heroine, task, outfit, exposure, mood, room. Click a part to jump to its settings. inspect opens the Prompt inspector for that picture; the round arrow refreshes the strip.
Bottom bar
- "No unsaved changes" or "1 unsaved change" with undo.
- redraw current picture: when ticked, Save & Apply also redraws the picture on screen so you see the change at once.
- Discard: forget the changes since the last save.
- Save & Apply to game: keep the changes and use them.
Inside a tab, each part is a box with a title. Click the title to fold it. Fold all / Open all at the top right of a tab fold or open every box. "More" links open a longer explanation. "Restore default" in a box only resets that box.
"Prompt from" (in many boxes): My settings = your own text. Picking an add-on there makes that box use the add-on's text instead (see Add-ons).
6. THE TWO SETTINGS MENUS (CLASSIC AND NEW)
Picture: mod-new-menu.png. The classic menu has tabs across the top. The new menu (4.15.10 and later) has a side list and a Home page that shows what is on screen, what you changed last, and where everything is. Same settings, different names:
Classic tab New menu page
--------------------- ------------------------------
Heroines & outfits Her looks
Scenes & NPCs Scenes and people
Locations Places
Negatives & tag rules Prompt rules
Bulk edit Change many at once
Pipeline & models Models and LoRAs
Animation Animation
Prompt inspector Prompt Inspector
Config Setup and backups, plus Pictures and image packs
(saved pictures, packs, drawing ahead, her face)
The new menu also links straight to the game's Daily Happenings, the Workshop and Projects & sharing. Lab's guides use the classic names.
7. EVERY SETTINGS TAB, ONE BY ONE
HEROINES & OUTFITS (new menu: Her looks)
- Character prompts: imageBase is how she looks (hair, eyes, body); imageNegative is what should never be on her. The name box renames her in the game too, after Save & Apply. Presets (Save As / Load) keep looks you like under a name; groups sort the presets.
- Only when exposed: tags that are only drawn when that part of her is bare (a navel piercing, a tattoo, pubic hair). While her clothes cover it, the tag is left out. "Never left out" overrides that.
- LoRA trigger words: click a word or group to add it to / take it out of her prompt. Faded = not in the prompt. Comes from the LoRAs in Pipeline & models.
- Clothing / outfit prompts: pick an outfit ("in game" marks the one she wears), then edit imageTags (what the outfit looks like) and negativeImageTags. Reset outfit / Restore all outfits undo your edits.
- Exposure: Top, Skirt and Underwear together make one exposure (covered, bodice open, topless, skirt up, nude ...). The box is the prompt for that one. Empty = keep the game's words. "For" chooses every outfit or only the selected one. Load exposure from current image picks what is on screen.
- Heroine state: the parts the game adds as the scene changes (Quality, Arousal, Aftermath, Gaze, Intoxication, Reaction, Focus, Narration, Gender, Group, Pose ...). Each can be the game's, your own text, or removed. "Live:" shows what is used right now.
SCENES & NPCS (new menu: Scenes and people)
- Currently active image: the picture key of the scene on screen (heroine__task__outfit__room__v-N) with Edit buttons for each part, Refresh live and Show prompt.
- Mood prompts: tags for each mood (content, shy, scared ...). Scenes use their own mood if they have one, else her mood meter.
- Scene prompts: the act tags of each scene (only the act; room, mood and exposure come from their own boxes). "+ Add a variant" gives a scene more than one prompt to pick from. "The game's own prompt" draws a scene from the game's prompt word for word, for every scene or just this one.
- Hidden scenes: tags added when she keeps a scene hidden, when you peek, or when she is touched (12 kinds).
- Scene participants (the men): who else is in the picture, for all scenes, patron scenes, scenes with you, or group scenes (two or more men). Start a box with a head count (1boy, 2boys, 1girl ...) and it replaces the game's count. Models often mix up who is who when two people share a picture; expect to redraw some.
- Named patrons in the tavern: give one man his own look and keep a face you like (Try a new face).
- NPC / patron portraits: the small round pictures of the men. Archetype tags per kind of man.
- Only when he is exposed: like hers, for the men.
LOCATIONS (new menu: Places)
- Location prompts for every room of the tavern and every place she goes.
NEGATIVES & TAG RULES (new menu: Prompt rules)
- The game's negative prompt: the base list for every picture, plus Clothes & exposure extras. "Same game negatives for all heroines" shares it.
- Tag rules: find a tag anywhere in the finished prompt (or negative) and replace or remove it, or with Find empty, always add tags. Use this when a word keeps showing up and you can't find where it comes from.
- Suggestions from your ratings: after a few thumbs up / thumbs down, the mod suggests tags to add or remove.
BULK EDIT (new menu: Change many at once)
- Many prompts at once: Scenes, Scene moods, Moods, Work moods, Outfits, Rooms, Looks, Tasks, Exposure. Find & replace across all of them, add or take out a tag everywhere, or edit the list as text (or as a .txt file). "Changes go to" says whose settings change.
PIPELINE & MODELS (new menu: Models and LoRAs)
- Settings profiles: everything on this tab under one name. Keep a "Fast" and a "Best quality" and switch: Load, then Save & Apply.
- Pipeline - model family: Anima / Z-Image, Illustrious XL, Pony, NoobAI. Each family remembers its own models, sampler, steps, CFG, turbo, quality tags and LoRAs; switching brings them back. "Draw with the game's look" uses what the game's own image server uses, so the mod's pictures match the game's art.
- Anima TeaCache (optional, Anima only): speeds drawing if the node is installed; skipped if not.
- Test this pipeline: one small test picture, with plain-word errors.
- Model / sampler presets: the model files (unet / clip / vae for Anima, a checkpoint for the SDXL families), sampler, scheduler, steps and CFG. Blank steps / CFG use the family's default (Anima: 30 steps, CFG 4). Lists come from ComfyUI, so start it and press Check.
- LoRA models & trigger words: + adds a LoRA, with its strength. Which LoRAs are used is per heroine; trigger words are shared.
- Turbo (steps / CFG override): for a lightning / turbo LoRA or model. When on, these steps and CFG are used instead (usually 4-8 steps, CFG 1-2).
- Picture size & quality tags: scene and portrait sizes. The tab tells you if a size suits the model family. "measure again" suggests a size from the game's picture box. Quality tags and the negative prompt for this family.
- Picture size for single scenes: another shape for some scenes (lying in bed, tall standing poses).
- Hi-res fix: draws, scales up, then adds detail. Denoise 0.3-0.5 keeps the picture.
- Face & hand fix: redraws faces and hands bigger, so small ones come out clean. Needs the detailer nodes in ComfyUI.
- Cut-out figures: sprites drawn on white and cut out, for visual novels.
ANIMATION (new menu: Animation) Picture: mod-animation.png. Needs WAN 2.2 image-to-video in ComfyUI and ffmpeg. "What you need for animations" lists the files.
- What gets animated - Animate: nothing (only the GIF button) only when you press the GIF button scenes with their own motion Workshop scenes that have a motion every scene picture every new scene picture It also shows how long an animation takes on your computer. Match the motion to the picture: the motion buttons pick a fitting template. Her own motion: a motion just for this heroine.
- Motion: templates (Everyday, At work, Shows & nights, Romance, Love scenes, Battle) fill in the motion prompt, speed and length. "Try these settings for 1 second" tests quickly. FPS and Seconds set the length.
- Size, file and loop: Quick / Sharp / HD / Same size as the picture; Animated WebP (smaller, every colour) or GIF (256 colours); smoother motion (needs frame interpolation nodes); keep the mouth still; perfect loop.
- Motion LoRA: WAN 2.2 alone barely knows love-scene movement. A motion LoRA teaches it. Set one for all love scenes, or one per position. Stronger motion makes movements bigger.
- Models and sampler: the WAN files and sampler. Empty / 0 keeps the workflow's own. "Use the 4-step speed LoRAs" makes it about a minute each.
- Restore animation defaults if it gets confusing.
PROMPT INSPECTOR (new menu: Prompt Inspector) Every picture drawn this session, with exactly how its prompt was built: each part, where it came from (her looks, outfit, scene, tag rule ...), what was left out (covered by her clothes) and what a box replaced. This is the first place to look when a picture shows something you didn't ask for.
CONFIG (new menu: Setup and backups; pictures parts under Pictures and image packs)
- Settings UI Mod 4.15.19: What's new.
- Game update check: after you put a new game version in the folder, lists the prompts the game changed and which of your edits were for the old text.
- ComfyUI: address, Check ComfyUI, Check my ComfyUI (small tests), Test an animation (tiny video).
- ComfyUI models folder: models folder, output folder, "Also keep a copy in ComfyUI's output\wayward folder" (off = each picture kept once), "Save pictures as WebP (smaller files)" (a third to a fifth the size of a .png), ffmpeg path.
- Auto-delete generated images: keeps only the newest N pictures. Saved pictures and Workshop project pictures are never auto-deleted. "Older pictures to WebP" converts the .png pictures you already have.
- Where her pictures come from: the order the game looks in: 1 Image packs -> 2 Saved pictures -> 3 Drawn new "Use Saved Images/GIFs" turns on 2. "A new picture each time a scene comes back" draws again instead of showing the one from earlier this session. Add a pack... puts a pack .zip (or an unzipped pack.json) into the game. Share a heroine as a pack makes an image pack of the pictures the mod drew of her (the newest of each scene, outfit, room and mood). The pictures you kept with the thumbs up (Saved pictures) are a separate collection: the newest drawing goes in, not necessarily the one you kept.
- Pictures ahead of time: draw before the game needs them, so nothing waits while you play. Generate all scenes for her in each outfit / room / mood / exposure you leave in (Ctrl+click to leave some out). Only fill gaps skips ones you have. Also: her pictures as an image pack, copy a pack's pictures into Saved pictures, Review pictures (J keep, K redo).
- Animation queue: animate right away, when the game is idle, or only at night hours, so pictures come quickly while you play.
- Saved scene images / GIFs: Save current image/GIF for active scene, and Clear Saved Image (narrow it by clothes, room and mood first).
- Her face (face anchor): one face picture every picture of her is drawn toward, so she looks the same. Needs IPAdapter FaceID nodes.
- Browse saved pictures, per heroine and scene.
- Send pictures to the game's makers: makes a folder of the pictures you kept, to send yourself if you want to. Nothing is sent by the mod.
- Speed and shortcuts: whole-tag Backspace / Delete, Ctrl+Up / Ctrl+Down for tag weight, keyboard keys on the game's picture (section 14).
- Misc: Always see what she does, money as yen, truly random dice, Workshop sounds and volume.
- Add-ons: bundle prompts (tick which parts) into an add-on, import add-ons, use several at once (lowest in the list wins).
- Save ID: each game has its own save ID and its own settings. "All saves the mod knows" lists them.
- Backup & move to another PC: Export everything / Import a backup (all settings, presets and add-ons, not the pictures).
- Server & defaults: restore defaults, Undo last save, import an old config.json or presets.json from the old mod.
- The Workshop: Workspace in its own tab, or the classic Workshop in the game.
- Copies of your save (on this PC): every 10 minutes and when you leave the page, plus a weekly zip (last 8 kept). Copy it now, download every copy, bring copies back from a zip.
- Diagnostics: mod version, game version, hooks, ComfyUI status, picture library and data folder. Read-only: no buttons, fixes nothing, and it shows full folder paths (see section 19). For a report, use the bug report.
8. THE GAME'S PICTURE: ITS BUTTONS
Hover over (or tap) the scene picture in the game. Some buttons are the mod's, some are the game's own.
The mod's bar on the picture:
- Generate new (Alt+G): draw this picture again with a new seed. The old one stays in the history. Always draws, even with Use Saved Images/GIFs on.
