Anima negative LoRA
This is my first attempt at creating a negative LoRA designed to improve overall image quality and detail retention.
Recommended Weight:
🟢 -0.1 to -0.4
Usage Notes
🟢 -0.1 to -0.4
Best balance between addition details and pose/shot preservation.
Recommended for most generations.
🟡 -0.5 to -0.6
May begin reducing saturation and stylization.
🔴 -0.8 to -1
Can alter the subject, reduce color intensity, or remove desired visual elements.
Tips
If using stronger weights, consider adding tags such as "flat color", "simple shading", "saturated colors", or similar style prompts to help maintain color consistency.
Description
more images should be better than the first version in more scenarios but might be worse when it comes to adding details
FAQ
Comments (4)
do I put this in the positive or negative prompt window?
It looks like it should go in the prompt but with a negative weight
You should put it in the positive prompt with negative weight something like -0.2, -0.4. or you could put it in the negative prompt with positive weight. It would be something like <lora:badart.safetensors:-1.0>, in the positive or <lora:badart.safetensors:1.0>, in the negative. the lora makes the image look worse if you use it with a positive weight in a positive prompt because it was trained off of bad images
any negative human anatomy next?