JK DoubleSelfie V01 LoRA (double viewpoint selfies)
Added 2023-10-04 05:19:04 +0000 UTC
I definitely don't want to miss every inch of your body. 🔍
⚖️LoRA prompt weight: 0.8~1
⚙️~30% effectiveness
🏫Model used for training: Realistic Vision V5.1
Another wish of mine, it was a challenge to make it viable, but now it's functional.
🔍Recomendations:
- CFG ~7
- Steps >=35
- DPM Family
I recommend start your prompts with:
close-up of a ("nationality":1.2) woman, [breasts and hair characteristics], nude, indoor, [body characteristics]...
LoRA details and tendencies (+resolution):
- I achieved good results with these neg embeddings: CyberRealistic_Negative-neg and FastNegative. But you can use normal neg prompts.
- Don't use viewpoints prompts.
- Faces are difficult to manipulate. I recommend to use adetailer.
Inside the folder I left some example images, you can use this on your prompts.
Comments
This is a configuration file for auto1111. I'm now implementing it to add a description of this model to the list of LoRAs.
Eu sou Chester
2023-10-06 09:52:35 +0000 UTCHey, Thanks for this great looking concept. What do I do with the .json file, should I put it somewhere? TIA, keep up the great work
Jacques D'Anjou
2023-10-06 08:22:27 +0000 UTC
