If you want to use more my checkpoint online generation, please visit here.

https://tensor.art/u/762555264535746522

Online generation has been added.

Style and Character reference

https://civitai.com/articles/8977/epis-embedding-maps-and-graphs

Fusion style reference for v7 (chinese)(Thanks to Endericedragon606105 for the provision.)

网友提供的v7融合画风参考( 感谢Endericedragon606105提供)

300Styles-waiNSFW.html · sieecc/SOKI at main

v8

• Optimized composition and background logic.

• Tried to fix some minor issues.

All example images are generated at 1024x1360,and Hires upscale: 1.5, Hires steps: 20, Hires upscaler: R-ESRGAN 4x+ Anime6B,Denoising strength: 0.5.

Recommended settings:

Steps: 30

CFG scale: 5-7

Sampler: Euler a

Not very necessary to use ADetailer correction

Positive Prompt

masterpiece,best quality,amazing quality,

Negative Prompt

Base:bad quality,worst quality,worst detail,sketch,censor,

v7

• Adjusted the color balance,

• Reduced the probability of overexposure,

• Added more composition elements, incorporated some special effects,

• More NSFW.

Some of the pony LoRA can be used, but please pay attention to the format when using them.

Please add a comma at the end.

Such as :

<lora:xxxxx:1>,

Recommended settings:

Steps: 30

CFG scale: 5-7

Sampler: Euler a

Not very necessary to use ADetailer correction

Positive Prompt

masterpiece,best quality,amazing quality,

Negative Prompt

Base:bad quality,worst quality,worst detail,sketch,censor,
Optional:simple background,transparent background

V6

Positive Prompt and Negative Prompt is Updated, please check again.

More NSFW , overall balance adjustment.

Recommended settings:

Steps: 30

CFG scale: 5-7

Sampler: Euler a

Not very necessary to use ADetailer correction

Positive Prompt

masterpiece,best quality,amazing quality,

Negative Prompt

Base:bad quality,worst quality,worst detail,sketch,censor,
Optional:simple background,transparent background

V5

I further adjusted the weights to better support the SDXL and Pony LoRA, optimizing some of the composition logic and backgrounds.

Recommended settings same with v2

V4

Increased more training materials and adjusted the default weights.

Some of the pony LoRA can also be used, and you may need to adjust the weight to over 1 for testing.

Recommended settings same with v2

V3

adjusted the weights and changed the default style.

Recommended settings same with v2

V2

I adjusted the default style weight to make the model easier to use. I optimized the composition and human anatomy, and added more NSFW materials.

The example images are all generated directly at 1360x1024. NO AD NO hires fix.

Please try experimenting with artists and characters on your own. I have only tested a portion of popular artists and characters that can be used directly.

Recommended settings:

Steps: 30

CFG scale: 5-7

Sampler: Euler a

Not very necessary to use ADetailer correction

Positive Prompt

masterpiece,best quality,amazing quality,very aesthetic,absurdres,newest,

Negative Prompt

lowres,(bad quality,worst quality:1.2),bad anatomy,sketch,jpeg artifacts,ugly, poorly drawn,(censor),blurry,watermark,simple background,transparent background

v1

Base: IllustriousXL

Since the adjustments were minimal, the artists and characters available in the base model can still be used directly.

Recommended settings:

Steps: 30

CFG scale: 7

Sampler: Euler a

Not very necessary to use ADetailer correction

Positive Prompt

masterpiece, best quality, high resolution, absurdres,uncensored,

Negative Prompt

worst quality,bad quality,bad hands,very displeasing,extra digit,fewer digits,jpeg artifacts,