Sign In

Elaine Auclair (Kuro no Kiseki)

19
133
3
Verified:
SafeTensor
Type
LoRA
Stats
133
0
Reviews
Published
Mar 28, 2024
Base Model
SDXL 1.0
Training
Steps: 1,540
Epochs: 7
Usage Tips
Strength: 1
Trigger Words
KnKElaineAuclair wearing her Signature Outfit
KnKElaineAuclair wearing her Casual Outfit
KnKElaineAuclair wearing her Bracer Outfit
KnKElaineAuclair wearing her Swimsuit Outfit
Hash
AutoV2
6486C782C8
default creator card background decoration
Z1Gamer's Avatar
Z1Gamer

Hi all! Here is my second LORA!

It includes 4 outfits, check the comparison picture (the last one up top) to see which trigger corresponds to which outfit.

It has a few issues left, and I am currently in the process of making an entirely new dataset, while also planning to use some new settings for what I hope will be better training. I am currently at 68 training runs for this LORA :D. So meanwhile I decided to upload this because even though it's not perfect, I have been able to make nice images with it. I hope you like it :)

Here are the triggers for the different outfits:

KnKElaineAuclair wearing her Signature Outfit
KnKElaineAuclair wearing her Casual Outfit
KnKElaineAuclair wearing her Bracer Outfit
KnKElaineAuclair wearing her Swimsuit Outfit

Some Recommendations:

LORA strength is working fine at 1 most of the time. Now if you are attempting major style changes (like fully realistic) you will probably need to go down to about 0.7. But it's not really made for that :D.

For the Signature and Casual outfits, the "thigh-highs" and "boots" don't always come out depending on the model used, so just put them in the prompt as well if needed. Other that that the outfits mostly work without any extra prompting, just the above basic trigger sentence.

I usually use DPM++ SDE Karras for generation, or Euler Ancestral. I have not tested the others much this time.

Tested and works fine with AAM XL, Animagine XL 3.1, JuggerMagine XL and Ghost XL. Although it will probably work with many models because it's decently flexible.

Regarding generational parameters, I generate images mostly with ComfyUI and it seems CivitAI has trouble getting the parameters, probably because I use some custom node addons. Anyway, if you are also using ComfyUI you should be able to get the workflow by left clicking on one of the images (to open it full size) and then right click on it - save as. This should lead you to save the .png which has the workflow.

Enjoy!

If you liked my work you can buy me a coffee:

Ko-Fi: https://ko-fi.com/z1gamer

or crypto: https://coindrop.to/z1gamer