Why Wan i2v video processor always float32 datatype?

[DamonsJ]

Describe the bug

I found

image = self.video_processor.preprocess(image, height=height, width=width).to(device, dtype=torch.float32)

diffusers/src/diffusers/pipelines/wan/pipeline_wan_i2v.py

Line 633 in 29d2afb

image = self.video_processor.preprocess(image, height=height, width=width).to(device, dtype=torch.float32)

in pipeline_wan_i2v.py

why datatype always float32, maybe it's a bug

Reproduction

just run

Logs

System Info

any platform

Who can help?

No response