- Same seed (Alt+S): draw it again with the same seed, so only your setting changes show. Use it right after a prompt edit.
- Show prompt (Alt+P): how this picture's prompt was built (the Prompt inspector, for this picture).
- Edit (Alt+E): opens this picture's prompts in the mod's menu, at the right place to change them.
- What's on screen?: names what the game is showing (heroine, scene, outfit, room, mood), so you know which prompt to change.
- < Saved > arrows (Alt+Left / Alt+Right, or Alt+A / Alt+D): flip between the saved pictures and GIFs of this scene (after the last it goes back to the first).
- Animate (the film button): animate this one picture now, with the Animation tab's settings. The animation is kept with the scene's pictures; the still stays. "Waiting in the animation queue" means it's coming.
The game's own buttons:
- Thumbs up (the mod adds Alt+K): keep it. Kept pictures show in Browse saved pictures and come back with Use Saved Images/GIFs.
- Thumbs down, "Mark as bad and regenerate" (the mod adds Alt+X or Alt+Delete): delete it and draw a new one.
- + (Draw another version): another version of the scene (2/2, 3/3...). If a picture for that version already exists, it is shown instead of drawn.
- Copy bug report (bottom right, only after the game has caught an error): the game's own report, for the game's maker.
Elsewhere:
- Review pictures (Config > Pictures ahead of time > Review pictures (Keep / Redo)): a page of its own at http://127.0.0.1:8788/bridge/review. One at a time (J keep, K redo) or By scene (every drawing of a scene side by side), for the game's scenes, the mod's own pictures and image packs. Lab opens it too: Settings UI Mod > Review pictures.
- Screen layout (top right of the game): move and resize the game's panels; pick the game's own layout to undo the mod's arrangement.
9. DAILY HAPPENINGS (THE ELENA BUTTON)
Her mind and life in the base game: what you tell her, her asking you first, her reports, hidden scenes, the patrons, her nights, her tastes. A switch at the top turns all of it off (and on again), whatever project is ticked. Its tabs:
- Her mind: what she and the others do, how hidden she keeps things, what people remember, the built-in actions.
- Her own person: she makes her own decisions (personality and loyalty first, then what you ask, money, the tavern and the men). Busy hours and brawls play out by her choices; you watch, nudge, encourage or stop it. Also: the debt and the books, blackmail, nights away, men who get too close in front of you, a rival keeper, Rowan's house (buy it, run it, the girls), and what is on her mind.
- Her life: the world around her, getting closer to her.
- Body & tastes: her days and her body, her tastes in men (optional), who comes in.
- Nights & hearts: theme nights, busy hours, tavern brawls (start one now), a man who falls for her (optional).
- Nights & costumes: the nights and the costumes.
- In this game: settings for this playthrough (what she is to you, what she calls you), the regulars and her moves, your theme nights, the week's ledger.
- Her reports: what she did out of your sight; "Always tell me" tells everything from the last week.
In the game itself you'll also meet: Plan a theme night and Ask her to keep the books (under Her afternoon), listening at her shut door, keep watching after a peek, the herbwife's eyebright drops, Watch from the bar and Have a drink under her card.
10. THE WORKSHOP (THE MARA BUTTON)
Make new things for the game and share them as projects. It opens the Workshop Workspace in a tab of its own; Config > The Workshop can switch to the classic Workshop inside the game. New to it: Projects & sharing > Make me a project... (a wizard), and the Glossary explains the words. Its eight groups:
- Events: all events, her asks, requests, routines at set times, actions of your own (they can take your time or hers, be a moment or a finish in the game's scenes with her), random encounters, quests.
- Story: visual novels, shows, scenarios (a place and its keeper, like Rowan's house), lines (your own wording for the mod's lines, with Try it), flags & her stats.
- Her: her stats (change her stats and traits as the game keeps them), her own person (her nights away, the men in front of you, answers left to her).
- Places: scenes and their pictures, rooms (the game's and rooms of your own, like a garden or a cellar, that count as real places), maps, dungeons, adventures.
- People: characters, tavern patrons (kinds of patron of your own, with their own days and lines), minds (what people do by themselves), bonds.
- Battle & magic: battle rules, the battle screen, skills & moves, spells, statuses & elements.
- Things: clothes (with how they come undone, drawn only at some hours or days, left out of her pictures in the bath, bed or intimate scenes), items & gear, pack & equipment, shops, sounds.
- Projects & sharing: tick projects on the shelf, share them and make bundles (with their pictures and GIFs), make one its own game, a theme (rename the game's words, a day and ledger of its own), art, a map of the project, Try it.
Built-in projects: Tavern Nights, Tavern Brawls and Rowan's House. Daily Happenings uses them, so unticking one asks first. When the mod updates them, your changes carry over.
11. OTHER THINGS THE MOD ADDS TO THE GAME
- Screen layout (top right): her picture as a tall panel, the rest in quadrants; drag borders, swap panels, or pick a ready layout; kept per screen size.
- Fights: tavern brawls and her own fight on a battle screen, with moves from her record, titles, lust and urges.
- Copies of your saves every 10 minutes, and a weekly zip (Config).
- Browser translation of the game page works.
- Money in pictures as yen, truly random dice, always see what she does, and Workshop sounds (Config > Misc).
12. EVERYDAY JOBS, STEP BY STEP
Change her hair, eyes or body
Heroines & outfits > Character prompts > imageBase. Change the tags, tick
redraw current picture, Save & Apply to game.
Change how an outfit looks
Get her into that outfit in the game (or pick it in the outfit list), then
Heroines & outfits > Clothing / outfit prompts > imageTags. Save & Apply.
A scene keeps drawing the wrong thing
Press inspect in the "In game" strip. The Prompt inspector shows which part
brought the wrong tag. Fix it there (Scene prompts, outfit, Heroine state),
or add a Tag rule to remove it everywhere.
A tag shows up everywhere and you want it gone
Negatives & tag rules > Tag rules > + Add rule for the prompt. Find: the tag,
Replace: empty. Or + Add rule for the negative to add it to the negative.
Keep pictures you like and see them again
Thumbs up (or Alt+K) on the game's picture saves it. Config > Where her
pictures come from > tick Use Saved Images/GIFs. The < Saved > arrows on the
picture flip between saved ones.
Draw everything before you play
(Making a whole image pack: section 22.)
Config > Pictures ahead of time > The game's scenes. Choose what varies,
tick Only fill gaps, Generate all scenes. Turn on Use Saved Images/GIFs.
To skip what you never want (an outfit you dislike, a room, a mood, a
scene): Ctrl+click it in its list to leave it out, or untick Vary outfits
and pick the one outfit to draw in. Exclude scenes leaves out scenes. The
confirm box says how many pictures that makes before anything starts.
Add an image pack someone sent you
Config > Where her pictures come from > Add a pack... and pick the .zip.
(Or in the game: title screen > Image packs. See SHARE-A-PACK.txt.)
Share your heroine as an image pack
Config > Where her pictures come from > Share a heroine as a pack: a .zip
with pack.json, ready to add in any game. Lab's Share a pack guide adds
the workflow and a README for the people you share it with
(SHARE-A-PACK.txt).
Make her look the same in every picture
Use her character LoRA (Pipeline & models > LoRA models) and/or
Config > Her face (face anchor) > Use the picture on screen, Strength 0.7.
Animate a picture
First time: Animation > Animate: nothing (only the GIF button), Save & Apply.
Press the GIF button on a game picture. Once that works, choose what gets
animated automatically. See GIFS-AND-VIDEOS.txt.
Fast while testing, best quality for keeps
Pipeline & models > Settings profiles. Set up a fast setup, Save As "Fast";
a slow one, Save As "Best". Load one, Save & Apply.
Smaller picture files
Config > Save pictures as WebP (smaller files) for new pictures, and
Auto-delete > Older pictures to WebP for the ones you have.
Don't run out of disk
Config > Auto-delete generated images, with a Max images number. Saved and
project pictures are never touched.
Back up, or move to another computer
Config > Backup & move to another PC > Export everything. On the other
computer: Import a backup. Pictures are separate: copy the
wayward-bridge-data folder too if you want them.
After updating the game
Config > Game update check lists prompts the game changed and which of your
edits were made for the old text.
13. WHAT IS SAVED WHERE
- Prompts (her looks, outfits, scenes ...) are saved for that heroine in the game being played (its save ID). Another game with the same heroine has its own copy; a new game starts from her most recent settings.
- Model family, models, sampler, which LoRAs, turbo: per heroine.
- LoRA trigger words and groups, picture sizes, quality tags, hi-res fix, face fix and animation: shared by all heroines.
- Everything lives in wayward-bridge-data next to the game. Back it up with Export everything (settings) and by copying the folder (pictures).
14. KEYBOARD SHORTCUTS
On the game's picture (not while typing or while the menu is open):
Alt+G Generate new
Alt+S Same seed (redraw with the same seed)
Alt+Left/Right Saved pictures (or Alt+A / Alt+D)
Alt+K Save (thumbs up)
Alt+Delete Delete (thumbs down) (or Alt+X)
Alt+P Show prompt
Alt+E Edit
In prompt boxes:
Backspace / Delete remove a whole tag (Shift+Backspace for one letter),
when turned on in Config > Speed and shortcuts.
Ctrl+Up / Ctrl+Down raise / lower a tag's weight: blush -> (blush:1.1).
In Review pictures: J keep, K redo.
15. WHEN SOMETHING GOES WRONG
No pictures at all
Is ComfyUI running? Config > ComfyUI > Check ComfyUI. Then Check my ComfyUI
(small tests): it names the missing node or model.
Pictures fail with a model error
Pipeline & models > Test this pipeline. Check the model files match the
model family you picked.
Wrong things in a picture
Prompt inspector (inspect in the "In game" strip).
Two people get mixed up
A limit of the image models. Name each person in Scene participants
("dark-skinned male", "old man"), use her LoRA, redraw.
"ffmpeg not found"
Config > ffmpeg path, full path to ffmpeg.exe.
Animations are slow
Animation > Models and sampler > Use the 4-step speed LoRAs, Size: Quick,
Animation queue: Wait until the game is idle.
An old picture keeps coming back
Config > Where her pictures come from: an image pack or a saved picture is
shown before anything is drawn. Untick Use Saved Images/GIFs, or tick
A new picture each time a scene comes back.
Changes don't show
You need Save & Apply to game. Tick redraw current picture to see it at once.
Still stuck
Config > Diagnostics shows the mod and game versions, hooks and ComfyUI
status. Lab's FAQ.txt has more answers.
16. USING IT TOGETHER WITH LAB
Set models, LoRAs, steps and CFG here in the mod (search "model family" or "LoRA models" in Pippa); Lab reads them from here. Lab is for what the mod has no editor for: a custom workflow, which Lab shows as a short list to edit and then adds to the mod's Add-ons. Keep Lab, Settings UI Mod and ComfyUI running while a custom workflow is ticked. Lab's FAQ has "Find in Settings UI Mod" buttons: with the Wayward Lab Helper on the game page they open Pippa and search for the place for you.
SHOW IT IN SETTINGS UI MOD
Keep Lab running. Open the game at Settings UI Mod's address, with the Helper on that game page. In Lab's Settings UI Mod page, the shortcuts are ComfyUI, Where her pictures come from, Scene prompts, Models and Prompt rules. Settings UI Mod's prompts also has Her look & outfits. The Connections card has Show it in Settings UI Mod beside ComfyUI.
Press the button, then switch to the game's tab. Keep it visible and wait a few seconds. The Helper presses the mod's own Pippa button, chooses the page and part, and points to the section when it can. It keeps your old or new menu choice. It does not edit or save settings.
