invoke-ai / invoke-ai/InvokeAI
[enhancement]: UI Tagging For User Objects
- Dominant language
- Python
- Stars
- 28.2k
- Forks
- 3k
- Avg merge
- 6d 5h
- Merged PRs (30d)
- 19
Description
### Is there an existing issue for this?
- [X] I have searched the existing issues
### Contact Details
admin@hunterchasens.com
### What should this feature add?
Right now if you have a large library of models, LORAs, and more, there's no way to filter them. It's just a long list of models and LORAs, and other objects that are too long to read easily. In my opinion, it's bad usability. It'd be great if you could tag them somehow. This could be done via user name tags or perhaps just a way to color code them similar to how you can color code files on macOSX systems. This would be great because, when hit the dropdown menu, it'll be easier to find the model you're looking for or models like it. It'd also be great if you could do the same with generated images. Having the ability to tag, and then search (perhaps they might inherit the model tags) would help workflow.
An example use case I have several sets of models(checkpoints). Some are for photo-realistic generation, some are for semi-realistic generation (such as render/cgi systyles), some are for anime/manga styles, lastly I have some for painting.
For each of these categories I have a few LORA which help. Some LORA work on multiple styles, but some don't transition well (e.g. anime LORA for photo-realistic generation). If I have 50 LORA, but only 3-4 are relevant to the style I'm working I shouldn't have to look through all fifty and try to remember the name of the ones that work. In this use case lets color the anime checkpoint blue and all related LORAS (which can hopefully have multiple tags) also blue. This way I can just scroll through the LORA list looking for the blue models which I know will work well with my style.
### Alternatives
_No response_
### Additional Content
### For reference here's MacOSX's color tag system.
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.