Contrast Controller

Now you can control contrast in SDXL just like using a slider on your monitor.

Most importantly, the effect of this LoRA is stable, linear, and has ZERO side effect (not an exaggeration, it's really mathematically linear and zero).

What does this LoRA do?

It can amplify (strength >0) or reduce (strength <0) the contrast of your base model.

Advance version:

It can control the "strength" of the U-net output and make it "smoother". It can do more than just adjusting contrast.

Sounds like FreeU, RescaleCFG and DynamicCFG-Fix etc., but much more simple and easy to use.

So you no longer have to choose between "creativity/low CFG scale" and "high contrast". Now you can have both.

You can increase base model contrast so you can lower CFG scale, and get more creative result and details.

Or vice versa, reduce base model contrast so you can use higher CFG to get a stable and clean result, without oversaturation.

Most importantly, this LoRA has ZERO side effect on style.

WTF is this 400KiB LoRA file. Is the file broken?

No. It is a fully functional LoRA. And probably the smallest LoRA. I even tried a 30KiB version, and did not feel any difference. But my inner heart told me not to go too crazy.

How to use it?

• Just apply it as normal LoRA.

• Recommended strength is -0.5~0.5.

• You don't have to set the patch strength for text encoder. This LoRA does not patch it.

• Also recommend adjusting your CFG scale.

Supported models?

All Illustrious based model.

What's the training data?

This LoRA does not come from training.

Share merges using this LoRA is allowed. However, you must credit the creator and provide a link to this page. It is easy for me to know whether a model merged my LoRA, because the weight pattern will become very unique after this LoRA applied. A normally trained model will never have such kind of pattern.

Feedback in comment section is welcome.

Have fun.