If nothing happens, read the message in Lab. Make sure the game is open at the mod's address, not a separate index.html tab. Put the Helper on that page using Image variants, step 3. After a refresh, click its bookmark again if you use one. Close the Workshop first if the Helper asks. Press the shortcut again after adding or reloading the Helper.
You can always open Pippa yourself. In the old menu use Config > ComfyUI, Config > Where her pictures come from, Scenes & NPCs, Pipeline & models, or Negatives & tag rules. In the new menu use Setup and backups, Pictures and image packs, Scenes and people, Models and LoRAs, or Prompt rules. Leave unsaved edits in the mod's own controls; these shortcuts do not apply or discard them for you.
Connections in Lab
In Settings UI Mod > Connection & folders, use Connections. Choose who may use the mod, its Port, and Where ComfyUI is. On this computer uses ComfyUI on your computer. On another computer asks for its address and ComfyUI port. Press Check to test that address, then Save into Settings UI Mod. Check does not draw a test picture.
Start the mod before saving its ComfyUI address. Lab writes it through the mod's settings. Host and port changes go into a start file Lab uses from then on. If Restart it now appears, confirm it and wait. Open the new game address, or refresh the game page once the mod is back. Keep Only this computer unless you want devices on your home network to use it. PHONE-AND-COMPUTER.txt has the home-network steps.
Image variants with the helper
Open Game tools > Image variants & Helper > Image variants in Lab. The status line says whether Lab has heard from the Helper, whether it is only added to the game's page, or whether you need step 3 to put it there.
How a scene's pictures are counted comes first. Choose Scene prompts and image variants, apart to get the two questions. Scene prompt variants: The game picks one at random, or A new one each visit, in turn. Image variants: + adds image variants, or None: one picture per scene prompt. With None, ‹ › go through the scene prompt variants; the game's pick only sets the starting picture. More drawing choices are in Fine-tune.
All just pictures counts every picture by name, including the scene prompt pictures. The scene prompt question is hidden. Choose whether + draws with this visit's scene prompt or each scene prompt in turn. Try ‹ ›, + and Next visit on the pretend scene. It does not generate anything or alter a pack. Off leaves + and ‹ › to the game or mod.
Choices save in Lab. Refresh the game page (F5). The Helper there offers Reload now within a few seconds; Later only hides the note. If you use its bookmark, click it again after the reload. An added Helper loads again while Lab runs. A standalone Helper must be made again to take new Lab settings. README.txt has the full steps.
Settings to take to another computer
Game tools > Saves & files > Lab storage > Your settings in one file > Export my settings makes wayward-settings.json. Start the mod first to include its shared prompts, pipeline and selected preferences. It does not export personal heroine/save layers, picture files, add-ons or Workshop projects.
Open a settings file… shows each part before anything changes. Open its preview, compare Now and From the file, tick the parts you want, then press Apply the ticked parts… and confirm. Read Applied and Not applied. A part needing a tool that is not running or found stays unavailable. Restart an image server if asked, then refresh the game page.
Folder and connection settings, addresses, ports, start files, the Helper key, save ids and history are left out. Set your folders and Connections on the other computer separately. Read the exported file before sharing it: prompts are your own words. This is separate from the mod's own Export everything / Import a backup. Its pictures need a separate copy.
Lab's Help > Guides > GIFs & videos and Extras checklist cover animation and nodes. Use Lab's Share a pack for picture packs; a settings file has no pictures.
17. CAN THE MOD DO IT? QUICK ANSWERS TO "CAN I ...?"
Look for the question closest to yours. "Where" is the classic menu name; section 6 has the new menu's names.
Patrons and other people
Change how the patrons look?
Yes. Scenes & NPCs > NPC / patron portraits: quality tags, negative, and
tags for each kind of patron (Patron archetype, e.g. Grizzled veteran).
How the men look in scene pictures: Scenes & NPCs > Scene participants
(all scenes, patron scenes, scenes with you, group scenes).
Give one particular patron his own look and the same face every time?
Yes. Scenes & NPCs > Named patrons in the tavern: his own prompt and
negative; Try a new face until you like it, and it is kept.
Make the man in a scene a woman, or two men, or someone specific?
Yes. Scene participants: start the box with a head count (1girl, 2boys,
1boy ...) and describe them. Models sometimes mix up who is who; redraw.
Add new kinds of patron, or characters of my own?
Yes. Workshop > People > Tavern patrons / Characters.
Hide something on a man until he undresses?
Yes. Scenes & NPCs > Only when he is exposed.
Her looks
Change her hair, eyes, body, or her name?
Yes. Heroines & outfits > Character prompts (imageBase; the name box).
Keep her face the same in every picture?
Yes, two ways: her character LoRA (Pipeline & models > LoRA models), and
Config > Her face (face anchor), which needs IPAdapter FaceID nodes.
Show a tattoo or piercing only when that part is bare?
Yes. Heroines & outfits > Only when exposed.
Change what an outfit looks like, or how undressed she is drawn?
Yes. Heroines & outfits > Clothing / outfit prompts, and Exposure.
Add new outfits?
Yes. Workshop > Things > Clothes.
Different settings for the same heroine in two playthroughs?
Yes. Settings are kept per save ID (section 13).
Scenes, places, prompts
Change the picture of one scene, a room, or a mood?
Yes. Scenes & NPCs > Scene prompts / Mood prompts; Locations.
Several different prompts for one scene, picked at random?
Yes. Scene prompts > + Add a variant.
Draw scenes exactly as the game's own image server would?
Yes. Scene prompts > The game's own prompt, or Pipeline & models > Draw
with the game's look.
Stop a word showing up in every picture?
Yes. Negatives & tag rules > Tag rules.
Change a word in all prompts at once?
Yes. Bulk edit.
See why a picture came out the way it did?
Yes. Prompt inspector (inspect in the "In game" strip).
Add new rooms or places?
Yes. Workshop > Places > Rooms (they count as real places in the game).
A different picture shape for some scenes?
Yes. Pipeline & models > Picture size for single scenes.
Models and quality
Use Illustrious, Pony, NoobAI or Anima models?
Yes. Pipeline & models > Pipeline - model family.
Use LoRAs?
Yes. Pipeline & models > LoRA models & trigger words.
Switch between a fast and a best-quality setup?
Yes. Pipeline & models > Settings profiles.
Sharper faces and hands, bigger pictures?
Yes. Face & hand fix, and Hi-res fix (Pipeline & models).
Use my own ComfyUI workflow file?
Through Lab: Custom workflow (a slot the mod's pictures go through). See
Lab's README.
Draw without ComfyUI, or have the mod start ComfyUI?
No. ComfyUI must be running; the mod only talks to it.
Pictures and packs
Reuse pictures instead of drawing again?
Yes. Keep them (thumbs up), then Config > Use Saved Images/GIFs.
Draw everything ahead of time?
Yes. Config > Pictures ahead of time > The game's scenes.
Export my pictures as a pack (with pack.json) to share?
Yes. Her pictures as an image pack: a .zip with pack.json (section 22).
Make a full image pack of her?
Yes. Section 22: Check what's missing, Generate all scenes with Only fill
gaps, Review, then Her pictures as an image pack.
Limit which outfits a batch draws, or filter some out?
Yes. Ctrl+click them in their list to leave them out, or untick Vary
outfits and pick one; Exclude scenes for scenes. (The game's own "Make
pictures ahead of time" has no such choice.)
Add an image pack someone made?
Yes. Config > Where her pictures come from > Add a pack... (or the game's
title screen > Image packs).
Share my heroine and her pictures?
Yes. Config > Share a heroine as a pack.
Smaller picture files?
Yes. Config > Save pictures as WebP; Older pictures to WebP.
Stop pictures filling my disk?
Yes. Config > Auto-delete generated images (kept pictures are safe).
Animation
Animate pictures?
Yes. Animation tab, or the GIF button on any picture. Needs WAN 2.2 and
ffmpeg.
Real movement in love scenes?
Yes, with a WAN 2.2 motion LoRA (Animation > Motion LoRA).
Save animations as MP4?
The mod saves animated WebP or GIF. Lab's GIFs & videos can make MP4 test
clips.
The game itself
Turn the mod's extra gameplay off and just get the pictures?
Yes. Daily Happenings has an on/off switch at the top.
Make my own events, stories, fights or a whole new game?
Yes. The Workshop (section 10).
Rename the game's words or give it my own theme?
Yes. Workshop > Projects & sharing > Theme.
Move the game's panels around?
Yes. Screen layout, top right.
Make dice rolls different after loading a save?
Yes. Config > Misc > Truly random dice.
Saves, backups, other devices
Back up my settings, or move them to another computer?
Yes. Config > Backup & move to another PC > Export everything.
Get back a lost save?
Often. Config > Copies of your save (every 10 minutes, weekly zips).
Play on my phone?
The phone can play; the computer still draws. See PHONE-AND-COMPUTER.txt
and Lab's Help > Guides > Play on your phone.
Use the game's web version with the mod?
No. Play from the address the mod shows (http://127.0.0.1:8788/).
Use a translated page?
Yes. Browser translation works with the game page.
Not listed? Search the settings menu (the Search prompts & settings box at the top), check the mod's What's new (Config), or the Settings UI Mod topic in Lab's FAQ (Help > Troubleshooting & FAQ, or docs\FAQ.txt).
18. WHERE IS IT? FIND A FEATURE BY WHAT YOU WANT TO DO
Classic menu path first, then the new menu's. "Pippa" is the settings menu.
Her and her clothes
Her hair, eyes, body, name ... Pippa > Heroines & outfits > Character prompts
(new: Her looks > Her prompt)
An outfit's look ............. Heroines & outfits > Clothing / outfit prompts
(new: Her looks > Her outfits)
How undressed she is drawn ... Heroines & outfits > Exposure
Tattoo only when bare ........ Heroines & outfits > Only when exposed
Mood, arousal, gaze words .... Heroines & outfits > Heroine state
(new: Her looks > Her state)
Same face every time ......... Config > Her face (face anchor)
(new: Pictures and image packs > Her face)
Scenes, men and places
One scene's picture .......... Scenes & NPCs > Scene prompts
(new: Scenes and people > Scene prompts)
Moods ........................ Scenes & NPCs > Mood prompts (new: ... > Moods)
The men in a scene ........... Scenes & NPCs > Scene participants
(new: Scenes and people > The men in scenes)
One patron's own look/face ... Scenes & NPCs > Named patrons in the tavern
Patron portraits ............. Scenes & NPCs > NPC / patron portraits
(new: Scenes and people > Portraits)
Rooms and places ............. Locations (new: Places)
Prompts everywhere
Remove/replace a word ........ Negatives & tag rules > Tag rules
(new: Prompt rules > Tag rules)
The game's negative prompt ... Negatives & tag rules (new: Prompt rules)
Change many prompts at once .. Bulk edit (new: Change many at once)
Why a picture looks so ....... Prompt inspector, or "inspect" in the In game
strip (new: Prompt Inspector)
Models and quality
Model family and files ....... Pipeline & models (new: Models and LoRAs >
Pipeline)
LoRAs ........................ Pipeline & models > LoRA models & trigger words
Fast / best setups ........... Pipeline & models > Settings profiles
Test one small picture ....... Pipeline & models > Test this pipeline
Picture size, hi-res, face &
hand fix, cut-outs ........... Pipeline & models (new: Models and LoRAs >
Picture quality)
Pictures and packs
Where pictures come from ..... Config > Where her pictures come from
(new: Pictures and image packs > Where her
pictures come from)
Add a pack ................... same place > Add a pack...
