majicMIX realistic

The fourth edition is re-integrated, and the third edition has a color problem that I personally think is a failure, but some friends use it well, so I keep it.

In the third version I continued to incorporate improved body models, as well as enhanced lighting and shadows.

I'm sorry that I used layered lora in the previous version of the example diagram to confuse everyone, and it also made it difficult for everyone to reproduce my example diagram. So I didn't use any lora for the new version of the illustration. If you want to know about lora layering, you can refer to: GitHub - hako-mikan/sd-webui-lora-block-weight

My layered parameters My lora block setting:

Body

BODY:1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1 BODY0.5:1,1,1,1,1,1 ,0.2,1,0.2,0,0,0.8,1,1,1,1,1

Face (face shape, hairstyle, eye shape, pupil color, etc.)

FACE:1,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0 FACE0.5:1,0,0,0,0,0 ,0,0,0.8,1,1,0.2,0,0,0,0,0 FACE0.2:1,0,0,0,0,0,0,0,0.2,0.6,0.8,0.2, 0,0,0,0,0

For repairing

HAND: 1,0,1,1,0.2,0,0,0,0,0,0,0,0,0,0,0,0

Clothing (use with tag)

CLOTHING: 1,1,1,1,1,0,0.2,0,0.8,1,1,0.2,0,0,0,0,0

Action (use with tag)

POSE:1,0,0,0,0,0,0.2,1,1,1,0,0,0,0,0,0,0

Coloring style (use with tag)

PALETTE: 1,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

character (de-stylized)

KEEPCHAR: 1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0

background (de-stylized)

KEEPBG:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,0,0

Reduce overfitting (equivalent to OUTALL)

REDUCEFIT: 1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1

In the second version, I improved the weight of certain components, and also improved the light and shadow

The new display picture uses film style In show case I used film style girl LoRA

A variety of models are integrated, which can generate beautiful faces and effectively deal with dark parts. Distant faces require inpaint for best results.

Listen to my advice, don't open facial repair! Do not prescribe facial restoration!

Recommended keywords recommended positive prompts: Best quality, masterpiece, ultra high res, (photorealistic:1.4), 1girl

If you want darker picture, add: in the dark, deep shadow, low key , etc.

Negative keywords use _ng_deepnegative_v1 75t and badhandv4 in negative prompt

I've used a bug-fixed version of DPM++ 2M Karras, you can check this out: https://civitai.com/models/35966/dpm-2m-alt-karras-sampler

Recommended Parameters Recommended Parameters:

Sampler: DPM++ 2M Karras (bug-fixed) or DPM++ SDE Karras

Steps: 20~40

Hires upscaler: R-ESRGAN 4x+ or 4x-UltraSharp

Hires upscale: 2

Hire steps: 15

Denoising strength: 0.2~0.5