v1.4.1

Please note:

• ComfyUI is now available as a backend. Install the extension from here: Comfy-Photoshop-SD

• Load any ComfyUI workflow and use it directly in Photoshop. This allows you to utilize the latest features from within Photoshop before they’re natively implemented into the plugin.

New Features:

• Added support for Real-time ComfyUI workflows. This feature works with any workflow that you run in ComfyUI using the Auto Queue option.
  
  

• The real_time_lcm_sketching workflow is quite useful for composing images using both prompts and sketching.

realtime_sketching.mp4

• The real_time_lcm_txt2img workflow is slower than the sdxl turbo, but it allows you to use your favorite SD 1.5 models, giving you better control over the style of the image.

real_time_txt2img.mp4

• The real_time_lcm_img2img workflow is ideal when working with img2img.

real_time_lcm_img2img.mp4

• The sdxl_turbo_txt2img workflow is by far the fastest, thanks to its use of sdxl turbo model, similar to lcm_txt2img but around 4 to 8 time faster.

•  Introduced two sets of buttons for rearranging the ComfyUI UI:

1. The set on the right will move the node one step up or down.
2. The set on the left will move the node to the topmost or bottommost position of the UI.
   

Bug Fixes:

• Fixed a bug that caused the plugin to check for ComfyUI multiple times on startup and prompt the user with error messages.

• Fixed Height: temporary workaround to prevent UI from collapsing on each image reload.
  

• Fix: All images uploaded to ComfyUI will now be stored inside the ComfyUI/input/Auto-Photoshop-SD directory.
• Fix: The searchable menu now shows all items when clicked on, so there’s no need to delete their content first.
• fix: The sizing adjustment of output image dimensions is now more accurate, increasing in steps of 8.
• fix: Reset workflow allows you to reset settings while keeping your custom UI layout intact.
  

v1.4.0

Updates and New Features Include:

New Features:

• ComfyUI is now available as a backend. Install the extension from here: Comfy-Photoshop-SD
• Load any ComfyUI workflow and use it directly in Photoshop. This allows you to utilize the latest features from within Photoshop before they’re natively implemented into the plugin.
• Custom ComfyUI workflows can be used regardless of whether you’re in Automatic1111 mode or ComfyUI mode.
• The Edit Nodes section in Custom ComfyUI Workflow lets you rearrange and relabel nodes, as well as save your settings or restore the original settings from the API file.
• Load LCM Lora and generate images in 1 to 4 steps instead of 20. This works with any model. Set CFG to low value of 1 or 2
• The added “Make Square” option in Lasso Inpaint allows you to generate in square or original ratio.
• A high-resolution fix option has been added to all modes (txt2img, img2img, inpaint, and outpaint) in ComfyUI only.
• Model, Lora, and ControlNet related Photoshop menus have been replaced with a custom searchable menu.
• The ‘DuckDuckGo_Search’ library has been updated to version v3.9.9 by deedy5.

Bug Fixes:

• Fixed a bug related to scrolling by conlanpatrek.
• Resolved an issue with saving prompts library by XiaoXiaoJiangYun.
• The prompts library will now create a file on its first run by XiaoXiaoJiangYun.
• The default value of the ControlNet module in ADetailer has been changed to None by Anson2048.
• Fixed an issue where some interface element labels were not affected by localization.

Important Notes:

• You can generate your own custom presets when using ComfyUI. However, the presets currently provided with the plugin will only work with Automatic1111. This will be addressed in a future update.
• Use the Refresh button after switching backends between “Automatic1111” and “ComfyUI” to refresh the interface with appropriate data.

Contributors: @AbdullahAlfaraj, @conlanpatrek, @deedy5, @XiaoXiaoJiangYun, @Anson2048

v1.3.3

Updates and New Features Include:

New Features:

• Allow a shorter minimum window height
• Add an option to disable server status alert messages.
• Add ‘Clip Skip’ slider to the settings tab.
• Add different minimum size options of 512, 768, and 1024 for the selection box, options are only visible in the "ratio mode".
• Update the width and height sliders when changes are made to selection modes or minimum size options.
• Resize the prompt box to fit larger text when changes are made.
• Enhance the lasso inpainting result by incorporating an offset into the lasso selection.
• Refactor the Lexica tab using Typescript and React and relocate it to the second window.
• Re-Select the last active layer upon interruption.
• Load settings from images generated by ‘Automatic1111’, must include metadata.
• Introduce custom width and height sliders for the ‘Ultimate SD Upscaler’
• Introduce an option for tiling textures.
• Set the custom preset panel to be closed by default.
• Introduce a quick access button for the History and Lexica panels.
• Update the ‘DuckDuckGo_Search’ library to version v3.9.3 by deedy5

Bug Fixes:

• Fix the issue where toggling the ‘Smart Object’ options has no effect.
• Resolve crashes related to React’s strict mode.
• Fix the issue where the prompt shortcut menu does not refresh after loading.
• Prevent changes to the model when using the Smart Presets.
• Fix the issue where the inpainting conditional mask is permanently set to a value of 1.
• Address the issue of missing ControlNet modules and models when using the smart presets.
• Address the issue of the missing extension type when the plugin is first started up.

