StabilityMatrix posts
Posts: 105
Files: 310
v2.7.0 Preview 4
Finally here after almost 2 months~ many backend improvements that should make it easier and quicker to roll out new features for Inference, like the new ImageToImage project type. You can now also add and reorder any number of custom steps in Inference, including guidance addons for samplers like ControlNet or saving intermediate images.We're hoping to add more step and addon options as we progress, so it will be possible to build out and sa...
v2.7.0 Preview 3
New model browser tab for Huggingface base model downloads, and more options for syncing connected model info for individual files, directories, or all models! Looking to get Inference modular steps and Image to Image out in next preview to be in time for 2.7.0 release.Thanks again for everyone's continued support 🤗Added Added "Find Connected Metadata" options for root-level and file-level scans to the Checkpoints page Added "U...
v2.7.0 Preview 2
Added Added System Information section to SettingsChanged Moved Inference Settings to subpageFixed Fixed crash when loading an empty settings file Improve Settings save and load performance with .NET 8 Source Generating Serialization Fixed ApplicationException during database shutdown
v2.7.0 Preview 1
Fixed Fixed control character decoding that caused some progress bars to show as \u2588 Fixed Python rich package's progress bars not showing in console Optimized ProgressRing animation bindings to reduce CPU usage Improved safety checks in custom control rendering to reduce potential graphical artifacts Improved console rendering safety with cursor line increment clamping, as potential fix for #111
v2.7.0 Dev 4
Fixed- Fixed #290 - Model browser crash due to text trimming certain unicode characters- Fixed #297 - Model browser LiteAsyncException occuring when fetching entries with unrecognized values from enum name changes- Fixed error when receiving unknown model format values from the Model Browser- Fixed process errors when installing or updating Pip packages using the Python packages dialogThis update also fixes a potential issue with the auto-updater...
v2.7.0 Dev 3
Added- New package: RuinedFooocusModel Browser- Right clicking anywhere on the model card will open the same menu as the three-dots button- New model downloads will save trigger words in metadata, if available- Model author username and avatar display, with clickable link to their profileCheckpoints Page- Added "Copy Trigger Words" option to the three-dots menu on the Checkpoints page (when data is available)- Added trigger words on check...
v2.7.0 Dev 2
Small update with UI improvements, drag-drop moving of checkpoint files, and many other fixes. This also merges the latest changes from 2.6.2AddedGeneral- Added an X button to all search fields to instantly clear them (Esc key also works) Outputs Page - Added Refresh button to update gallery from file system changesCheckpoints Page- Added the ability to drag & drop checkpoints between different foldersChangedOutputs Page- Updated button and menu ...
Receiving Dev/Preview auto-updates
Starting with v2.7.0-dev.1 release of Stability Matrix, you can connect your Patreon account to receive auto-updates for the Dev and Preview channels.Please follow this guide below, if you have other questions feel free to message us here or on Discord.1. Go to the Settings and click Accounts2. Connect your Lykos account3. If you don't have an account yet, click the Sign up tab and enter your desired email, username, and password to Continue4...
v2.7.0 Dev 1
Hey all, due to some models recently requiring CivitAI logins to download, we've pushed forward the adding of the accounts integrations feature. You can now connect your CivitAI account via API Key under Settings > Accounts. There is also the new Settings > Updates page to toggle auto-update notifications and choosing the update channel.You can now create a Lykos account in the app and connect with Patreon to receive Preview / Dev updates thr...
v2.6.0 Preview 2
Hi all, here with another preview build for v2.6.0, this time with some bug fixes & a new package option.Added- New installable package, kohya_ss- Added Upgrade & Downgrade buttons to the package manager dialogFixed- Package manager now respects index URLsThank you for your continued support & feedback!
v2.6.0 Preview 1
Hi all, we've been working on getting ControlNet and custom step support added to Inference for v2.6.0, but due to preprocessor support not being built-in we're needing to work on common elements like plugin management before being able to support ControlNet fully as a simple and intuitive feature.We're tidying up v2.6.0 for stable release soon, with ControlNet and other Inference features planned for v2.7.0. Many fixes in this previe...
v2.6.0 Dev 3
Hi again o/ thanks for everyone's continued support! Mostly UI improvements and bug fixes in this update, including adjustable scaling for images on Output page and Inference gallery.Simply hold the Ctrl key while scrolling your mouse wheel to make images larger or smaller, the scaling level is saved to your settings.Added Reworked Package installer with GPU profile compatibility badges Automatic1111 Package will now install the lshqqytiger/s...
v2.6.0 Dev 2
Added New Installable Package anapnoe/stable-diffusion-webui-ux Outputs page multiple selection and search Custom file names conflicts should now have correct file extensions Inference gallery hover tooltip now has more information like model, sampler, seed, and full path Inference model selection now has a tooltip on hover, showing full path and preview imageChanged Package installs and updates in now use git operations for better compatibility ...
