stable-diffusion-webui
stable-diffusion-webui copied to clipboard
AUTOMATIC1111
[Feature Request]: Save full Settings of Text2IMG or IMG2IMG page (Input Image aswell)
Is there an existing issue for this?
- [X] I have searched the existing issues and checked the recent builds/commits
What would your feature do ?
Save full Settings of Text2IMG or IMG2IMG page (Input Image aswell): so one can load them in like a Preset/save them if one finds good settings. Especially with the merging of Control Net-it would help alot to not get confused which Init Control Net image was used. At least the Image File NAME could be saved-but loading in the Preset file and having the init images load in too would be amazing for workflow!
Best wishes
Proposed workflow
Similar like the Prompt Save Button-that it would be saved in an folder directory one choose. Then there would be a button like ''load in preset'' and one can choose from these files :)
Additional information
No response
In addition, or even separately, the web-ui should save the "Current State" of a panel when Generate is pressed. This state should be recalled, and pre-filled for the user, on startup.
This would include:
- Positive prompt
- Negative prompt
- The state of all checkboxes, sliders, and comboboxes
The theory being that not only do many of us use the same settings from session to session and that our previous settings are more likely to be correct than the default settings, but also that starting the web-ui with two empty prompt boxes is guaranteed to be wrong 100% of the time. :)
I know this would improve my "startup time" significantly.
yes! i see myself often coming back to generations/styles months ago. it would help alot to save everytime full generation settings
The settings are already saved in output images. Use the PNG Info tab, or drag an image to the prompt and click ↙️.
@Alchete Clicking ↙️ with the prompt blank does exactly that,
Potentially worth noting, the class used for image processing & presentation does have fields for scripts and script_args ... they're just not used, at least by the XYZ Grid script I've been dealing with.
In txt2img, everything outside of script is saved in images, ready to be retrieved with PNG Info.
Beyond even that, what you're looking for is the ui-config.json file. It already has fields for you to put your desired default prompts, along with I think every other UI item and state in the whole darn program.
It would be nice if this was separate from ui-config.json. In my experience, crashes can be common. If there was a ".tmp.ui.json" or similar that if sdw needed to be restarted for... reasons... the next restart would check for that file and reload the settings (including images in img2img and controlnet). Perhaps a flag could be introduced "--ignore-tmp-ui-config" or something.
Now with Tiled Diffusion, Multi Control Net, etc. if things are crashing (usually because some parameter conflict) having to bring in add those fields manually can take 5+ minutes.
I would really appreciate having such a feature. Saving successful experiments with various Controlnet Units by taking screenshots and storing the prompts in a TXT file is quite cumbersome, and there's a risk of overlooking something.
