Colorful_1

Translated from Google Translate：

This is used to emulate a Japanese anime style of drawing that can make characters look softer.

Although this Lora is trained on the pony, its performance on the [Pony Diffusion V6 XL] is not satisfactory.

I recommend you use [WAI-NSFW-illustrious-SDXL], although the basic model of [WAI-NSFW-illustrious-SDXL] is not Pony.

Maybe I'll try to fix this later, but I don't have enough buzz right now.

================

Be sure to use Trigger Words: colorful_V1

===========================

It has poor background generation capabilities,so using "white background" in Prompt might be a good choice.

I haven't trained on Negative Prompt, so please fill in the Negative Prompt on your own.

If you are using this lora on a local PC:

Sampling method: DPM++3M SDE

Schedule type: Karras

Sampling steps: 50

If you are using this lora on CivitAi:

DDIM and DPM++ 2M Karras are OK (DDIM maybe better).Euler a occasionally has some flaws.

Sampling steps:38(This is the result of combining BUZZ cost-effectiveness. If you are relatively wealthy, you can add even more.）

CFG Scale: 7.5