Skip to content

vllm_omni.diffusion.utils.image_output ¶

ImageOutput `module-attribute` ¶

ImageOutput = OmniRequestOutput | RequestOutput

ImageOutputs `module-attribute` ¶

ImageOutputs = ImageOutput | Sequence[ImageOutput] | None

extract_images_from_outputs ¶

extract_images_from_outputs(
    outputs: ImageOutputs,
) -> list[Image]

Extract PIL images from known Omni output types.

Supported inputs: - OmniRequestOutput / list[OmniRequestOutput] - RequestOutput / list[RequestOutput] (compatibility fallback)