Contributors:
@AbdullahAlfaraj , @zombieyang , @deedy5 , @alinsavix

v1.3.2

Updates and New Features Include:
New Features:

• Added a new masking mode “Fully Transparent”, now enabled by default by (ZombieYang).
• History is now segmented by session and relocated to the plugin's second panel.
• Loading an image from history will automatically apply its mask.
• Added a Textual Inversion menu.
• Fixed adetailer issue by (Bing-su).
• Refactored the main sd tab using TypeScript and React.
• Added four prompt text areas, allowing work on multiple prompts without losing any of them.
• Fixed a bug: changing the width and height sliders after starting a session will not affect the final image size.
• added new presets that mimic Photoshop’s generative fills feature.
• Added a new custom preset panel to record and edit your own presets.
• Introduced a new expand Photoshop mask slider in the viewer panel used during placement of an image.
• Alter the color of active controlnet units from green to red upon deactivation of the controlnet tab.
• The sampler section will display the selected sampler even when the section is hidden.
• Fixed: switching between languages will now update viewer and history panel labels.
• Fixed: one button prompt no longer requires restarting Photoshop if it fails to load config from Auto1111 on first attempt.
• Fixed: checkbox in setting tab only toggles when cursor hovers on the text.

Contributors:
@AbdullahAlfaraj  , @zombieyang , @Bing-su

v1.3.1

A Quickfix for breaking change of ControlNet:

1. use control_net_unit_count to get the count of controlnet. #362

v1.3.0

Updates and New Features Include:
New Features:

• Added new Lasso Inpaint mode, allowing inpainting from lasso selection without the need to paint a mask.
• Complete refactoring of ControlNet tab in TypeScript, React, and MobX by zombieYang
• Reimplemented Preview, Viewer, History, and Image Search sections in TypeScript and React.
• Added quick access panel for ControlNet, Viewer, and Preview tabs
• Added toolbar for quick access to buttons and actions
• Added support for ControlNet ‘Generative Fill’ in inpaint/outpaint modes
• New icons design by amir167
• Chinese localization mainly done by MKhichain with contributions from zombieYang, and implemented by zombieYang.
• Added option to add custom mask for inpaint ControlNet models, useful in combination with txt2img mode
• Added ‘Auto Image’ option for ControlNet to automatically load input image from selection area on canvas
• Redesigned Preview section to be collapsible with progress bar on top
• Added support to automaticly generating mask with segment anything using keywords and prompt
• Added ‘Disable ControlNet Tab’ option to quickly disable ControlNet without disabling units individually
• Added error message to inform users to download ‘Auto-Photoshop-SD’ from automatic1111 extension tab for better inpaint and outpaint
• Third-party extensions/scripts checked for installation and notify users if any is missing
• Added mask preview that updates after changing Mask Expansion or Mask Blur sliders
• ‘Generate More’ buttons no longer require a selection area
• Added option to log errors to a log file, enabled from Settings tab
• Renamed ‘Continue Session’ button to ‘Reuse Selection’
• Added ControlNet Filters menu for quick setup of ControlNet unit (requires ‘Auto-Photoshop-SD’ extension installed inside Auto1111)
• Added option to automatically mask output image layers in inpaint/outpaint mode to prevent loss of resolution from diffusion operation
• Added support for ‘One Button Prompt’ for easy generation and exploration of unique prompts with one click
• Plugin now supports layer-less mode, initializing without generating intermediary/temporary layers
• Improved ControlNet user interface by making it collapsible (by AbdullahAlfaraj and zombieYang)
• Added option to select interpolation method for scaling images (bicubic, bilinear, or nearest neighbor)
• Added option to switch languages from Settings tab (English, Chinese) by zombieyang
• Added option for algorithm used to mask generated image when placed as layer (keep borders, keep corners)
• Added support for ‘After Detailer’ extension for easy inpainting of characters, faces, and people without drawing a mask
• ‘Scripts’ and ‘Always on Scripts’ display error message when unavailable, instructing user to update if not installed
• Added VAE dropdown menu in ‘Settings’ tab
• Added support for ‘Pixel Perfect’ option in ControlNet tab
• Added ‘Control Mode’ options in ControlNet tab for more precise control over image generation by shifting emphasis from prompt to ControlNet by AbdullahAlfaraj and zombieYang
• Improved ‘Always on Scripts’ interface by making it collapsible
• Sped up plugin load time when working with remote Auto1111 instance

Bug Fixes:

