Repository navigation
Releases
Better Comfy 1.5.1
A small update with fixes for the gallery.
- Albums inside albums: sub-collections can hold their own now, a few levels deep. Moving a collection from the right click menu puts it exactly where you picked (it could land in the wrong one before).
- Duplicates where you are: Find duplicates only looks through the collection or sub-collection that is open, and works in the vault and its collections too.
- Duplicates first: a new sort that puts the stacks of look-alikes on top.
- Calmer hints: no more long prompt pop-ups over pictures, and no hint over the empty preview.
- Dots is the default background now. Off is still in Settings.
If Windows Defender flags the download with a name ending in !ml, that is a guess by its machine learning check on a new, unsigned program, not a known virus.
Download BetterComfySetup-1.5.1.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.5.1-win-x64.zip:
1041ea413c8fd783faa67fa16878362dac6885bfd91bf9251d6b64570a1a6cb4
Better Comfy 1.5.0
A big update: two new tabs that are still being worked on, a Storage tab, duplicates, export, undo, and a fresh look.
Work in progress
These two tabs work, but aren't finished yet. They are marked WIP in the app.
- Characters: make a character from a few pictures, pick it on the Image page and put it in any scene, from free to 1-to-1.
- Train: train a LoRA of your own from pictures, right in the app, with the training built into ComfyUI. It takes a long time on most graphics cards.
Library
- Storage: a new tab that shows what takes up space, biggest first. Delete, move or compress big pictures and videos.
- Duplicates: finds look-alike pictures without AI and stacks them. Nothing is deleted: open a stack, keep what you want, mark the rest.
- Export many pictures at once, with the prompt and settings taken out if you like. New results can also get a second, clean copy.
- Side by side: select 2 to 4 pictures and zoom and move them together.
- Use as start: send any picture from the gallery to the Image page as the start picture.
Everyday
- Undo for deletes, a cleared prompt and Unmark all (Ctrl+Z too).
- Quick search with Ctrl+K, and every shortcut with F1.
- Models by picture: the model list on the Image page is a grid with covers and a search.
- Panic key: hides the app and locks the vault, from anywhere. Set it in Settings, Privacy.
- Connection log in Settings, Privacy: everything the app connected to since it opened.
- Backup of settings, presets, notes and characters in one zip.
- A clean start each time the app opens, boxes on the Image and Video pages that fold away, and tab groups in the rail.
Look
- Light mode, with a switch at the top.
- Moving backgrounds: Dots, Constellation and Binary city. Pictures you put in
Pictures\Better Comfy\Dot artbecome Dot art backgrounds made of drifting particles. - An interface size setting, and panels you can see the background through.
Faster
- The gallery opens right away even with thousands of pictures.
- Changing colours or light and dark no longer freezes the app.
- Downloads from the Market show up in the library at once, with a steady progress bar, also for files over 2 GB.
Download BetterComfySetup-1.5.0.exe and run it, or let Better Comfy update itself.
Windows Defender sometimes flags new versions with a name ending in !ml. That is its automatic guess about a new, unsigned program, not a known threat. The full source is in this repository.
SHA-256 of BetterComfy-1.5.0-win-x64.zip:
f3b0458b30613ff373c322e4de4d17cac08226194b73816c9bd4194c265c4c29
Better Comfy 1.4.1
A small gallery update: a real video player, small edits and dragging vault pictures out.
Video player
- Videos play in the big viewer with their sound, a time bar you can drag, the time, volume and mute, loop, and a moment back or on (
,and.). Vault videos play straight from memory. - The small preview in the gallery stays a quiet loop.
- GIFs get a frame bar: pause and go through them frame by frame.
- Save this frame as a picture, right from the player.
- Slideshow: press S in the big viewer.
Small edits
- Trim, crop, turn and mirror pictures, GIFs and videos (right click, Trim, crop, turn, or E in the big viewer). Crop freely or to 1:1, 4:5, 2:3, 3:2, 16:9 or 9:16.
- The result is always a new one next to the original, in the same folder, drive, vault and collections. The sound of a video is kept. Vault items are edited in memory only.
Drag and drop
- Drag pictures and videos out of the vault into a folder or onto the desktop. The copy made for the drop is shredded a few minutes later (and when Better Comfy closes).
Download BetterComfySetup-1.4.1.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.4.1-win-x64.zip:
ea8bcb77fe2e4e8f0d9691d2e9544e4b9c367feada8fcce0b2808b0aedc1d5c9
Better Comfy 1.4.0
Faster videos, drives, sub-collections, GIFs and a prompt helper that writes what you ask for.
Faster videos
- New presets from Ultra Fast to Best. Fast HD makes a video small and quick, then sharpens it to Balanced size with an upscale model: close to Balanced in about half the time. Quality and Best are made larger and then sharpened to one and a half times their size.
- Speed models (Lightning, lightx2v) get the 4 steps they are made for: Balanced is about a third faster with no visible difference.
- Faster maths is now on by default: about 15 to 20 % faster, no visible difference.