Share her as a pack .......... same place > Share a heroine as a pack
Reuse kept pictures .......... Config > Use Saved Images/GIFs
Browse kept pictures ......... Config > Browse saved pictures
(new: Pictures and image packs > Saved pictures)
Draw everything ahead ........ Config > Pictures ahead of time
Review: keep / redo .......... Config > Pictures ahead of time > Review
pictures (Keep / Redo)
Save as WebP ................. Config > ComfyUI models folder > Save pictures
as WebP (new: Setup and backups > ComfyUI)
Convert old PNGs / auto-delete Config > Auto-delete generated images
(new: Pictures and image packs > Clean up)
Animation
Everything about GIFs ........ Animation (same in both)
ffmpeg ....................... Config > ffmpeg path (new: Setup and backups >
ComfyUI)
When animations are made ..... Config > Animation queue
(new: Pictures and image packs > Animation queue)
Setup, saves and help
ComfyUI address and checks ... Config > ComfyUI (new: Setup and backups >
ComfyUI)
Backups, other computer ...... Config > Backup & move to another PC
Copies of your saves ......... Config > Copies of your save
Settings per save ............ Config > Save ID
After a game update .......... Config > Game update check
Keyboard options ............. Config > Speed and shortcuts
Add-ons ...................... Config > Add-ons
Versions and status .......... Config > Diagnostics (see section 19)
What's new ................... Config > What's new (new: Setup and backups >
About the mod)
Bug report for the mod ....... Elena button > In this game > Something wrong
with the mod? (see section 21)
Move the game's panels ....... Screen layout, top right of the game
The mod's extra gameplay ..... Elena button (Daily Happenings), switch at top
Make new things .............. Mara button (the Workshop)
19. OLD MENU AND NEW MENU: WHAT'S DIFFERENT
Both menus change the same settings, so you can switch any time (Try the new menu / Back to the old menu, at the top of the menu). What differs:
- Layout: the classic menu has tabs across the top and each tab is one long page of boxes (Fold all / Open all). The new menu has a list on the left; each page is split into parts, shown one at a time.
- Home page (new menu only): what is on screen, what you changed last, and where everything is.
- Config is split in the new menu: pictures, packs, saved pictures, drawing ahead, her face and clean-up are under "Pictures and image packs"; the rest (ComfyUI, folders, backups, save IDs, add-ons, Diagnostics) is under "Setup and backups".
- WebP: "Save pictures as WebP" sits under Config > ComfyUI models folder in the classic menu and under Setup and backups > ComfyUI in the new one; "Older pictures to WebP" is under Auto-delete (classic) or Clean up (new).
- Links: the new menu also links straight to Daily Happenings, the Workshop and Projects & sharing.
- Names: see the table in section 6. Lab's guides use the classic names.
- Diagnostics is the same in both, and it only shows information: the mod and Node versions, the game folder and version, whether its hooks are in place, ComfyUI's status, the picture library and the data folder. It has no buttons and fixes nothing. Use it to read a version or to see whether ComfyUI is reachable. It shows your full folder paths, which can include your Windows user name: don't post a screenshot of it. To report a problem, use the bug report (section 21) instead.
20. EXTRAS: WHAT EACH OPTIONAL FEATURE NEEDS IN COMFYUI
Lab can check these for you: Lab > Settings UI Mod > Extras checklist. It asks ComfyUI what it has and changes nothing. Install node packs with ComfyUI Manager (Custom Nodes Manager > search the name > Install), then restart ComfyUI. Lab doesn't install node packs: they are programs that run inside ComfyUI, and the Manager is the safe way to get and update them.
Feature (where in the mod) Needs in ComfyUI
------------------------------- ---------------------------------------------
Face & hand fix ComfyUI Impact Pack + Impact Subpack, and
(Pipeline & models) detector files in models\ultralytics\bbox
(face_yolov8m.pt, hand_yolov8s.pt: Manager >
Model Manager)
Her face, face anchor (Config) ComfyUI_IPAdapter_plus, with InsightFace and
the FaceID model files (see its README)
Cut-out figures optional: ComfyUI-Inspyrenet-Rembg (gets its
(Pipeline & models) own model); without it, rougher edges
Animation (Animation tab) an up-to-date ComfyUI (WanImageToVideo),
ComfyUI-VideoHelperSuite, ComfyUI-GGUF for
.gguf models, the WAN 2.2 files (Lab > GIFs &
videos > Check requirements downloads them),
and ffmpeg
Smoother animation (Animation) ComfyUI-Frame-Interpolation (RIFE)
Perfect loop (Animation) an up-to-date ComfyUI
Anima TeaCache Anima TeaCache (BSS), from ComfyUI Manager;
(Pipeline & models) optional, Anima only
After installing: Lab's Extras checklist > Check again, then in the mod Config > Check my ComfyUI (small tests), then turn the feature on.
21. REPORTING A PROBLEM WITH THE MOD
- Update first: the newest Settings UI Mod (and the game, and Lab). The problem may already be fixed, and the current mod has its own bug report.
- The mod's own bug report: Elena button > In this game > "Something wrong with the mod?". Write a line about what happened in the box, then press Copy a bug report for the mod. It holds the mod and game versions, whether ComfyUI was reachable, which of the mod's hooks found their place in the game, your ticked projects and recent page errors. No folder paths.
- Easier: Wayward Lab > Help > Report a problem. Pick "Settings UI Mod" (or "Not sure"). It checks your versions, adds the mod's own report by itself while the mod runs, and adds what Lab can see (two image servers at once, missing models, ComfyUI down...). You read it, then copy it.
- Post it where the report page says: the mod's author takes Settings UI Mod reports in its support thread. Don't send a screenshot of Diagnostics: it shows your folder paths.
22. IMAGE PACKS WITH THE MOD: DRAW, CHECK, SHARE, ADD
An image pack is the game's own format: a folder with pack.json and the pictures, usually zipped. The mod makes packs of the pictures it drew, so the work is in drawing them; packing is one click.
Where it is
Classic menu: Pippa > Config > Pictures ahead of time
Pippa > Config > Where her pictures come from
New menu: Pictures and image packs > Pictures ahead of time
Pictures and image packs > Where her pictures come from
Pictures ahead of time has four tabs:
- The game's scenes: every scene of the game for the heroine you pick (the heroine picker at the top of the menu), in each outfit, room, mood and exposure you leave in. This is the one for a pack.
- While you play: Draw ahead while idle, and how many pictures to keep ready per scene. Fresh pictures while you play, not a pack.
- A Workshop project: a project's scenes, portraits, sprites, backgrounds and battle looks are drawn from its card on the Workshop's shelf (Open the Workshop's projects).
- Brawls & busy hours: Preload her fight's pictures (the stills) and Animate her fight's pictures (after the stills). Turn on Tavern brawls or Busy hours first (Daily Happenings > Nights & hearts).
Make a whole pack, step by step
- Pick the heroine at the top of the menu, and check her prompts and model (a test picture or two first saves hours).
- Pictures ahead of time > The game's scenes > Check what's missing: which of the game's scenes she has no picture of yet.
- Choose what varies:
- Vary outfits / rooms / moods: on = every choice you leave in; off = the one you pick in its list ("as in the game now" = what the game last showed).
- Ctrl+click a choice to leave it out (an outfit you dislike, a room).
- Vary exposure: also draw her with clothes off, each level in its own folder so the game shows the right one. Off = dressed only.
- Exclude scenes: leave out whole scenes.
- Per scene: how many pictures of each combination.
- Only fill gaps: a combination that has pictures already only gets the missing ones. Leave it on when you come back to a pack.
- Generate all scenes. The confirm box says how many pictures that is (check it before you say yes: it can be thousands). Stop pauses it; it comes back paused after a restart.
- Review pictures (Keep / Redo), J keeps, K redraws. By scene shows every drawing of a scene side by side.
- Her pictures as an image pack (also Where her pictures come from > Share her as a pack): a .zip of the newest picture of each scene, outfit, room and mood the mod has of her, as pack.json, plus her character if she was built in the game.
- The .zip has its pack.json already, so anyone can add it: unzip it into the game's packs folder, then title screen > Image packs > Add a pack and pick its pack.json (or the mod's Add a pack..., which takes the .zip). Optional: Lab's Share a pack guide adds the workflow and a README (docs\SHARE-A-PACK.txt).
Someone else's pack
- Add a pack... (Where her pictures come from): puts the pack and its character into the game. The game shows a pack's pictures while it is installed.
- Copy a pack's pictures into Saved pictures (Pictures ahead of time): the pictures go into Saved pictures, each to the heroine it is of, so the mod shows them like its own (with Use Saved Images/GIFs on), even without the pack installed. "Also where she has one already" keeps them behind yours. Packs mix: a zip of Official Gen's Export or images folder works too.
Good to know
- The game's own Settings > Image generation > Make pictures ahead of time is drawn by the mod too while it runs, with your prompts, and it then draws your ticked projects' missing pictures. It always draws the game's usual set; use The game's scenes to choose.
- Pictures you keep (thumbs up, Keep in Review) are in Browse saved pictures.
- A pack's size: Save pictures as WebP makes it a third to a fifth of PNG.
23. EVERY BUTTON, PLACE BY PLACE
The three round buttons at the side of the game (fixed names, whoever you play):
Pippa ... the settings menu (section 7).
Elena ... Daily Happenings: her mind and life in the base game (section 9).
Mara .... the Workshop: make new things and share them (section 10).
Pippa, the settings menu
Try the new menu / Back to the old menu ... switch menus (section 19).
Heroine picker (top) ...... whose prompts and pictures you are editing.
Tabs (classic) / list (new) the pages; section 18 finds things by task.
Fold all / Open all ....... (classic) close or open every box on a page.
Restore default(s) ........ puts a setting (or a page) back as the mod
had it.
Check ComfyUI / Check my ComfyUI (small tests) ... Config > ComfyUI.
Test this pipeline ........ Pipeline & models: one small picture.
Generate all scenes, Stop, Check what's missing, Her pictures as an image
pack, Copy a pack's pictures into Saved pictures, Review pictures ... the
pack buttons (section 22).
Add a pack..., Share her as a pack ... Where her pictures come from.
Animate all / Retry failed animations ... the Animation tab.
What's new ................ Config: what each mod version changed.
The game's picture
Section 8: Generate new, Same seed, Show prompt, Edit, What's on screen?,
< Saved >, Animate (the mod's); thumbs, +, Copy bug report (the game's).
Elena, daily happenings
The switch at the top ..... all of Daily Happenings off or on.
Tabs ...................... section 9.
> Start a brawl now / Start a busy hour now ... Nights & hearts: try one
at once instead of waiting for it.
Always tell me ............ Her reports: she tells everything from the
last week.
Something wrong with the mod? > Copy a bug report for the mod ... In this
game (section 21).
In the game itself: Plan a theme night, Ask her to keep the books, Watch
from the bar, Have a drink (section 9).
MARA, THE WORKSHOP WORKSPACE (opens in its own tab)
Find in all your projects (top) ... searches every project.
The game (top) ............ back to the game's tab; reload it after saving
to play what you changed.
Your game:
Add-ons ................. only what you switch on is in your game; off
means gone (no buttons, events or words in her
pictures). Built-in projects: what differs from
the mod's copy, keep yours or take the mod's.
Projects ................ tick what is in your game, or play one alone.
"Bring in a shared piece" (4.15.22): a shared
.action.json, .show.json or .scene.json goes
into the project you pick, or a new one called
"Shared with me".
Clashes ................. where ticked projects step on each other, and
what to do about it.
Make:
Story (boxes and arrows) the story as a chart.
Story parts (forms) ..... Her asks, Your requests, Routines, Scenes,
Visual novels, Quests, Shows, Actions (4.15.22:
your own buttons in the action list or Elena's
menu, and finishes or moments in your scenes
with her, with every setting the classic
Workshop has; they can be shared from here).
