v0.9.0-beta2

• Fix Flux inpainting and tested with Flux Fill
• Try different mask approach for Flux and Sd3.5 regional prompting

To install new version with existing installation, change the version number in version.txt to v0.9.0-beta2 and then run the bat install script,
OR
Use the one click installer included in the release.

Check previous release notes for more details about v0.9.0

v0.9.0-beta1

• Revamp LORA; Civit integration, dedicated section, LORA modal
• Support SD 3.5
• Multiple select text encoder for convenience
• Add Model config for apply parameters presets
• Add memory stats for Comfy integration through ComfyUI-Crystools
• Fix Flux generation with second pass (txt2img & img2img) including upscaling
• CTRL + Enter shortcut to interrupt ongoing generation
• UI tweaks and color palette updates
• Fix Forge integration with latest version

New extension required: https://github.com/crystian/ComfyUI-Crystools

To install new version with existing installation, change the version number in version.txt to v0.9.0-beta1 and then run the bat install script,
OR
Use the one click installer included in the release.

Check previous release notes for more details about v0.9.0

v0.9.0-alpha4

v0.9.0-alpha4

• Add Flux regional prompting support (Comfy)
• Use random noise seed for noise injection (instead of hardcoded 1)

To install new version with existing installation, change the version number in version.txt to v0.9.0-alpha4 and then run the bat install script,
OR
Use the one click installer included in the release.

Check previous release notes for more details about v0.9.0

v0.9.0-alpha3

Changes:

• Add dropdown groups support (ex: model types)
• Refine dropdown search with a higher match threshold for better matches
• Save last generation main params and load them on refresh/restart
• Add Skimmed CFG (Comfy)
• Add Split rendering for noise injection (Comfy Flux only)
• Separate A1111 and Forge (ground work for updating Forge integration)
• Increase status polling interval to reduce backend CPU usage
• Add CTRL + SHIFT + L shortcut to un/lock seed

To install new version with existing installation, change the version number in version.txt to v0.9.0-alpha3 and then run the bat install script,
OR
Use the one click installer included in the release.

NOTE:
Two new Comfy extensions need to be added to your comfy installation:

• https://github.com/Extraltodeus/Skimmed_CFG
• https://github.com/cubiq/ComfyUI_essentials

Check previous release notes for more details about v0.9.0

v0.9.0-alpha2

• Add scaling support for Comfy Flux inapinting (without model)
• Add t5 GGUF clip model support (download from https://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/tree/main)

Only GGUF quantized model format is supported for the time being.

Make sure to install ComfyUI-GGUF to your ComfyUI installation.

You can grab a model from here.

Note: you'll need to install the flux vae and place it under Comfy/models/vae, and also 2 clip models clip_l and t5xxl and place them under Comfy/models/clip.

To install new version with existing installation, change the version number in version.txt to v0.9.0-alpha2 and then run the bat install script,
OR
Use the one click installer included in the release.

v0.9.0-alpha

• Add initial Flux support including txt2img and LORA support.

Only GGUF quantized model format is supported for the time being.

Make sure to install ComfyUI-GGUF to your ComfyUI installation.

You can grab a model from here.

Note: you'll need to install the flux vae and place it under Comfy/models/vae, and also 2 clip models clip_l and t5xxl and place them under Comfy/models/clip.

To install new version with existing installation, change the version number in version.txt to v0.9.0-alpha and then run the bat install script,
OR
Use the one click installer included in the release.

v0.8.0

• Revamped the img2img workflow
• Added soft inapinting with mask blur control
• Added fooocus inpainting
• Added IPA integration with masking
• Added CN union integration

Make sure to install those extensions to your ComfyUI installation:
comfyui-inpaint-nodes
ComfyUI_IPAdapter_plus

v0.8.0-beta2

• Update to work with latest version of Comfy (fixed the breaking regression)
• Defaulting to base layer image if no image is set on an IPA layer
• Soft inpainting was showing on txt2img; now only shows with img2img
• Fixed default value of mask blur
• Switched the order of CN fields to show model then union type then preprocessor

Also, pushed an update to A8R8_ComfyUI_nodes (UPDATE EXTENSION) with a work around for using IPA with Attention Couple with an even number of regions.

To install new version with existing installation, change the version number in version.txt to v0.8.0-beta2 and then run the install script,
OR
Use the one click installer.

v0.8.0-beta

v0.8.0-beta

• Revamped the img2img workflow
• Added soft inapinting
• Added fooocus inpainting
• Added IPA integration with masking
• Added CN union integration

To install new version with existing installation, change the version number in version.txt to v0.8.0-beta and then run the install script,
OR
Use the one click installer.

New Comfy extension required:
comfyui-inpaint-nodes
ComfyUI_IPAdapter_plus

Also, make sure you A8R8 ComfyUI nodes installation is updated to latest.

v0.7.1

• Revamped bundled release for much faster and more compact installation
• Add DEFAULT_BACKEND env variable support to configure which backend is used on startup (set in user.sh)
• A number of optimizations for async internal calls

For Windows with WSL:

1. Download A8R8_One-Click-Installer.zip
2. Unzip the downloaded archive
3. Run one-click-installer-wsl.bat
4. Run run-windows-wsl.bat to run the application

You can now set the default backend in user.sh; check example_user.sh for reference.

Make sure to check the main readme for more details.

For Ubuntu 22.04:

A new package installer has been added, check readme for more details.