Style Strength Controller

Can you control the style strength in SDXL. Even if you are a user and cannot control the training process? Now you can. And it only needs a LoRA.

NOTE: Forget about v0.1. It is alpha version.

Use v1. It's much better and stabler. Supports all tools. All info in this page has been updated for v1.

What can this LoRA do?

Short answer: This LoRA can reduce (strength > 0) or amplify (strength < 0) the style of your base model. Notice the direction is reversed!

Tech answer: This LoRA can pull your base model weight back toward original SDXL.

So you can

• reduce the style strength a little bit to avoid overfitting effects and bring creativity back. Very useful when your base model is heavily overfitted first, or you stacked tons of LoRAs.

• amplify the style. (just want something crazy)

This lora will not bring any side effects to the style, mathematically.

(Maybe this is the mathematically true stabilizer you guys are asking for?)

How to use?

• Just apply it as normal LoRA.

• Working strength is around -0.5~1. Recommended -0.2 and 0.2.

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

• Some styles affect CFG scale. So you may also need to adjust the CFG scales because the style strength changed.

What's the training data?

This LoRA is "calculated/calibrated" from SDXL and Illustrious v0.1. No training process.

Supported models?

This LoRA only supports models that are close to the base model it was calibrated for.

Currently there is only one version for Illustrious v0.1. I'm still testing this idea. Other versions, like NoobAI etc. will come later.

Sounds like some kind of black magic.

Yes, it is, math. And it always works.

Remember to leave feedback in the comment section.

Have fun.

Old info for v0.1:

• Only support ComfyUI and new tools. A1111 and online generator is not supported.

• Strength: ~0.5~0.5. >0 will amplify the style, <0 will reduce it.