- Added Evolve which generates variations on your generated picture.
- Fixed wild imports
- Added instant One Button Prompt
- Added keybind of
ctrl+shift
to toggle `hurt me plenty`` mode - Code Cleanup
- Fixed Issue with dropdown box's being case sensitive when sorting
- Adds token count (thanks Lavorther)
- Fixed lora keywords only working the first time the lora is loaded
- Use OneButtonPrompt with infinite mode, simply put a tick in
BYPASS SAFETY PROTOCOLS
- Inpainting Update: Adds inpainting to the powerup tab
- Now comes with a built in clip interrogator, just drag your image onto the main image to generate the prompt
- Π Time - Click the small Π symbol to get a fullscreen image of your last generation great for slideshows
- updated comfy version, NEW SAMPLER TIME
- Added wildcard helper, just start typing __ and your wildcards will display
- Added slideshow
- Fixed old bug of switching models when the stop button is pressed (old code from OG-F)
- Old experimental lcm-pipeline removed
- Generate forever if Image Number is set to 0
- Updated comfy version to latest
- Nested wildcards now supported
- Random styles now correctly applying to each image
- Gradio Rollback to v3 until v4 is fixed
- WildCard Fixes
- Automatcially downloads LCM Models
- Now checks subdirectories for models
- Gradio update to V4
- New Random Style Selection
- Adding one button prompt overrides in wildcards now
- Added wildcards official
- Other stuff i'll have to get arljen to explain
- Limit seeds to 32 bits
- Sort model dropdowns
- Use caches better
- removed groop and faceswap as it was causing dependency issues on some systems
- Changed minimum cfg scale to 0
- Updated to latest comfy and diffusers (Now supports LCM Loras)
- You NEEED to set the custom settings to use lcm and sgm_sampler, steps of 4 and REALLY low config (between 0 and 1.5)
- Facewapping
- Groop
- Updated Comfy Version
- Different pipelines supporting lcm and sdxl/ssd
- Let async_worker handle model preload
- Lots of small fixes
- fixed metadata bug when stopped
- Fixed small issue with metadata not updating
- Added Metadata Viewer for Gallery items (Viewable in
Info
Tab) - Refresh Files now also reloads your
styles.csv
file
- Automatically download 4xUltrasharp Upscaler
- Added the ability to upscale images wth upscaler of your choosing
- Changed Powerup Settings so if there is a missing key from defaults it adds it to your custome settings.
- Refactored backend code to allow for future pipeline changes
- Automatically read triggerwords from <lora_filename>.txt
- Updated Comfy Version
- Added support for SSD-1B Models
- Removed ref redundant code.
- Update Comfy version and fix changes :D
- Improved image2image and allowed settings to be changed when "custom" is selected form the PowerUp Tab.
- Added the basics for image 2 image
- Renamed Controlnet to PowerUp
- Now uses
powerup.json
as default
- Wildcards can now use subdirectories
- Fixed issue where if you placed 2 placeholders with the same name, you got the same results, a new one is now chosen
- Updated status to show model loading / vae decoding
- Update to one button prompt (provided by Alrjen)
- Custom Controlnet Modes
- minor bugfixes
- moved the controlnet tab to its own ui file.
- Added sketch controlnet!
- Updated gradio version
- Added recolour controlnet!
- Restored gallery preview on all images
- renamed more variables to make sense
- bugfixes
- Added all the settings/customization to their own
settings
folder NOTE: you will need to move / copy your settings into the new directory - Bugfix where clicking stop mid-generation stopped working
- code cleanup and some renaming
- inference model speed up
- now only shows gallery when needed
- removed metadata toggle, it will now always save metadata
- save your own custom performances
- tidied ui
- fix crash when failing to load loras
- hide all but one lora dropdown showing "None"
- change fooocusversion.py to version.py for easier updating
- Moved controlnet to its own tab for easier updates
- updated gradio version
- minor wording changes
paths.json
will now be updated if there are any missing defaults paths
- Now supports controlnet
- Updated onebutton prompt so you can now add multiple random prompts by clicking the
Add To Prompt
button
- Update comfy version - Lora weights are now calculated on the gpu so should apply faster
- Bug fixes and backend updates
- changed
resolutions.csv
toresolutions.json
- updated readme
- Prompt now splits correctly using
---
- added the ability to change styles in the prompt by using style:stylename
- Added init image
- Fixed issue with wildcards if file not found.
- Fixed sorting on subfolders, so directories are displayed first
- Allowed main image window to recieve drag and drop
- Added a gallery preview underneath that will activate image window.
- Added support for subdirectories with models/loras so you can get all organised!
- showed imported image in gallery
- moved
---
split into prompt generation - correctly updates progressbar
- fixed importing of width / height
- In the json prompt, setting a seed of
-1
allows you to generate a random seed
- Render different subjects so you can process a whole list of prompts. Seperate each prompt by placing
---
on a new line
- New Beginnings. The official start of the updates!