Please Read before test!
This is a testing lora, some issue is possible, follow instruction below on how to use.
How to use:
add it at the end of your prompt.
Suggested value:
Realistic model: from 0.2 to 0.5
Anime and toon: from 0.5 to 1
What it can do on realistic model:
After many test, it seem giving better result and fix some issue like double subject, many harms, fused parts.
What it can do on anime/toon model:
same as realistic plus add clear border line on character, emphatize image.
This is really alpha stage, and having more test on different checkpoint help me to make it better.
All sample image are made with UniPc sampler, 10 step, cfg 5.6 at 1152x896 in different checkpoint i have.
Attached image without and with this lora for comparison.
Description
v1.0 alpha
FAQ
Comments (5)
I don't know what kind of witchcraft/voodoo you are doing with this, but it is incredible. EVERY DAY, I fight with token weight to cram as much detail and color into my Faerie images as possible. I would say, without a doubt, that your work & this LORA are some of the most game-changing I've seen in the past six months. If you need any assistance in just running mass quantities of prompts through the various checkpoints and loras to see how this interacts, let me know and I'll dedicate some time to it and post the results here or anywhere you specify. You are doing incredible work. Keep at it!
I say just run with it anyway so others can compare. I've also very interested how this pans out.
thank you very much for taking time and test this LoRa. i saw your review an i like test you have made. I'm doing same test with different checkpoint, but i have only my selfmade model, and see how it work in other checkpoint help me to doing better work in a future update, so if you want spending time and test it in any way you think is truly appreciated. I'm also doing some test with other LoRa and embedding, to see where mix split out and where need to be fixed.
@devilkkw I'll spend some time running it through the more popular checkpoints and use the same prompt on each with various LORAs in combination and run X,Y grids so we can all see which benefits the most and if there are any issues. I'll post results in the gallery for your LORA.
@Inland389 thank you so much.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.