Maps / Dungeons ......... places on a map (click, drag); paint a dungeon.
People .................. each character as a form.
Things .................. Items, Clothes, Shops.
Rules ................... Flags, Statuses, Elements, Skills, Team attacks,
Battle rules, Weak spots, Her traits, Titles,
Positions, Fights in the tavern.
Project health .......... what appears nowhere, checks that can never be
true, each with a fix; the count opens a list.
On every part: Save (keeps a version), Undo / Redo (Ctrl+Z / Ctrl+Y), and
the > buttons (Try it, Play it now, Show it in the game now ...) play it in
the game's tab.
More: A short tour (five steps), The classic Workshop (opens it in the
game; Pippa > Config > The Workshop > Classic look makes Mara open it).
THE CLASSIC WORKSHOP (inside the game)
Eight groups (section 10). Useful buttons:
Make me a project... ...... Projects & sharing: a wizard for a first one.
Glossary .................. the Workshop's words explained.
Preview its pictures ...... every picture a project has, drawn or not.
Draw its pictures ......... draws what a scene or project is missing (it
saves first).
Play this project / This project from here ... try it now.
Fill in from its tags, Draw her like this ... a piece of clothing.
Share / bundles ........... Projects & sharing: a project with its
pictures and GIFs, for someone else.
Your own character pack
In Lab: Game tools > Custom Image Pack Creator (it was called Share Plus).
Straight there: start Lab, then open http://127.0.0.1:8799/?creator=1
The creator keeps everything for one character's pack together: her record, her look, her default outfit, the scene prompts, the workflow and the pictures. Go through its tabs from left to right: Character, Outfit, Prompts, Workflow, Pictures, Export.
What you need
- Lab. Making her record, her outfit and the export needs nothing else.
- Settings UI Mod, to draw the pictures and to send her outfit to it.
- ComfyUI, to draw pictures or test a workflow.
A finished pack plays in the base game without any of these.
Lab helps alongside Settings UI Mod; it doesn't replace it. The mod still edits the scene prompts, decides which pictures to draw, and draws them.
1. Character
Start one of three ways:
- New custom character: type her name. Lab gives her an ID (custom_...).
- Import character / pack JSON...: a record or a pack.json you already have.
- Load character: a custom character from your game. This needs the Wayward Lab Helper on the game page (Game tools > Image variants & Helper).
Her look. Under Appearance source choose (a new character starts with the second):
- Build from appearance fields: pick hair, eyes, build and so on. The box beside them shows the imageBase the game will make from them.
- Write my own imageBase: type her whole look as tags. Lab keeps your picked fields too, so you can switch back. There is no length limit.
Her story, personality and ID are under "Game record: identity, story and personality". Her name is the only box you have to fill in yourself.
2. Outfit
The default outfit is what she wears normally. Her current one is shown first.
- The game's outfits show as cards: name, group, what it looks like, and its top and bottom. Group (All outfits, Everyday, Revealing...) and Find an outfit narrow the cards. Click one to see its wording for every clothing state (top loosened, skirt lifted, and so on).
- Press Replace default with this outfit. Lab copies the outfit and all of its clothing states into her setup, ready to edit.
- Change the words to describe her own clothes. Every clothing state is in one list: type in its box (Game's words puts the game's wording back), and untick Include for a state that doesn't fit her clothes.
- Press Save pack setup (at the bottom) to keep it with her.
With Settings UI Mod:
- Send default outfit + exposure states to Settings UI Mod puts her outfit and every clothing state into the mod, for her save (or for her, if there is no save). Start the mod and have her selected in the game first.
- Lab refuses, and changes nothing, when the mod's settings changed since Lab read them, when another character is selected, or when her clothing comes from an add-on. Choose your own clothing settings in the mod first.
- Load current default from Settings UI Mod copies the mod's version back into Lab, including the game's own wording for states nobody changed.
What it doesn't do
- It changes how her normal outfit is drawn. It doesn't make a new item in the game: what the outfit does in the game stays the same.
- The game still decides when a clothing state happens.
- In the mod, a clothing state's words are added to the normal outfit's words. If a picture shows clothes that should be off, check the mod's prompt preview and remove the clashing words.
3. Prompts
- Scene & task catalog: every scene and task prompt, to read and copy, in one list (All); each shows the start of its prompt, open it for all its variants. Find a prompt searches the words, and Scene or task picks one. Pick where from: the base game, the mod's saved prompts, or the character selected in the running mod. Refresh sources reads them again.
- This pack's prompt profile: the scene prompts kept with this pack. Update pack snapshot copies them from Settings UI Mod; Use base-game defaults copies the game's own. It is a copy: later changes in the mod don't change it until you update it again.
- Try a prompt: pick a scene or task (the variant's words show under it), its variant, outfit, room and expression, then Preview complete prompt. Use in workflow test opens it in Lab's Workflow test; nothing is drawn until you press Run test there. The game adds some things by itself (who else is in the scene, clothing state changes); add those in the boxes if you want them in the test.
4. Workflow
Link the workflow her pictures are drawn with: choose one from your library, or Use the open workflow. Lab keeps a copy with her, so a recipe can carry it. Edit linked workflow opens it in Custom workflow.
5. Pictures
- Generate pictures in Settings UI Mod opens the mod and its review. Draw and review her pictures there.
- Back in Lab, press Import pictures into this pack. It reads the pictures the mod drew for her on this computer (the mod's library folder for her ID) and lists them under New pictures.
- Pictures from another computer: in the mod, export her pictures as an image pack and unzip it, then put that folder (the one with pack.json) in the folder box above and press Use it.
- Export > Save as a pack... copies the new pictures into a new pack folder (packs\<her ID>-additions). Her own pack is not changed.
New pictures and Pack review work as for any pack: see what's missing, look at the new pictures beside the pack's, OK or reject scene by scene.
6. Export
- Save pack setup: writes packs\<her ID>\pack.json with her record, and pack-prompts.json beside it. Pictures already in that folder stay.
- Export character for base game: one pack.json with only her. In the game: title screen > Image packs > Add a pack, and choose it. She is then in the New game list. Your saves don't change.
- Export creation recipe: a zip for someone who wants to make more of her pictures: her pack.json, the outfit, the scene prompts, the workflow and its reference pictures. ComfyUI has to be running to add the reference pictures. No model files and no drawn pictures are in it; on the other computer, upload the reference pictures into ComfyUI.
Kept small for the game The game keeps every custom character's record in the browser and sends it to the image server. So the record in a pack.json leaves out the scene prompts and the workflow: those are in pack-prompts.json and the recipe. When you open her again in Lab from her pack, Lab reads pack-prompts.json back in.
Not in Lab yet Drawing a whole pack with your own workflow straight from Lab. Use Settings UI Mod to draw, then Import pictures into this pack.
A character pack with a LoRA
In Lab: Help > Guides > A character pack with a LoRA
Make pictures of a known character (from an anime, a game...) with a character LoRA: set up the workflow, write her look and outfits from reference pictures, test in Lab, then draw the pack with Settings UI Mod. Each step has a worked example: Boa Hancock from One Piece, on Anima.
What you need: Lab with the Helper on the game page, the character as a custom character in your game, Settings UI Mod (it draws the pictures), and ComfyUI with your base model and the LoRA file. The finished pack plays in the base game without any of these.
Where things are: her look, outfit, prompts and pictures are in Game tools > Custom Image Pack Creator. The workflow is in Custom workflow; tests are in Workflow test and A/B compare.
1. Gather references and read the LoRA's page
- Find a model sheet (front, side and back) for each outfit you want. It shows details one picture hides, like the back of a cape or the shoes.
- On the LoRA's download page, note its trigger word (the name that switches the character on), the recommended strength, the base model it was made for, and what its example pictures look like. A LoRA leans toward what it was trained on (a body type, an outfit), even when your words ask for something else.
- Keep the exact page with its version ID; it goes in the workflow in step 2.
Example (Boa Hancock): LoRA Boa-Hancock-ANIMA_epoch_12.safetensors (Boa-ANIMA-v1.0) for Anima. Trigger word boa-hancock; the author recommends 1.0-1.2. With Anima Turbo at 6 steps, 0.7 was enough. Its training pictures have a very full figure, so it pulls her bust up whatever you write.
2. Set up the workflow
- In Custom workflow, import the workflow you draw with and add a LoRA loader for her LoRA. Start a little under the recommended strength on fast or "turbo" models.
- Put the trigger word in the workflow's own positive text, before {{positive}}. Then it is in every picture, even if a mod's word filter trims it out of her look.
- Open Downloads and add the exact download page for every model and the LoRA. Save.
- Make it the workflow the game's pictures use: Settings UI Mod > Workflow & dependencies.
This workflow adds her LoRA to every picture it draws, other characters too. Use it while you draw her pack, then switch back. At CFG 1 (fast models) the negative prompt does nothing: control her body and clothes with positive words.
Example (Boa Hancock): Anima Turbo v1.1 Plain INT8, LoRA at 0.7, 832x1216, 6 steps, CFG 1, euler_ancestral, simple. Workflow positive text:
anime style, 2d illustration, clean cel shading, clean lineart, best
quality, boa-hancock, {{positive}}
Why it matters: Settings UI Mod leaves out look words that seem to name a body part while she's dressed ("covered by her clothes"), and boa-hancock has "cock" in it, so it went missing from every dressed picture. The test bench's Game check shows such drops and the mod's reason. Fix: fill in her trigger word (Creator > Workflow > Her LoRA) and press Send to Settings UI Mod under her outfits: Lab puts the word on the mod's keep list for her. The workflow's own text is a second safety net.
3. Her look (Creator > Character)
- Load her from the game, or make a new custom character.
- Under Appearance source choose Write my own imageBase. Start with the trigger word, then her face, hair, eyes and body: what never changes.
- No clothes here. Clothes come from the outfit (step 4), so the game's clothing states keep working.
- If the LoRA exaggerates something, name the size you want with a weight, like (medium breasts:1.4), and check it in step 5.
Example (Boa Hancock):
boa-hancock, one piece, adult woman, very long straight black hair parted
in the center, high forehead, narrow dark blue eyes, long eyelashes, gold
snake-shaped dangling earrings, tall slender figure, slim waist, long
legs, (medium breasts:1.4), elegant bearing
Plain "large breasts" drew her noticeably bigger than her sheet.
4. Her outfit (Creator > Outfit)
- Pick the game outfit closest in shape: a dress for a one-piece, separates for a top and a skirt. Press Replace default with this outfit; that copies its clothing states.
- Write her outfit from the sheet, piece by piece: top (sleeves, neckline, fastening), bottom (length, slit, pattern), layers (sash, cape, underskirt), shoes. Name colours and patterns plainly; avoid words that are also things (a "snake emblem" can draw a snake).
- Rewrite each clothing state for her clothes and untick states that can't happen. A layer like a cape is gone in the states where her top is off.
- Press Save pack setup, then Send default outfit + exposure states to Settings UI Mod with her selected in the game.
More than one outfit, worn in the game: the default outfit changes how her normal clothes are drawn. For her lore outfits use Her outfits (wardrobe) under the outfit editor: each outfit with its words, top-only and bottom-only words and clothing states. Save to the game's wardrobe, put Lab's Character wardrobe edit on the game's page once, and they are hers in the game, worn without the boldness check if you tick "she's willing". Send to Settings UI Mod gives the mod their clothing-state words.
