This was an experiment but came out it can add a very nice details to the images also make it more vivid.

The sample grids were created with different checkpoints, some base models are adjusting more easily but in general it makes a nice output.

I would say use the weight low. 0.2-0.3 should be the sweet spot but you can experiment. If you adjust higher the colors will be more vivid but the image might brake. Maybe not... who knows you need to try it :)

Feel free to comment and I would love to see the created images. I hope all of you will have fun with this lora.

Recommended settings (this is what I used with sample images):
Steps: 30

Sampler: Euler a

Schedule Type: Karras

Cfg: 5

Clip Skip: 2

Hires Upscaler: R-ESRGAN 4x+ Anime6B

No need to add any words to activate the lora.

Thank you if you use or download this model, I hope you will enjoy.