WAN2.2 is unrivaled in generating "Oriental Classical" style character portraits to date.

It also boasts exceptional support for Chinese prompt words, and still outperforms the current Z IMAGE and QWEN IMAGE by a landslide.

Generating images with the high and low noise dual models still holds distinct advantages over using a pure low noise model alone. The combination of high and low noise in image generation helps enrich the fine details in the frame.

This workflow adopts the AdvancedSampler for sampling, featuring fewer parameters for easy adjustment. Its default parameters strike a sensible balance between image expressiveness and visual diversity.

Users have reported issues with encrypted node groups, including downloading and unpacking problems, so I have not used them. This workflow has already been unpacked.

Test Demo Video:

https://www.bilibili.com/video/BV1ns6FBqEQn?vd_source=a5db6342ae91bf786f8799336a9e794a

WAN2.2 is unrivaled in generating character portraits in the Oriental Classical style to date.

It also boasts exceptional support for Chinese prompt words, and still outperforms the current Z IMAGE and QWEN IMAGE by a landslide.

Generating images with the combined high and low noise dual models still holds distinct advantages over using a pure low noise model alone. The combination of high and low noise in image generation helps enrich the fine details in the frame.

This workflow adopts the Advanced Sampler for sampling, featuring fewer parameters for easy adjustment. Its default parameters strike a sensible balance between image expressiveness and visual diversity.

Test Demo Video：

https://www.bilibili.com/video/BV1ns6FBqEQn?vd_source=a5db6342ae91bf786f8799336a9e794a