[Feature request] Implement Style Aligned Image Generation via Shared Attention published by google recently. ( https://github.com/google/style-aligned)

[lanyusan]

Please take a look at https://github.com/google/style-aligned .

This is a great tool for style consistency.

The demo code supports sdxl and sdxl controlnet with diffusers.

Please consider implementing it in Fooocus.

[mashb1t]

relates to https://github.com/lllyasviel/Fooocus/pull/1503 for prompt input of various subjects with same style