yinzo/sd-webui-lora-queue-helper

A Script that help you generate a batch of Lora, using same prompt & settings. Helpful for compare Lora, Or just switching Lora easily without switch tab.

sd-webui-lora-queue-helper's Issues

Problem : Grid do not save (+feature request)

I love this script it's really something I needed, Auto row number is a simple but very nice feature too. BUT...

• the grid is not saving. I can find it in appdata/local/temp/gradio then I copy that somewhre manually but it's not cool
• it would be nice to have the lora names in the grid, because if a lot are tested it quickly become difficult to tell what lora was used for each image
  I'm using webUI Forge :

Python 3.10.6 (tags/v3.10.6:9c7b4bd, Aug  1 2022, 21:53:49) [MSC v.1932 64 bit (AMD64)]
Version: f0.0.17v1.8.0rc-latest-276-g29be1da7
Commit hash: 29be1da7cf2b5dccfc70fbdd33eb35c56a31ffb7

Thank you

Feature Request: Change Default Weight via UI

"Use built-in Lora configuration to store Activation Text & Preferred Weight, which will be automatically used in this script. Otherwise, it will only apply the Lora with default 1 weight."

Can you add a slider/input box to let the user easily select a different default weight? I have 100+ LoRAs I haven't set the preferred weight for and most of them are cooked at weight = 1. I'd rather test them around 0.5, but I don't want to go configure that as the default for all of them in their configs. It's also beneficial to be able to compare LoRA's quickly across different weights.

Lora-Queue-Helper and Lora-Block-Weights

I tested out your script and it worked very well, but the Lora weights turned out to be either 1 or 0. I think because I accidentally had "Lora-Block-Weights" enabled.

Examples taken from the Stable Diffusion output text file:

"<lora:Celeb-Zh_Sjenkins_V1:0>", ZH_SJenkins,,(masterpiece, best quality, highres, high resolution:1.2), extremely detailed, realistic, intricate details

"<lora:Celeb-Melonie_Mac:1>", mlnemc,,(masterpiece, best quality, highres, high resolution:1.2), extremely detailed, realistic, intricate `details

So it might not be a problem. I'm running it again without Lora-Block-Weights enabled to see what happens.

Feature Request: Nested subfolders support

Great script but doesn't show nested subfolders. eg fictional/dccomics/batman.

Thanks for your efforts though and I can understand why nesting would be a pain programatically.

TypeError after update.

Using Automatic1111 version: 1.6.1 and python: 3.10.6.

I opened the Extensions and checked for updates. Lora Queue Helper was the only update. I applied and restarted.
When I tried to use the script I get:

TypeError: Script.run() missing 9 required positional arguments: 'is_use_custom_path', 'custom_path', 'directories', 'selected_loras', 'checkbox_iterate', 'checkbox_iterate_batch', 'is_save_grid', 'is_auto_row_number', and 'row_number'

I used the Backup/Restore to roll back the change to commit be67ef53

And now get a slightly different error, but the same issue:

TypeError: Script.run() missing 7 required positional arguments: 'is_use_custom_path', 'custom_path', 'directories', 'selected_loras', 'checkbox_iterate', 'checkbox_iterate_batch', and 'is_save_grid'

I tried to manually remove the folder and update manually, but get the same issue regardless.

Feature Request: Textual inversion support

Would be nice if you could to textual inversions please.

Enhancement? The ability to use multiple Loras for even more dynamic results?

I'm not sure how hard it would be to implement, but the idea of being able to run multiple Loras from the list at a time would be pretty cool. I am not a programmer and I am new to AI, so I am unsure how the process works in the software. I've used multiple Loras in my prompts and it can create some pretty cool styles and effects.

Thanks!

Feature Request: support glob and exclude