Example (Boa Hancock), classic outfit (separates):
red cropped long-sleeved top, deep V neckline with gold ring clasp, pink
scale-patterned flared sleeve cuffs, bare midriff, navel, long red sarong
skirt with green and white swirl pattern, lavender sash tied at the hip,
high side slit, purple underskirt, white cape with gold epaulettes draped
over shoulders, red high heels
Marineford outfit (dress):
purple sleeveless cheongsam, long purple china dress, high mandarin collar
with small V notch, bare shoulders, bare arms, white skull emblems with
pink swirl pattern on the chest, hip and skirt, high side slit, bare leg
through the slit, form-fitting dress, lavender fur trim at the hem, black
high heels
What went wrong first: the classic outfit written as a one-piece dress with a high collar (the sheet shows a crop top and a sarong), and "white Kuja snake emblem", which drew a real snake.
5. Test before drawing the pack
- Creator > Test (the test bench): her look, outfits and trigger word come in from the steps before. Pick test sets: the three game scenes on the game's seeds, an outfit check, face and likeness, clothing states, problem shots (hands, crossed legs, the back, lying down, a man beside her, low light, feet) and game scene sets made from the game's own words (the game's own scene sets). Add any game scene by name to your own set.
- To choose between wordings or LoRA strengths, add variants: they are drawn on the same seeds, so only your change differs. Blind mode hides which is which until you have picked.
- Check: does she look like the character? Is every outfit piece there? Is her body what you asked for? Does the scene still come through, or do her words drown it? Fix the words and test again.
- For one quick prompt: Creator > Prompts > Try a prompt, then Use in workflow test and Run test.
Example (Boa Hancock): the Quick check (bedroom, serving, chatting at a table, on the game's seeds). Without the trigger she came out off-model; with it, the LoRA's own outfit details came through (pink scale cuffs), so the outfit words were rewritten to match the sheet. Problem shots and game scenes then showed the full outfit words drowning the scene (the same front-facing pose); a shorter outfit wording let the poses through.
6. Draw, collect and check the pictures
- Creator > Pictures > Generate pictures in Settings UI Mod, and draw her scenes there.
- Back in Lab, Import pictures into this pack, then go through New pictures and Pack review.
- A scene that already had a picture can keep showing the old one. Check the mod's library shows the new picture before judging it.
- Check one real game picture with the test bench's Game check: it lists any of her words the game or a mod dropped.
7. Save and share
Save pack setup writes her pack. Export character for base game gives a pack.json that adds her to New game; Export creation recipe carries her outfit, prompts and workflow for someone who wants to draw more. Put the LoRA's page and trigger word in your README: see SHARE-A-PACK.txt.
If something looks wrong
She doesn't look like the character
Trigger word in the workflow text; raise the LoRA strength a little.
Her outfit is the LoRA's, not yours
Describe every piece; lower the LoRA strength a little.
A body part is exaggerated
Name the size with a weight in her look; at CFG 1 negatives won't help.
Clothes show in a state where they should be off
Check the mod's prompt preview; remove the clashing words from that state.
Other characters look like her
Switch back from her workflow when her pack is done.
The picture didn't change
The game kept its saved picture; check the mod's library.
Share a pack & workflow
In Lab: Help > Guides > Share a pack & workflow
Beginner guide
Share a pack & its workflow
Follow four parts: export the workflow, add its download links, package your pictures and extras, then check the bundle. This lets others make matching replacements or missing scenes.
Small starter bundle: your complete picture pack, the API workflow, and a short README listing exact model versions, download links, LoRA weights/trigger words and any extra nodes or input pictures. People can view a complete pack without installing its generation models; those dependencies are for generating more pictures.
- Keep the workflow that actually made the pictures
In Lab · Settings UI Mod → Workflow & dependencies
- Generate one representative picture with your intended setup. Before that job disappears from ComfyUI history, choose Export last-used API workflow. This captures the filled-in graph, rather than assuming current settings match an older pack.
- Export workflow template is for reuse with placeholders; it needs values filled before direct execution. If you used multiple setups for the pack, include each workflow and explain which pictures it made.
Screenshot: docs/screenshots/beginner-export-workflow.png
Click the screenshot to enlarge. Example filenames and installed states
are demonstrations.
For Official Gen, Custom Image-Gen Server, or a custom graph
- Export the API workflow from ComfyUI while generating. For a workflow already open in Lab, use Share… → Workflow only (.json).
- You can drag an original image onto the ComfyUI canvas to recover a workflow only if the required metadata survived. PNG alone is not proof that metadata exists; a WebP can display in-game without carrying a recoverable workflow.
- Export before conversion or metadata stripping. Do not rely on recovering the graph from shared pack pictures later. If you cannot recover the actual graph, label a reconstruction as a reconstruction.
Also export normal ComfyUI workflow JSON if you want to share the arranged canvas. API JSON is the format Lab edits and submits; it does not include model weights or referenced source pictures.
- Add exact download pages in Lab
Custom workflow → import the API file → Downloads
- Import your exported API JSON into Lab. Open Downloads to inspect every model and custom-node requirement.
- Use Add download page or edit on each model row. Paste the exact creator/version page, including the version ID where available. Add node-pack repository links too. Save the workflow, then export it again so the saved links travel with it.
Screenshot: docs/screenshots/beginner-workflow-links.png
Click the screenshot to enlarge. Example filenames and installed states
are demonstrations.
Custom workflow → Share…
- Choose Workflow only (.json) for the small export with its saved links. Keep models unticked when recipients can download the matching versions themselves.
- Or choose Workflow + files (.zip) and tick only the extras you want to include. Check a model’s redistribution terms before bundling its weights. Node folders may contain local settings: inspect what you include.
Screenshot: docs/screenshots/beginner-share-files.png
Click the screenshot to enlarge. Example filenames and installed states
are demonstrations.
Important: Lab’s workflow ZIP contains the workflow and selected dependencies. It does not automatically add your picture pack. Combine the exported workflow/ZIP with the complete pack yourself.
- Put the pack and instructions together
Keep the pack’s pack.json and its pictures in their correct folder. Do not send pack.json alone. One possible sharing archive, extracted beside the game’s index.html, is:
MyPack-share.zip
packs/custom_mycharacter/pack.json packs/custom_mycharacter/[the picture
files] workflows/my-pack.api.json workflows/my-pack.comfyui.json (if you
also exported the canvas) README.txt
Use your pack’s actual folder name, including the folder named by its pack.json. Avoid accidentally nesting packs/packs . Adding a workflow does not repair an incorrect picture path.
Want a smaller pack? If your pictures are PNG, WebP is the way to go: it is several times smaller, works the same in packs, and keeps big packs under the 4 GB limit of the game's Share button. To share a WebP copy:
- Convert a copy, never your own library. Image servers keep records that point at those exact files; redo and review depend on them. Your own pictures stay as they are.
- Export the workflow first. Converting removes the workflow stored inside each picture.
- Fix the copy's pack.json. Its pack.json still names the .png files. Run Rebuild pack.json on the copy so it points at the .webp files, then check the result before sharing.
- If your image server has a WebP option, using it for new pictures saves converting later.
A simple README you can copy Made with: [Official Gen / Settings UI Mod / Custom Image-Gen Server] Pipeline and exact checkpoint/diffusion model: [name + version] Download pages: [one exact page per model, VAE/encoder if separate, and LoRA] LoRAs: [filename, version, strength, trigger words] Custom nodes: [repository + version/commit if known] Workflow: [API filename; say whether fixed or a template] Inputs: [reference pictures needed by the graph, if any] Reproduction example: [prompt, negative, seed, size, steps, CFG, sampler] Pack install: needs game version 2026-11-01 or newer (update the game first).
Unzip into the packs folder next to the game's index.html so you have
packs\<pack name>\pack.json, then on the title screen: Image packs > Add a
pack, and choose that pack.json.
Known gaps: [missing scenes or pictures you want others to patch]
Choose your extras: a model-link list, model/LoRA notes, a sample original with metadata, reference inputs needed by the graph, or a separate character-record JSON if you want to share it. Character records and model files are not automatically included by exporting an API workflow. Use Character record → Export JSON for a record you choose to include.
- Check the bundle before posting
- Open the archive and confirm the actual pictures, pack.json, workflow and README are there.
- Re-import the workflow into Lab and check that the download pages point to the exact intended versions. A graph needs its own input images too, not just models.
- If you share a template, name its required placeholders. If you share a filled API, give one prompt/seed example. A successful export is not proof that another installation can render it.
OPTIONAL: SHARE YOUR SETTINGS AS WELL
Game tools > Saves & files > Lab storage > Your settings in one file > Export my settings makes wayward-settings.json. Start Settings UI Mod first if you want its shared prompt, pipeline and preference choices included. Read the file before sharing: prompts are your own words.
The file leaves out folder and connection settings, addresses, ports, start files, the Helper key, save ids and history. It does not include pictures, models, workflow files, add-ons or Workshop projects. Keep the pack and the API workflow in your bundle as described above. Add this settings file as a separate optional file; it is not added to a pack ZIP just because you exported it.
The other player opens it at Lab storage > Your settings in one file > Open a settings file…. The preview shows Setting, Now and From the file. They tick parts, then press Apply the ticked parts… and confirm. Only those parts are applied. They must set their own folders and Connections and have the required model files. They should read the Applied and Not applied result, restart an image server if asked, and refresh the game page. A settings file is not the mod's Export everything backup.
For a shortcut to the mod's own page while checking a setup, use Show it in Settings UI Mod in Lab's Connections card. The nearby shortcuts open models, scene prompts and prompt rules too. Put the Helper on that game page first, press the shortcut, then switch to the game's tab. It opens the menu but saves nothing. If it cannot find the place, open Pippa and choose the section yourself.
GIFs and videos
In Lab, open Help > Guides > GIFs & videos. Its tabs separate Settings UI Mod, Official Gen and Custom Image-Gen Server. The redundant Custom workflows video option was removed; the main reusable workflow editor is unchanged. Common troubleshooting now has its own Help > Troubleshooting & FAQ section (docs/FAQ.txt). This is a setup guide; navigation buttons do not start generation, install models or change source settings automatically.
Quick test without leaving Lab
- Open GIFs & videos, select the image source and choose GIF or MP4 output.
- Check requirements first. The provided Wan 2.2 graph checks installed nodes, models and output-format availability on that source's renderer. Missing files and destination folders are listed here. Install missing models or matching files from Downloads; install node packs separately and restart ComfyUI, then Check requirements again.
- Once the check passes, Choose a picture to animate unlocks. Pick an existing PNG/JPEG/WebP still, such as a portrait, and choose 1/2/4 seconds. Enter a motion prompt, e.g. slow breathing, gentle hair movement, fixed camera. Changing the prompt/duration does not require a new installation check; changing source/output format does.
- Press Test picture in Lab. This submits a real render; wait for the preview and Download link here. The starter uses 400x608, 16 FPS and seed 1.
- Download provided API workflow if you want to keep/edit/share that graph.
The backend checks readiness again before uploading/submitting the test. This test does not install a persistent override or start an image server. Native in-game animation settings remain source-specific; advanced details are folded below the main flow. Readiness is not proof of a completed render.
The same Lab flow for each source
Official Gen: select Official Gen, then follow Check requirements > install missing dependencies > choose picture > Test in Lab. Lab reads that source's ComfyUI connection. Its still-image Illustrious pipeline is unchanged. Native in-game video additionally uses its configured Wan workflow and node IDs.
Settings UI Mod: select Settings UI Mod and use the same Lab steps. For native in-game animation afterwards, use that mod's animation/GIF controls and output processing. Those native details are folded under Advanced source setup.
Custom Image-Gen Server: select that source and use the same Lab steps through its ComfyUI URL, including custom ports. The Custom Image-Gen Server has no native video API; the Lab test does not add one. Native game playback requires support in that server version.