- The upscale model for videos can be picked on the Video page (anime or realistic).
- Settings, Performance looks at your PC (page file, memory, power mode, where the models are, the graphics driver) and Better Comfy's own options, and opens the right setting for each recommendation.
Drives
- Keep pictures, videos and vault files on a USB stick or another drive (Settings, Drives). What is on it shows in the gallery while it is plugged in, vault files once the vault is open too, sorted as before. Nothing is forgotten while it is out.
- Right click, Move to: onto a drive and back. Files are copied, checked byte for byte, and only then removed. Vault files stay encrypted on the drive, which also keeps an encrypted copy of what is on it.
Gallery and vault
- Sub-collections, in the gallery and the vault: a collection shows everything in it and its sub-collections, which fold out when you open it. Drag collections to reorder them or to put one inside another, or use right click.
- GIFs play in a loop, in the gallery and the big viewer, at a speed you set in Settings or for each GIF.
- Import: drop pictures and videos from a folder or another app onto the gallery, or straight into the vault (the originals can be shredded after).
- Favourites are never deleted, also not with Delete all marked.
- Vault items can be favourites and marked for deletion, and show up in Favourites and Marked for deletion while the vault is open. F, X and Delete work in the big viewer too.
- Show in folder shows where a vault item's encrypted file is.
Prompt helper
- Sentence prompts are written by the writing model (TIPO only when picked in Settings). It keeps every person, how many there are, their positions and what they do, and describes the picture plainly, the way a camera would see it.
- Nothing it wrote is changed or left out afterwards anymore, apart from your tag blacklist.
- Tags written without commas are split up again, and words like "two men" are counted correctly.
Market and models
- Safety at a glance: every file shows its format (
.safetensorsholds only numbers and can't run code) and Civitai's virus and hidden code scans. A flagged file asks before it downloads. - Downloads and likes on every tile.
- With NSFW on, searches find NSFW models, and a search word no longer comes back empty with "this month".
- Missing parts as downloads: the text encoders and VAEs Anima, Flux, Qwen-Image and Z-Image need, the official files, pinned and checked. The Checkpoints tab says what a model is missing.
Fixes
- Newest first in the vault is in order again, and a new private picture shows at the top.
Download BetterComfySetup-1.4.0.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.4.0-win-x64.zip:
6a9d8303157c35d7176df3c25df8c1df4d03772cc9526016b07be26d48b33d3e
Better Comfy 1.3.0
A market for LoRAs and checkpoints, a model library, vault collections and a clean up for everything a picture leaves behind.
Market
- Browse Civitai inside the app (civitai.red or civitai.com), in the LoRAs tab and the new Checkpoints tab: search, filter by what a model is made for, sort by downloads, rating or date, and keep favourites.
- Downloads go straight into ComfyUI, are checked against Civitai's checksum and arrive with their link, trigger words, cover and what they are made for. A checkpoint without its own VAE and text encoder goes to diffusion_models by itself.
- SFW only until you switch NSFW on. The NSFW blur from Settings covers the Market too.
- Civitai API key (optional, Settings, Civitai, with a short how-to): some creators only allow downloads when logged in. The key is kept encrypted for your Windows account and only sent to Civitai with downloads.
- The Market only goes online while it is open. Covers you see there stay in memory and are never saved.
Checkpoints
- A library for your models: checkpoints and diffusion models side by side, each marked as such, with what they are made for, tags, notes, a cover, the page you got them from and their size.
- Fetch, favourites, sort by size or date, delete and Use on Image.
LoRA library
- Fetch: with a Civitai link saved, one click fills in the trigger words, a small cover and what the LoRA was made for. Fetch all does every LoRA with a link at once.
- Covers show in Add LoRA on the Image and Video pages.
- Tags show on the tiles, sort by size or date, delete a LoRA from the drive.
- Krea 2 and Anima LoRAs are recognized from the file.
Vault and gallery
- Collections inside the vault. Their names are encrypted too, and they only show under Vault while it is open.
- Pictures you put into a collection leave the general list (switchable in Settings, App, also for vault collections).
Clean up
- Settings, Clean up shows what making pictures leaves behind (prompt history, finished jobs, uploads and temporary files in ComfyUI, its job history, unused thumbnails, masks, video frames, logs) and removes each part or all at once.
- ComfyUI is tidied after every job: uploaded pictures, temporary copies and its history entry go.
- Deleting a picture also takes its prompt out of the prompt history and its job out of the queue list.
Little things
- Generate splits into Generate and Stop while something is being made. Stop only stops the current picture or video, the rest of the queue goes on.
- The count arrows double and halve (1, 2, 4, 8).
- Settings spread over two or three columns.
Download BetterComfySetup-1.3.0.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.3.0-win-x64.zip:
81429d53ec9b48a0961df74da9ae8b7a6049a739788c442c0eb7aa58a1645939
Better Comfy 1.2.0
Better prompts, a smarter LoRA library and Krea 2.