• Fixed bug where guessmode was wrongly used with ControlNet API v2
• Fixed issue where mask expansion slider would blur the mask when set to value 0
• Fixed performance issue when switching between modes caused by operating on large history states
• Fixed issue related to TypeError: ‘<’ not supported between instances of ‘str’ and ‘int’
• Added error message if Ultimate SD Upscale is not installed
• Fixed issue where ControlNet annotation map would load on canvas and in Viewer tab when current script is active but not installed
• Fixed VAE menu list crash and disappearance from Settings tab
• Fixed inability to enter text into text field when window size is large
• Fixed sizing of plugin, UI no longer overflows outside of plugin border
• Updated to use new Lora API to work with both Auto1111 and AutoVlad
• Fixed missing plugin icon from Photoshop plugin panel
• Fixed error ‘command Show is not available’, usually encountered when starting with document with image present as background
• Fixed empty ‘Hires Upscaler’ list when using plugin with remote instance of Automatic1111
• Removed dependency on API config/endpoint
• Fixed ‘img2prompt’ Interrogate button for remote Auto1111 users

Contributors:
@AbdullahAlfaraj , @zombieyang, @MKhichain, @amir167

Support us on Patreon to help us continue developing the Auto-Photoshop-StableDiffusion-Plugin and get early access to new features, exclusive art tutorials and more!

v1.2.5

Updates and New Features Include:

New Features:

• Added support for the “Ultimate SD Upscale” script, allowing images to be upscaled to super resolutions while adding details. make sure to install it from Auto1111)
• Controlnet resolution, threshold A & B sliders are now hidden by default (by zombieyang).
• Removed the model option when using the “reference” controlnet module. (by zombieyang).
• Module names now use similar naming conventions the Auto1111 Webui to avoid confusion. (by zombieyang).

Bug Fixes:

• Improvements to the image search code and updated the duckduckgo_search dependency to version 3.7.1 (by deedy5)
• Implemented a workaround for an issue where Controlnet detection maps would not be imported into the plugin when used together with scripts like “Ultimate SD Upscale”.

Check Auto-Photoshop-SD Wiki for Guides on how to use the Plugin.

Contributors:
@AbdullahAlfaraj, @zombieyang, @deedy5

Support us on Patreon to help us continue developing the Auto-Photoshop-StableDiffusion-Plugin and get early access to new features, exclusive art tutorials and more!

v1.2.4

Updates and New Features Include:

New Features:

• The ControlNet tab now displays an error message if the ControlNet extension is not installed.
• We’ve updated the Lexica image interface with appropriate icons.
• The ControlNet preprocessor thumbnail interface has also been updated.

Bug Fixes:

• Fixed a bug where images wouldn’t generate due to issues with the ControlNet tab.
• Fixed a bug where the ControlNet tab would try to load even when ControlNet was not installed.
• Fixed a bug where ControlNet would only work when the number of ControlNet units was set to 3 or higher.
• Fixed an outdated dependency that caused the image search feature to stop working on 5/12/2023.
• Fixed an issue where native horde wouldn’t work with a random seed.

Support us on Patreon to help us continue developing the Auto-Photoshop-StableDiffusion-Plugin and get early access to new features, exclusive art tutorials and more!

v1.2.3

Updates and new Features include:

• Updated the plugin to use the latest ControlNet endpoint (by @AbdullahAlfaraj and @zombieyang)
• You can preview the ControlNet Preprocessor’s output without generating an image first
• Added support for new ControlNet 1.1 preprocessors (by @AbdullahAlfaraj and @zombieyang)
• Added a “Disable ControlNet Tab” button to allow you to turn off ControlNet without losing your settings
• Added support for Lora. Find all of your Lora models from the Lora dropdown menu
• Added support for Lexica search so you can search for prompt inspiration and download images directly to your canvas
• Added reverse image search for Lexica - find images in Lexica similar to the image you have on the canvas
• Added an image interrogation (txt2prompt) button. Convert any image to a prompt
• Mask expansion slider preview: you can see what your expanded mask looks like as you change the slider value
• Batch count support: You can generate x number of batches subsequently (by @zombieyang)
• Added support for pix2pix model: Toggle Image CFG slider from settings tab
• Fixed an issue caused by the latest Auto1111 update where the image size was not divisible by 64 during inpainting and outpainting.
• Easy access to your generated images folder from the settings tab
• You can enable or disable the “Preview Image” feature from the checkbox next to the progress slider
• Improved the interface for Hi-res fix and added support for scale by ratio
• New “Clear History Results” Button
• The session folder will be locked while the session is active to prevent the user from accidentally moving the output image or selection area.
• Tested with ControlNet version c9c8ca6e

Support us on Patreon to help us continue developing the Auto-Photoshop-StableDiffusion-Plugin and get early access to new features, exclusive art tutorials and more!

v1.2.2

New features include:

• Multi ControlNet - to provide you with more precision control over the generated images.
• New ControlNet presets - enable you to populate Multi ControlNet settings with one click.
• Added Custom Presets tab - allows you to create your own favorite presets/settings.
• New option to speed up the plugin for Slower PC - fined it in the settings tab