HuanCai||Color Fantasy (Auxiliary Fine-Tuning Model)

Welcome to useColorful Lora ModelThis modelsupports art stylemicro-adjustmentand is compatible

  2D, 2.5D, realistic models; can be used independently as character Lora

-------------------------------------------------------------------------------------------------------------------------------

Recommendations for Use:

Recommended to use in combination with anime or realistic models compatible with 2.5D models

When combined with other Lora suggestions, weight of this model should be 0.8-1; other Lora weight 0.6-0.8

If using the ADetailer plugin, make sure to enable it when exporting images

Model adaptation to base model conditions and sampler compatibility chart

As shown in the chart, it is not recommended to use the DPM++ 2M SDE sampler

The cover image is based on MiaoKa_reality【Fusion】drawn

-------------------------------------------------------------------------------------------------------------------------------

Recommended starting prompts: best quality, masterpiece, realistic, HDR, UHD, 8K, masterpiece, high quality, exquisite wallpaper, 16K, animation, illustration, (white_skin),

Suggested image parameters

Sampling method (Sampler):

Euler a: 25-30 steps

DPM++ SDE Karras: 20-30 steps

DPM++ 2M Karras: 20-40 steps

Prompt word guide factor (CFG Scale): 7

Character resolution: 512x768, 768x768, 512x1024

Background and landscape resolution: 768x512, 1024x512, 1536x512

High-definition repair: R-ESRGAN 4x, high-end iteration steps 15-30, redraw range 0.3-0.5

Clip Skip: 2

Negative prompt words:

Recommended to use EasyNegative in the embedded embeddings model, negative_hand-neg

Two sets of additional negative prompt words:

Set 1: (worst quality, low quality: 2), monochrome, zombie, (interlocked fingers: 1.2),

Set 2: NSFW, logo, text, blurry, low quality, bad anatomy, lowres, worst quality, watermark, bad proportions, username, EasyNegative,

-------------------------------------------------------------------------------------------------------------------------------

Model feedback AI technology exchange QQ group: 377209248

This model does not include any real-life image sources or training code from the internet. The model is only for research and interest exchange