Prompt helper
- New models to pick from in Settings, Prompt helper, Models: uncensored writing models (Qwen3.5 2B, 4B and 9B) and TIPO, a model made for picture prompts. Each is an optional download to a folder you pick, with a short note on what it is good for.
- TIPO writes the tags for Illustrious, Pony and Anima and concrete sentences for the other models, in about a second. Tags stay true to your prompt: who is in it stays the same, nothing gets censored, no text or watermark tags.
- Turn into tags always gives tags now, also from long sentences.
- Refusals and empty answers are retried by themselves.
LoRA library
- Made for: Illustrious, Pony and plain SDXL LoRAs are told apart from their training details, and you can set it by hand. Add LoRA only lists the ones that fit the model, unclear ones with a ?.
- Tags to sort your LoRAs: character, art style, pose, clothing, detailer and your own, with suggestions from the file. Filter by them on the LoRAs page and in Add LoRA.
- Blur NSFW LoRAs (Settings, Privacy): names and previews of LoRAs tagged NSFW are blurred everywhere.
- Trigger words show up only once, also when they are saved twice or used by two LoRAs.
Models
- Krea 2 support. Its text encoder and VAE can be downloaded from the Model card if they are missing, and a Krea 2 file in the checkpoints folder can be moved to diffusion_models with one click. It needs a ComfyUI from June 2026 or newer.
Fixes
- The download bar now moves for files bigger than 2 GB.
- A deleted picture no longer stays in "Start from a picture".
- A stopped private job is also taken out of ComfyUI's history.
Download BetterComfySetup-1.2.0.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.2.0-win-x64.zip:
2d0f867baafceb83866370864d711017541d2abd0b5fc5538a33f2b6fe237b12
Better Comfy 1.1.0
A bigger gallery, a vault for what only you should see, and a helper for writing prompts.
Gallery
- Sort it out: favourite with F, mark with X and delete everything marked at once. Sort by date, type, model or LoRA, filter and search.
- Collections you fill yourself, or smart ones that fill themselves by model or LoRA.
Vault
- Pictures and videos only you can open, encrypted with your password (Argon2id and AES-256-GCM) and only ever decrypted in memory.
- Private next to Generate: the result goes straight into the vault, without a copy in your folders or in ComfyUI's.
- Move pictures in later or take them out again. Animate, upscale, edit and extend them without them leaving the vault.
- Recovery key, auto lock (it waits while the queue works), and a re-encrypt option when you change the password.
Prompt helper
- Improves, extends, shortens or writes prompts, as tags or as sentences depending on the model, and writes motion for videos.
- Runs on the processor, so generating stays as fast as before. Optional: set it up in Settings and it downloads once (0.5 to 2.6 GB) to a folder you pick, or use a .gguf model you already have.
- Tag suggestions while typing for tag based models, with a tag blacklist you can edit in Settings.
Privacy
- Screen capture blocking for the whole app or only while the vault is open.
- Deleting for good: shred files or hand them to Eraser instead of the Recycle Bin.
- This PC check in Settings: drive encryption, page file, hibernation and folders synced to the cloud.
Download BetterComfySetup-1.1.0.exe and run it, or let Better Comfy update itself.
SHA-256 of BetterComfy-1.1.0-win-x64.zip:
7bfa50742b67b2b7ce00d830e3ef7b803bb4b60808dd4adaccb926172b05827b
Better Comfy 1.0.0
The first release of Better Comfy: pictures and videos with the ComfyUI on your PC, without wiring a single node.
Pictures
- Every popular model: SDXL, Pony, Illustrious, SD 1.5, Flux, Qwen-Image, Anima and Z-Image. Better Comfy reads what a model is from inside the file and sets size, steps, sampler and guidance for it.
- Presets from Ultra Fast to Best, with an upscale pass for the top two. Speed models are spotted and run with few steps by themselves.
- Edit a part: paint over it, say what should be there, and only that part is drawn new.
- Upscale any picture with its own model, prompt and seed, or just enlarge it.
- Compare: the same picture with one setting changed, side by side.
Videos
- WAN 2.2 and 2.1, from a picture or from words only.
- Seamless loops that find the smoothest loop point, plus ping-pong, crossfade and start to end picture.
- Extend a video from its last frame and get one longer video. Animate a whole folder of pictures at once.
Everything around it
- LoRA library with trigger words, notes, links and previews. WAN high and low files are paired up.
- Queue with time estimates, single pictures you can take out or stop, and an action for when it is done (sleep, shut down and more).
- Gallery with every result and its settings, a big viewer with zoom, and joining videos.
- Finds ComfyUI by itself (Pinokio, portable or ComfyUI Desktop). If it can't, it asks for the folder on the first start.
Download BetterComfySetup-1.0.0.exe and run it. Better Comfy keeps itself up to date from then on.
SHA-256 of BetterComfy-1.0.0-win-x64.zip:
77fb1f8e3c97e69227d888bee45a76a1456dcc3768abbb66e6302a4931f0abb1