Next features for Inference ✨
Thanks for voting on our last poll for ReadyToRun, pretty obvious everyone wanted enabling it, so ReadyToRun was enabled for our releases from v2.5.0 onwards. We'll look to use compressed downloads soon to reduce download time for updates as well.Here's a new poll for what we should work on next for Inference. You can pick multiple options but try not to pick all of them so we can prioritize something :)Feel free to suggest other ideas in...
v2.6.0 Dev 1
Hi again, thanks for all your support and suggestions after our Inference release. We're working on adding support for shared output image viewing next, along with custom file name formatting.Added- Added new shared outputs page, configure packages to use shared output directories by toggling "Output sharing" in their config menus- Inference image loading is now done over API instead of file paths, the latency from generation completi...
v2.5.0 Preview 3
Added Added FreeU option to Inference (https://github.com/ChenyangSi/FreeU)Fixed Fixed bug where gallery did not show images when in floating mode
v2.5.0 Preview 2
Drag-drop import for Inference now supports files with Inference or Automatic1111 metadata, more metadata types may be supported in the future as well. Just a few more things to finish until release :3Added- You can drag in external .png files to load project states, similar to the image gallery. Supports images generated from Inference or with Automatic1111 metadata- Added "Restore Default Layout" menu option for resetting dock layoutFix...
Should we build releases with ReadyToRun?
ReadyToRun is an option that precompiles some JIT code ahead of time to reduce startup time and latency in loading content / pages for the first time.In our testing this reduced the time for starting Stability Matrix from ~1.6s to ~0.7s. The difference may be smaller or larger depending on the amount of models / packages you have.However this also increases the size of the app and updates:exe: (150 MB) ➝ (250 MB)zip: (60 MB) ➝ (100 MB)
v2.5.0 Preview 1
Inference is finally merged as we get ready for 2.5.0 release 🎉Added- Added Inference, a built-in native Stable Diffusion interface, powered by ComfyUI- Added option to change the Shared Folder method for packages using the three-dots menu on the Packages page- Added the ability to Favorite models in the Model Browser- Added "Favorites" sort option to the Model BrowserChanged- Model Browser page size is now 20 instead of 14Fixed- Fixed...
v2.5.0 - Dev 5
Added for Inference UI Upscalers will now show some default models if you don't have them yet, simply click to download and useFixes Fixed Api Error when using extra networks (Lora / LyCORIS) with nested subfolder paths Image viewer now scales up more to fit available space Fixed issue where gallery would have rendering flickering Fixed directory not exist errors on startup for model folder Performance and visual enhancements for the transpar...
v2.5.0 Dev 4
Added- Add Shift+Insert hotkey for pastingIncludes Fixes from v2.4.5- Fixed "Library Dir not set" error on launch
v2.5.0 - Dev 3
Generation parameters and arrow key navigations for gallery viewer, and some miscellaneous enhancements and fixes. Should be close to release of v2.5.0 now.Added for Inference UI- Generation parameters popup for image viewer (press top left info button to toggle)- Arrow key navigation for image viewer (up/right for forward, down/left for back), can also click outside the viewer to close it- Add context menu for prompt for copy/paste/cut, Shift+De...
v2.5.0 - Dev 2
Now includes an upscaler tab in the Inference UI. Click the + button in the top tab bar and select "Upscale".Added for Inference UI- Added Upscaler tab with sharpening optionsFixes- Fix syntax error when using negative LORA weightsThank you for your support!
v2.5.0 - Dev 1
Now you can drag images from the gallery to load the project state. This also includes latest changes from 2.4.2 including the Japanese language option.Added for Inference UI - Added loading of project states by dragging the images onto the project space, currently this only works by dragging from the gallery, but this should eventually work for normal files, including ones only with A1111-style metadata.- Search bar for images is now implemented...
v2.4.0 Dev 6
o/ Image gallery for Inference is here. Also includes a reworked installer with more options, and support for a new Package.Added - New installable Package - Fooocus-MREAdded for Inference UI - Added Image gallery panel, shows a grid of generated outputs. New generated images in Inference will now be at "(Data)/Images/Inference/" (Note search bar is not functional yet) - Left clicking the images will open a bigger preview, and right click...
v2.4.0 Dev 5
Fixes - Fix batch images error message and not showing output- Hotkeys for save/load should now work without focusing the menu
v2.4.0 Dev 4
Added for Inference UI - SDXL Refiner support, toggle the refiner selection in the gear icon next to models, in this mode you can adjust refiner steps in the sampler config- Scheduler selection (Normal, Karras, etc.)- Open project file tabs now are persistent through relaunches, due to significant changes to the project file format previous project files will no longer be compatible. There should be no further backwards-incompatible file format c...
v2.4.0 Dev 3
More features completed for Inference UI, including Lora / LyCORIS support and VAE. Getting closer to initial release now!Added for Inference UI- Lora / LyCORIS support, using syntax of <lora:model:weight> or <lyco:model:weight>- Settings button next to models can now toggle custom VAE selection- Settings option to enable/disable image viewer pixel grids- New tokenizer and language syntax for extra networks support. A prompt will show syntax and ...
v2.4.0 Dev 2
Fixes for Inference UI- Image pan/zoom should work properly with multiple images nowAlso includes some in-progress fixes for v2.3.3 - Fixed GPU recognition for Nvidia Tesla GPUs- Fix checkpoint file index extension identification with some path names- Fixed issue where config file may be overwritten during Automatic1111 package updates- Fixed "Directory Not Found" error on startup when previously selected Data directory does not exist
v2.4.0 Dev 1
Bunch of updates for Inference UI! This also includes merged changes from the v2.3.0 release.Added- New tokenizer-based prompt auto-complete. Compatible with any CSV for the a1111-sd-webui-tagcomplete plugin, you can import them in a new settings section. - Here are some common ones (danbooru.csv, e621.csv)- Zoom/Pan enabled image viewer, switches to pixel grid at high zoom levels- Hires fix supports ESRGAN / model-based upscalers, simply add mo...