Picture selection stays locked until the current source/output check passes. Switching sources invalidates old download buttons, preventing installation for a previously selected source. While a video is submitted/running, its source is locked. If polling loses the connection, Check existing test resumes tracking that same job instead of submitting another video.
Side-by-side setup
In GIFs & videos, select your source and press Show Lab ↔ game setup steps. Each task pairs the Lab controls with the corresponding actual game/mod panel, numbered steps and an expected result. Click a screenshot to enlarge it. Settings UI Mod's native motion controls are in its Animation section (4.15.2 to 4.15.19). There is no separate page gallery. Custom Image-Gen Server has no native video API; its walkthrough distinguishes a Lab test from in-game playback. Each screenshot has numbered markers that match its steps, so it can be followed in any language. Screenshots are setup examples and do not show completed GPU renders.
Before you start
ComfyUI is the renderer. Run one game-facing image-generation server at a time. An image-to-video graph needs a source still, motion prompt, appropriate models, text encoder, VAE and an animation/video output node. A still-image checkpoint alone is not an animation setup. Start with one short, small test in the source that will actually generate/play it before requesting batches.
Worked example 1: one portrait in settings UI mod
Goal: a short blink/breathing animation, without animating every new picture.
- Show an existing portrait in the mod-hosted game.
- Animation: Animate = nothing (only the GIF button).
- Example first settings: Quick size, 2 seconds, 12 FPS, animated WebP. Keep the graph's matching models/LoRAs/sampling; dependencies must already work.
- Motion: slow breathing, one gentle blink, slight hair movement, fixed camera. Save & Apply. Disable automatic motion matching for this controlled test if it would select a different scene template.
- Click the portrait's GIF/animation action once and wait for its queue item.
Expected: roughly two seconds of motion. .webp is correct when animated WebP was selected. Look in the mod library/review, commonly under wayward-bridge-data/library/<character id>/. Choose GIF and generate again if you specifically want .gif; renaming does not convert it. The mod's one-second motion template preview is a quicker check, but that preview is not saved.
Worked example 2: a gif directly in ComfyUI
Start with a working Wan i2v graph and a still named portrait.png. To inspect the mod's graph, open its animation requirements in Lab > Inspect animation workflow, then Share its API JSON. Ensure that graph's source input and required files exist in your ComfyUI installation before trying it. A raw mod template may need loader/ node adjustments outside the mod: some substitutions happen only at mod runtime.
- Load Image = portrait.png.
- Positive motion = slow breathing, gentle hair movement, fixed camera.
- For the standard Wan example: width 400, height 608, length 33 frames. Other pipelines can require different dimensions/frame counts. Keep the matched high/low models, acceleration LoRAs and sampling schedule.
- VHS_VideoCombine: format image/gif, frame rate 16, save output enabled, filename prefix lab-motion-demo.
- Run/queue in ComfyUI and wait.
Expected: a GIF about two seconds long, in ComfyUI's output folder, with a filename beginning lab-motion-demo. Open it in a browser to check movement. MP4 version: if the output node offers video/h264-mp4, select that and run again. If absent/failing, check VideoHelperSuite and its ffmpeg support. Keep the JSON with your shared media. This creates a file, not an automatic game installation.
Worked example 3: official gen video in the normal game
Start with a tested Official-compatible API graph, for example: C:\Wayward\wayward-lab-data\official-wan-api.json Replace that sample path with your actual file; the example does not supply it.
- Official Gen > Connection & folders > Moving pictures: set Wan workflow path to that file, save the config and restart Official Gen.
- In the game's image-server settings press Check again. At desktop width, show a still portrait and use its video action.
- The server uploads that picture and runs the selected Wan graph. Its own settings determine motion/length/output; still-image Illustrious CFG/steps do not configure this video.
Expected: the returned animation plays through the game's video route and is cached under the generator's image folder/videos/<character id>/. An MP4 graph returns MP4; a GIF graph returns GIF, subject to browser playback support. A missing-node error such as 241 means the graph does not match the generator's expected IDs. Use a compatible graph/remap; changing its filename does not help.
Worked example 4: an Anima-server picture as the source
Generate/save portrait.png with Custom Image-Gen Server, then select it as the source of the separate ComfyUI animation graph in example 2. Expected: an output GIF/MP4 from ComfyUI; the Anima server still works as before. This alone does not register the new animation for in-game playback. Use your server's documented video route if it has one, or the integrated mod/Official route described below.
Reference setup per source
Settings UI mod
- In Lab's GIFs & videos > Settings UI Mod, open animation requirements. Install the missing model files in their displayed folders and node packs in ComfyUI; restart ComfyUI after adding nodes, then check again.
- The default Wan 2.2 graph needs the high- AND low-noise models, their matching LightX2V acceleration LoRAs, UMT5 text encoder and Wan VAE. A GGUF loader and VideoHelperSuite are needed; rgthree, frame interpolation and loop options have their own requirements/fallbacks. Custom graphs may differ.
- In Settings UI Mod, Config > ComfyUI > Check my ComfyUI checks the renderer, formats and ffmpeg. This diagnostic can run small real test jobs; Lab's requirements checker only checks availability. Configure the mod's ffmpeg path for its conversion/processing features.
- Animation controls motion, length/FPS, size and format. Animate can be manual, scenes with their own motion, or every scene picture. Save & Apply in the mod. Use the GIF action on a game picture or its Test an animation diagnostic; the latter makes a real short ComfyUI job.
- For project animations, Use animated project pictures controls display of existing animations. It does not itself generate them.
- Use Saved Images/GIFs enables saved-media reuse. Select the same heroine/scene and check the correct store: mod library versus saved-scenes.
Finished output can be animated WebP or GIF. MP4 may be an intermediate hand-off when ffmpeg/output support allows it. Output GIF is a panel name, not a guarantee that the finished filename ends in .gif.
Official gen / normal index.html
- Obtain a compatible Wan image-to-video API workflow and its model/node files. The inspected generator expects the node IDs from ComfyUI_00033_.json. An arbitrary Wan export is not a drop-in replacement.
- Official Gen > Connection & folders > Moving pictures: set Wan workflow path to the absolute filename, save wayward-imagegen.config.json, then restart Official Gen. CLI alternative: --wan-workflow <absolute workflow path>.
- Official Gen > Downloads & requirements > Check again also checks the selected video graph. It has its own models/VAE/sampling, separate from Illustrious still images and Lab's still-image override.
- In the game's image-server settings, press Check again after enabling video. A video-capable connection exposes the portrait video action at desktop width. Show a source picture, request the video and wait for completion.
- The server uploads the source image and caches the animation in its configured image folder under videos/<character id>. It recognizes MP4, WebM, MOV and GIF; actual browser playback depends on the codec/container.
A missing-node error means the workflow does not match the patcher's expected IDs. This implementation patches image 241, positive 207:93, negative 207:89, seed 216, samplers 207:86 / 207:178, and output prefix 221. Use the matching graph or deliberately remap it; renaming a file does not change its IDs.
Custom image-gen server
Lab's Custom Image-Gen Server integration edits still-image models and character/outfit prompts. It does not expose native animation setup or establish that the installed server implements the game's video API. If your version documents motion support, follow its own workflow/output instructions and requirements. Otherwise run the animation graph in ComfyUI directly, or choose an integrated animation server. Stop the current image-generation server before starting another. ComfyUI can remain running. Replacing a still-image graph does not add a missing video API.
Custom ComfyUI workflows
- Start with an image-to-video graph with a source image and animation output.
- Test it directly in ComfyUI with a short motion prompt and small frame count.
- Import its API JSON in Lab to inspect/edit nodes and requirements. Keep the workflow JSON alongside shared results. Fill required placeholders before standalone use; the template export is not an unchanged standalone job.
- Lab's current test preview handles image/GIF-style outputs. Inspect video files directly in ComfyUI.
- For in-game use, use the source's actual animation route: Settings UI's moving- pictures slot (gif.json / i2v.json in its workflows folder), or Official Gen's compatible wanWorkflowPath graph. Start from that source's animation graph. A generic still-image override does not create the necessary server features.
Troubleshooting & FAQ
Common questions have their own section in Lab: Help > Troubleshooting & FAQ. See docs/FAQ.txt for output-format problems, files that exist but are not reused, incognito, library versus saved-scenes, missing indexes and pack recovery. Model-file roles/pipeline labels have their own guide: docs/MODEL-GUIDE.txt.
These instructions were checked against the bundled game, Official Gen source and local Settings UI source/documentation. Features in another server/mod build can differ. No animation generation was performed while writing this guide.
In-app visual guide
Open Help > Guides > GIFs & videos > select your source > Lab ↔ game setup for labelled examples and source-specific steps. The sidebar group headings fold/unfold and remember your choice. Opening a page through another link expands its sidebar group automatically.
Phone and computer
The game runs in a browser, so a phone or tablet (Android or iPhone/iPad) can play it. Lab itself, Official Gen, ComfyUI and Settings UI Mod only run on a computer. The same guide is in Lab under Help > Guides > Play on your phone.
Which way to play
From your computer Game copied onto sintention.com
over Wi-Fi the phone (the website)
Android yes yes, with JoiPlay yes
or a web server app
iPhone / iPad yes no (see below) yes
New pictures drawn yes no no
Image packs yes, by address yes (JoiPlay or only one put
web server app) online (D)
Mods (mod folder) yes should (in JoiPlay) no
Settings UI Mod yes, its address no no
Cheat mods yes yes (in JoiPlay) no
Changing gold/stats with a cheat mod, or without: bookmark or save file
- From your computer: the best way. The phone plays the game your computer serves, with your pictures, packs and mods. Needs the computer on, same Wi-Fi.
- Game copied onto the phone: no computer needed, but no new pictures: only the game's own pictures and packs you put in its folder.
- sintention.com: the plain game, plus packs someone put online (D).
- Saves: each way keeps its own saves in that browser or app. To move one: in the game, Load Game > Export current to JSON, then Import from JSON on the other side.
A. from your computer over wi-fi (android and iphone)
First get one image source working on your computer. Keep your computer on while you play. Put the phone or other computer on the same home network. Use a network you trust: other devices on it may use the server and change its settings. Do not open these ports to the internet.
1. open the right connections card
In Lab, open your image source: Official Gen, Custom Image-Gen Server, or
Settings UI Mod. Open Connection & folders and find Connections. These are
separate tools. Run only the one you use for the game.
2. connect that source to ComfyUI
Under Where ComfyUI is, choose On this computer if your computer draws the
pictures. Leave its ComfyUI port at 8188 unless yours is different. If a
different computer draws them, choose On another computer. Type that
computer's address, for example 192.168.1.20, and its ComfyUI port. Start
ComfyUI there with --listen and allow it through its firewall on private
networks. Press Check on the card. It checks ComfyUI; it does not check the
phone, save settings or draw a picture.
3. let your home network use the image server
Under Who may use it, choose Every device on my home network. Check the
server's Port. This port is separate from ComfyUI's port. Press Save into
Official Gen, Save into the Custom Image-Gen Server, or Save into Settings
UI Mod, on that source's card. Confirm the home-network choice.
- Official Gen: Lab writes its config. Stop and start Official Gen.
- Custom Image-Gen Server: Lab writes its config.json. Stop and start that server if the saved message asks.
- Settings UI Mod: start the mod before saving its ComfyUI address. Lab saves that address in the mod. Its host and port go into a start file Lab uses. Press Restart it now if offered, confirm, and wait for it to start. Refresh or reopen the game afterwards. Typing or saving a host alone does not move the running server.
