StabilityMatrix

Fixed

• Fixed prerequisite install not unpacking due to improperly formatted 7z argument (Caused the "python310._pth FileNotFoundException")
• Fixed #301 - Package updates failing silently because of a PortableGit error

Fixed

• Fixed #297 - Model browser LiteAsyncException occuring when fetching entries with unrecognized values from enum name changes

Fixed

• 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 dialog

Fixed

• Fixed errors preventing Model Browser from finding results with certain search queries

Fixed

• Fixed InvalidOperationException during prerequisite installs on certain platforms where process name and duration reporting are not supported

v2.6.2

Changed

• Backend changes for auto-update schema v3, supporting customizable release channels and faster downloads with zip compression

Fixed

• Better error reporting including outputs for git subprocess errors during package install / update
• Fixed 'accelerate' is not recognized as an internal or external command error when starting training in kohya_ss
• Fixed some instances of ModuleNotFoundError: No module named 'bitsandbytes.cuda_setup.paths' error when using 8-bit optimizers in kohya_ss
• Fixed errors preventing Inference outputs from loading in the img2img tabs of other packages

Changed

• NVIDIA GPU users will be updated to use CUDA 12.1 for the InvokeAI package for a slight performance improvement
  • Update will occur the next time the package is updated, or on a fresh install
  • Note: CUDA 12.1 is only available on Maxwell (GTX 900 series) and newer GPUs

Fixed

• Reduced the amount of calls to GitHub to help prevent rate limiting
• Fixed rate limit crash on startup preventing app from starting

Added

• Added Output Sharing option for all packages in the three-dots menu on the Packages page
  • This will link the package's output folders to the relevant subfolders in the "Outputs" directory
    • When a package only has a generic "outputs" folder, all generated images from that package will be linked to the "Outputs\Text2Img" folder when this option is enabled
• Added Outputs page for viewing generated images from any package, or the shared output folder
• Added Stable Diffusion WebUI/UX package
• Added Stable Diffusion WebUI-DirectML package
• Added kohya_ss package
• Added Fooocus-ControlNet-SDXL package
• Added GPU compatibility badges to the installers
• Added filtering of "incompatible" packages (ones that do not support your GPU) to all installers
  • This can be overridden by checking the new "Show All Packages" checkbox
• Added more launch options for Fooocus, such as the --preset option
• Added Ctrl+ScrollWheel to change image size in the inference output gallery and new Outputs page
• Added "No Images Found" placeholder for non-connected models on the Checkpoints tab
• Added "Open on GitHub" option to the three-dots menu on the Packages page

Changed

• If ComfyUI for Inference is chosen during the One-Click Installer, the Inference page will be opened after installation instead of the Launch page
• Changed all package installs & updates to use git commands instead of downloading zip files
• The One-Click Installer now uses the new progress dialog with console
• NVIDIA GPU users will be updated to use CUDA 12.1 for ComfyUI & Fooocus packages for a slight performance improvement
  • Update will occur the next time the package is updated, or on a fresh install
  • Note: CUDA 12.1 is only available on Maxwell (GTX 900 series) and newer GPUs
• Improved Model Browser download stability with automatic retries for download errors
• Optimized page navigation and syntax formatting configurations to improve startup time

Fixed

• Fixed crash when clicking Inference gallery image after the image is deleted externally in file explorer
• Fixed Inference popup Install button not working on One-Click Installer
• Fixed Inference Prompt Completion window sometimes not showing while typing
• Fixed "Show Model Images" toggle on Checkpoints page sometimes displaying cut-off model images
• Fixed missing httpx package during Automatic1111 install
• Fixed some instances of localized text being cut off from controls being too small

Fixed

• Fixed error got an unexpected keyword argument 'socket_options' on fresh installs of Automatic1111 Stable Diffusion WebUI due to missing httpx dependency specification from gradio

Added

• Added Russian UI language option, thanks to aolko for the translation

Added

• Added Spanish UI language options, thanks to Carlos Baena and Lautaroturina for the translations
• Manual input prompt popup on package input requests besides Y/n confirmations
• Added --disable-gpu launch argument to disable hardware accelerated rendering

Fixed

• Fixed infinite progress wheel when package uninstall fails

Fixed

• Fixed #208 - error when installing xformers

Added

• Added French UI language option, thanks to eephyne for the translation

Fixed

• Fixed Automatic 1111 missing dependencies on startup by no longer enabling --skip-install by default.

Added

• Right click Inference Batch options to enable selecting a "Batch Index". This can be used to reproduce a specific image from a batch generation. The field will be automatically populated in metadata of individual images from a batch generation.
  • The index is 1-based, so the first image in a batch is index 1, and the last image is the batch size.
  • Currently this generates different individual images for batches using Ancestral samplers, due to an upstream ComfyUI issue with noise masking. Looking into fixing this.
• Inference Batches option now is implemented, previously the setting had no effect

Changed

• Default upscale factor for Inference is now 2x instead of 1x

Fixed

• Fixed batch combined image grids not showing metadata and not being importable
• Fixed "Call from invalid thread" errors that sometimes occured during update notifications

Added

• --skip-install default launch argument for Automatic1111 Package

Fixed

• Fixed Prompt weights showing syntax error in locales where decimal separator is not a period

We're introducing Inference in this update, a built-in Stable Diffusion interface, powered by any running ComfyUI package.
It's still in early development, so try it out and let us know of any issues or suggestions!

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 Browser
• Added notification flyout for new available updates. Dismiss to hide until the next update version.
• Added Italian UI language options, thanks to Marco Capelli for the translations

Changed

• Model Browser page size is now 20 instead of 14
• Update changelog now only shows the difference between the current version and the latest version

Fixed

• Fixed #141 - Search not working when sorting by Installed on Model Browser
• Fixed SD.Next not showing "Open Web UI" button when finished loading
• Fixed model index startup errors when ./Models contains unknown custom folder names
• Fixed ストップ button being cut off in Japanese translation
• Fixed update progress freezing in some cases
• Fixed light theme being default in first time setup window
• Fixed shared folder links not recreating fully when partially missing

Added

• LDSR / ADetailer shared folder links for Automatic1111 Package

Changed

• Made Dark Mode background slightly lighter

Fixed

• Fixed "Library Dir not set" error on launch

Added

• Added button to toggle automatic scrolling of console output

Fixed

• Fixed #130 ComfyUI extra_model_paths.yaml file being overwritten on each launch
• Fixed some package updates not showing any console output
• Fixed auto-close of update dialog when package update is complete

Added

• Added "--no-download-sd-model" launch argument option for Stable Diffusion Web UI
• Added Chinese (Simplified) and Chinese (Traditional) UI language options, thanks to jimlovewine for the translations

Changed

• Package updates now use the new progress dialog with console output

Fixed

• Updated Japanese translation for some terms