InvokeAI
InvokeAI copied to clipboard
invoke-ai
[enhancement]: Grouping LoRAs by category
Is there an existing issue for this?
- [X] I have searched the existing issues
Contact Details
What should this feature add?
When you have a large number of LoRA, it becomes hard to find the one you want from the list. It would be nice to be able to group them by categories. A good way to do this would be to put them in subfolders with the name of the category like this:
Lora
├── Characters
│ └── hatsune_miku.safetensors
├── Poses
│ └── cool_pose.safetensors
└── Styles
├── greyscale.safetensors
└── some_other_style.safetensors
Then, in the WebUI, they would show grouped by category.
Alternatives
Renaming every LoRA to be prefixed with a category name. That makes it harder to keep track of each LoRA though, and doesn't look good in the WebUI.
Aditional Content
No response
Is this encoded in the lora or listed on hugging face in some way? Otherwise it would require a manual list to be maintained or something which is not feasible to do automatically. Maybe an optional setting the user could manually set up, but that might get too clunky.
Personally using the alternative method the OP suggested here and it's working just fine for me. I agree with @Void2258 that you'd need an external source to obtain this data, and it might not match with your expectations and could require additional translation work etc.
My categories are fairly self-explanatory, such as:
- "Char - x"
- "Pose - x"
- "Mod - x" (used for styles or sliders, enhancers etc)
- "Prop - x" (used for clothing or items)
... and this keeps them nicely grouped in the UI.
I also tend to put the URL of the page for the model into the notes field, so I can keep track of where it's actually from etc... a separate field for this would be a nice enhancement though.
Naming the LoRAs in a way that lets you filter for them is indeed the best way to do this.
