webui-plugin improvment: group generated images by prompt, lora/hypernetworks/embedding support

[jtydhr88]

previously, only create one group for all generated images.
However they should be grouped by prompt:

[jtydhr88]

Regarding extra network, I am using style to place the list, I am not sure it is the right place.

About LoRA, currently this is a bug in webui with api mode, see more detail here: AUTOMATIC1111/stable-diffusion-webui#9274, until today, this issue still does not get fixed, that means, if the user want to apply LoRA, they have to mofidy webui.py by theirselves:
modules.script_callbacks.before_ui_callback()

[harrywang]

Thanks for working on this - looking forward to test this out.

[jtydhr88]

update, a1111 updated webui to latest 1.3.0, which fixed the lora api issue, the list of loras api provided by default now, no need to change source code

[harrywang]

@jtydhr88 how can we use lora? use the <lora:lora_name:0.8> syntax? I don't see a place to see the lora models in the frontend. Thanks!

[jtydhr88]

Hey @harrywang you can go to settings, and choose style type from lora or hypernetwork, then you can go back and choose your Lora from style drop-down, currently I am using the style as Lora holder, I am not sure it is right place, and it only support to select one Lora at once time. If we want support multi-extras, we have to change StableStudio deep code. Right now I hardcode the weight to 1, we may need a slider to allow to change weight to other value. This PR is still pending for review.

[harrywang]

Hey @harrywang you can go to settings, and choose style type from lora or hypernetwork, then you can go back and choose your Lora from style drop-down, currently I am using the style as Lora holder, I am not sure it is right place, and it only support to select one Lora at once time. If we want support multi-extras, we have to change StableStudio deep code. Right now I hardcode the weight to 1, we may need a slider to allow to change weight to other value. This PR is still pending for review.

Thanks a lot! This is a great starting point for using lora - we will need to able to adjust weights and use multiple loras eventully :).