4. find your computer's address
Press Show this computer's address on the Connections card. It is only
looked up when you press the button. Hide it hides it; leaving the page
makes Lab forget the displayed address. Use your own address in place of
the example 192.168.1.20 below. Or use Windows Settings > Network &
internet > your connection > Properties > IPv4 address. A command window's
ipconfig shows it too. This is the computer serving the game, which need
not be the computer running ComfyUI.
5. open the game on the phone or other computer
Official Gen does not serve the game page. In a command window in the game
folder, run one of these and keep that window open:
npx http-server -p 8080
python -m http.server 8080
The first needs Node.js; the second needs Python. Open this in the other
device's browser:
http://192.168.1.20:8080/index.html
The game then reaches Official Gen on your computer. Use the port your
setup expects; the usual Official Gen port is 8189. If you chose another
port, check the game's image-server address too.
Settings UI Mod and the Custom Image-Gen Server serve the game themselves.
Do not start the separate web server for them. Open the address from that
source's Connections card, for example:
http://192.168.1.20:8788/
If you changed its port, use the saved port after restarting it. Use
http:// for these local examples, not https://.
6. add a pack if you want one
Packs belong to the browser and address where you add them. On the other
device's title screen, open Image packs and add the pack by its address.
With the separate web server above, an example is:
http://192.168.1.20:8080/packs/<pack name>/pack.json
Keep the pack unzipped in the packs folder on your computer. Use your own
computer address and the port that serves that folder. It is a pack.json
address, not a ZIP download. Do not assume a pack added in your computer's
browser is already added in the phone's browser.
With settings UI mod instead of official gen
Use its own Connections card and game address. Show it in Settings UI Mod
can open ComfyUI in Pippa on a game page with the Helper. Press it, then
switch to that game page. It opens the menu; it does not Save. If it does
not open, use Pippa yourself. See SETTINGS-UI-MOD-GUIDE.txt.
If it doesn't connect
- Press Check again on the source's Connections card. Check the address and port of ComfyUI separately from the address that serves the game.
- Check that you saved, restarted the right server, and opened its new address. A saved host or port takes effect when it starts again.
- Allow the image server and, for Official Gen, the separate web server through your computer's firewall on private networks. Remote ComfyUI also needs --listen and its own firewall permission.
- Your computer's network address can change. Press Show this computer's address again, or check its network settings. Use the new address.
- If you change image server or game address, add your packs at that address. Saves also stay in that browser. Export a save to JSON and import it on the other device if you want to move it.
- Away from home, use a pack put online instead of opening your home server to the internet. See D below.
B. game copied onto an android phone (joiplay)
- Use JoiPlay (an Android app that runs HTML games from a folder). Copy the whole game folder to the phone and add index.html in JoiPlay. Keep the folders and files in the same places as on your computer.
- For an image pack, unzip it into packs/<pack name>/ beside index.html. On the title screen, open Image packs, then Add a pack, and choose its pack.json. Keep a mod's files where that mod's instructions put them.
- No JoiPlay? A small web server app on the phone does the same as way A, without the computer. The game counts a page opened from localhost as the downloaded game, so Image packs works the normal way. (Follows from how the game checks it; not tried on a phone yet.) 1. Unzip the game on the phone (its Files app can), and the pack into a folder named packs next to index.html: packs/<pack name>/pack.json. 2. Serve that folder: a "simple HTTP server" app from the Play Store pointed at the game folder, or Termux: termux-setup-storage once, then pkg install python, cd into the game folder, python -m http.server 8000 3. Open http://localhost:8000/index.html in Chrome (or the app's address). 4. Title screen > Image packs > Add a pack, and pick that pack.json. The server app has to run while you play; saves stay in that browser.
- New pictures aren't drawn this way (Official Gen is on the computer). Use way A for that.
C. iphone and ipad
- Use way A (from your computer) or sintention.com.
- A game copied onto an iPhone isn't a working way: JoiPlay is Android only, and opening index.html from the Files app doesn't load the files next to it (packs, mods).
- Mods and packs work on an iPhone through way A, because the computer serves them.
D. a pack put online (the website, or away from home)
The game on sintention.com (and any phone browser) can't open a pack saved on the phone. It can add a pack someone put online, by the address of its pack.json. The usual way: the pack's author puts it online once and posts the address with the zip; everyone else only pastes it.
- Players: title screen > Image packs, paste the pack.json address, Add. The browser keeps the pack's list; pictures load from the site as they're shown (needs the internet while playing).
- Updates: a picture replaced on the site under the same name shows by itself (the browser may keep the old one a while). New pictures or a new pack.json: add the address again; it updates the same pack.
- The site needs: an https:// address; the pack unzipped, pictures next to pack.json as in the zip (a .zip can't be added); pack.json readable by other sites. Many file-sharing sites don't allow that last part.
- Use a host that gives direct HTTPS links and lets other sites read pack.json. Set its cross-origin access rules if needed. Check the host's current file, storage and traffic limits before uploading.
Mods on a phone, step by step (cheat mods and other mod builds)
- Set up the mod on your computer, following that mod's own instructions. Keep a copy of the original game folder. Use the game version the mod supports. A phone cannot run your computer's build or start commands.
- Check the mod in the browser on your computer first.
- With your computer on, follow A and serve the modded game folder. Settings UI Mod serves its own game page. On Android without your computer, copy the whole prepared folder and follow B with JoiPlay or a local web server app. On iPhone or iPad, use A.
- The other device keeps its own browser saves. Move a save with Load Game > Export current to JSON, then Import from JSON on that device.
Cheats without a mod: changing gold or stats
The game has console commands (game 2026-11-01): wayward.help() lists them, for example wayward.set("economy.gold", 500). On a computer: F12 > Console. Phones have no console, so use one of these. Back up your saves first.
- A bookmark (Chrome on Android, Safari on iPhone): 1. Bookmark any page. 2. Edit the bookmark: name it gold, and replace its address with javascript:wayward.set("economy.gold",500);void 0 3. Open the game and load your save. 4. Run the bookmark: Android Chrome: type gold in the address bar and tap the bookmark in the list (the one with the star). iPhone Safari: open Bookmarks and tap it. One bookmark per change. Some browsers block these; JoiPlay has no bookmarks.
- The save file (works everywhere, JoiPlay too): Load Game > Export current to JSON, open the .json in a text editor app, change the number (for example after "gold":), save it, then Load Game > Import from JSON. Keep a copy of the original file.
- With a computer and a cable: Android Chrome's remote debugging (chrome://inspect on the computer) or, on a Mac, Safari's Web Inspector for an iPhone, give the full console.
Fix pictures after an update
In Lab: Help > Guides > Fix pictures after an update.
Pictures were fine, then ComfyUI updated or you installed something, and now they come out crunchy, blurry, burned or wrong. Go through this in order.
1. Is it really the update?
- It is likely when the pictures changed right after a ComfyUI update, a node pack install or a click on "Update" or "Install" in ComfyUI Manager, and you changed nothing else.
- It is likely not the update when you also changed the checkpoint, a LoRA, steps, CFG or the workflow. Undo that first. Crunchy, over-detailed pictures often mean the picture is "overcooked": two models or adapters doing the same job, too many LoRAs, or CFG or steps too high. See FAQ.txt: "Pictures come out blurry, distorted or with odd colours".
- Write down what you installed or updated, and when. You need it in step 3.
2. Back up before you change anything
- Stop ComfyUI and the image server (Official Gen, Settings UI Mod or the Custom Image-Gen Server).
- Copy these somewhere safe: your exported workflows, ComfyUI's user folder, Lab's wayward-lab-data, the mod's wayward-bridge-data, your image packs and your game saves (Lab: Game tools > Saves & files > Backups).
- Leave the models folder where it is. You will point the repaired ComfyUI at it; there is nothing to download again.
3. Look for two copies of the same node pack
The most common cause: two versions of one node pack installed side by side, for example two IPAdapter packs after pressing the wrong Install. They fight over the same nodes.
- Start ComfyUI. Open Manager > Custom Nodes Manager and show only Installed.
- Look for two entries that do the same job, or the one you installed just before it broke.
- Uninstall the one you didn't mean to install (or the newer one, if both are yours). Restart ComfyUI.
- Test a picture that used to work (step 6).
Uninstalling often does not fix it. A node pack also installs Python packages into ComfyUI's Python, and removing the pack leaves those behind. If pictures are still wrong, go on to step 4.
4. Clean reinstall of ComfyUI
Installing over the old copy keeps the same problem. Start fresh, and only bring back what you need.
- Install ComfyUI again into a new folder, with its own new Python. Don't copy the old custom_nodes folder or the old Python across: that brings the conflict back.
- Point it at your existing models. ComfyUI Desktop asks where your models are during setup. The portable version reads extra_model_paths.yaml in its ComfyUI folder (rename the .example copy and put your models folder in it).
- Test the plain workflow first (step 6).
- Add the node packs your workflow needs, one at a time, with ComfyUI Manager. Restart and test after each. If one breaks it again, that is the pack to leave out or replace.
- When everything works and your files are all there, remove the old ComfyUI folder. Don't delete a shared models folder just because its path has "ComfyUI" in it.
Reinstall only what is broken, usually ComfyUI. The game, Lab and the mods keep working. A reinstall fixes a broken install; it does not fix a LoRA made for another model, or a wrong workflow setting.
5. Put the model files in the right folders
After a reinstall, check that the files your workflow uses are where ComfyUI looks for them. The folder names go inside your models folder, or inside your shared models folder if you use one. Create a folder if it isn't there.
Example: SDXL IPAdapter FaceID Plus v2. Only if your workflow uses it. It is not a fix for Anima, for other models or for workflows without FaceID.
CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors -> clip_vision
ip-adapter-faceid-plusv2_sdxl.bin -> ipadapter
ip-adapter-faceid-plusv2_sdxl_lora.safetensors -> loras
- The LoRA goes in loras only. The FaceID loader finds it there; a second copy in ipadapter isn't needed.
- FaceID also needs InsightFace, installed into the Python that runs ComfyUI.
- The CLIP-ViT file reads the reference picture. It is not the checkpoint's text encoder.
- Exact names and the other FaceID variants: https://github.com/cubiq/ComfyUI_IPAdapter_plus#installation
Lab can check this for you: open the workflow in Custom workflow and press Check. It lists missing nodes and model files, and a missing file already in your Downloads folder moves into the right folder with one click. With Settings UI Mod, its Extras checklist says what each feature (face anchor, face & hand fix...) still needs.
6. Test with a picture that used to work
- In Lab's Workflow test, use the same prompt, seed and size as a good picture from before.
- Turn off the extras first: adapters, extra LoRAs, detail passes, speed-up nodes such as TeaCache. Then turn them back on one at a time.
- Use the checkpoint author's sampler, steps and CFG. More steps are not always better.
7. Settings UI Mod: the custom workflow isn't the one drawing
- The switch is in Pippa (the mod's image-settings menu), not in Workshop Workspace > Add-ons. Classic menu: Config > Add-ons. Newer menu: Setup and backups > Add-ons. Or type "add" in Pippa's Search box.
- Imported several versions of a workflow? Each one gets its own "Custom workflow: <name>" tick. Untick the ones that don't work, so the game draws with the one you want.
- No entry there at all? Create it in Lab first: Custom workflow > open your workflow > Save to library > Image server tab > In the game > Add it to the game's menu.
8. Still broken? Ask with the right details
"Default everything except the checkpoint" isn't enough for someone else to help. Share:
- The workflow itself (Custom workflow > Share; the workflow alone is enough). Your setup may differ from the default ones.
- The checkpoint (or diffusion model), VAE, text encoder and LoRA names, plus steps, CFG, sampler and size.
- A bad picture, and a good one from before if you have it.
- What you installed or updated just before it broke.
Lab's Help > Report a problem collects most of this and takes your user name out of paths. Read it before posting.