v1.1.0 release: ComfyUI support, ControlNet improvements

New features:

ComfyUI image generator support

• ComfyUI can now be used in place of the Stable Diffusion WebUI for image generation, present as a new entry in the list of image generators. All major features are supported, with the "Interrogate" button being the only major exception.
• Dynamic ComfyUI workflow generator created to handle ComfyUI's unique API, and provide some flexibility for things like ControlNet use.
• "Extras" tab added to the image generation panel, holding controls for generator-specific features. Config selection, tiled VAE options, inpainting model loading, and clear memory button added to the ComfyUI extras tab.

Improved image scaling controls

• New dropdown added to explicitly select between basic scaling, image generator powered scaling, and advanced latent upscaling.
• Add extended support for selecting a ControlNet tile model and preprocessor, and setting tile preprocessor parameters.
• Use of the "Ultimate SD Upscale" script can now be directly enabled and disabled.
• When using latent upscaling, denoising strength and step count can now be set in the image scaling window.

Improved ControlNet support

• Improved management of ControlNet data and added useful defaults, so the "Control Type" dropdown will always be available.
• Add hard-coded preprocessor parameters for WebUI APIs that don't provide those, so parameters from standard preprocessors should always be available.
• Add a "preprocessor preview" button, showing ControlNet preprocessor output directly in the ControlNet panel.
• WebUI ControlNet resolution, image scaling mode, and control mode options can now be set in the ControlNet panel.

Other

• Added a Help menu, linking to GitHub documentation
• Stable Diffusion model selection and CLIP skip are now available directly in the Image Generation panel
• Added "randomize", "reuse last value" buttons to the seed input field.
• When using the WebUI generator, rovide access to batch variation seed settings, seed resizing, and tile/face restore toggles within the new "extras" tab within the Image Generation panel.

Bugfixes

• Changing the URL of an image generator within the image generator selection window should be possible again.
• Fixed some errors where ControlNet widgets weren't properly disabled when the ControlNet unit is disabled.
• Image generator activation failures should be a lot clearer now.
• Fixed some issues caused by changing lists of various options when switching between image generators.
• Fixed an error blocking access to settings if the Stable Diffusion WebUI stops responding.

Development changes, documentation and cleanup

• Increased minimum Python version from 3.9 to 3.11 to take advantage of typing system improvements.
• In the settings, the "Stable Diffusion" and "Connected generator" options have been merged, and some items have been moved to the Image Generation panel and Scale Image window.
• Added expanded Stable Diffusion installation guide,
• Improved documentation layout, setup instructions in the image generator selection window.
• When upscaling adds a new image layer, that layer now becomes the active layer.
• Placeholder preview image in the LoRA panel is no longer excessively tall.

v1.0.0 release

v1.0.0: Initial release as a full image editor and Stable-Diffusion interface

This is nearly a complete rewrite, introducing the following changes and more:

• Support for Stable-Diffusion inpainting.
• Support for ControlNet.
• A complete redesign of the interface.
• A full-featured layer stack.
• Support for the libmypaint brush engine.
• A full suite of editing tools, in-line with what you'd expect in any image editor.

Complete documentation is available for all features, see the README for details.

v0.1.0 Release

Initial public release as IntraPaint.