lllyasviel/Fooocus · error · ValueError

Cannot process this value as an Image

Error message

Cannot process this value as an Image

What it means

get_config()/serialize of the vendored gradio Image converts an example value y into base64 for the frontend; it accepts np.ndarray, PIL.Image, and str/Path (file or URL) and raises ValueError('Cannot process this value as an Image') for anything else. This typically fires when preparing example outputs for display, not during prediction.

Solutions

  1. Convert tensors: img = tensor.mul(255).clamp(0,255).byte().cpu().numpy() (HWC) or wrap with PIL.Image.fromarray before returning/exemplifying.
  2. Give examples as filesystem paths or URLs (str/Path), not raw bytes.
  3. Ensure list-of-images goes to gr.Gallery, not Image.

Example fix

# before
return torch_tensor  # into gr.Image output
# after
return Image.fromarray(torch_tensor.mul(255).clamp(0,255).byte().cpu().numpy())
Defensive patterns

Strategy: type-guard

Validate before calling

import numpy as np
from PIL import Image
from pathlib import Path
def to_image_value(v):
    if isinstance(v, Image.Image) or isinstance(v, (str, Path)) or isinstance(v, np.ndarray):
        return v
    if hasattr(v, 'cpu'):  # torch tensor
        return Image.fromarray(v.mul(255).clamp(0, 255).byte().cpu().numpy())
    raise TypeError(f'not a serializable image value: {type(v)}')

Type guard

def is_image_value(v) -> bool:
    import numpy as np
    from PIL import Image as PILImage
    from pathlib import Path
    return isinstance(v, (np.ndarray, PILImage.Image, str, Path))

Prevention

When it happens

Trigger: Passing examples=[...] or returning values containing bytes, a torch.Tensor, a list of images, or None-with-wrong-wrapper to the component's serialization path (e.g. examples=[[42]] or examples=[[tensor]]).

Common situations: Returning a torch.Tensor from a prediction function that feeds an Image output; examples referencing file-like objects instead of paths; numpy scalars or float arrays of unexpected shape.

Related errors

AI-assisted analysis of lllyasviel/Fooocus@ae05379cc9 (2026-08-15). Data as JSON: /api/errors/78ed0beaa266e37f. Report an issue: GitHub.

Appendix: source

Thrown at modules/gradio_hijack.py:332

    def postprocess(
        self, y: np.ndarray | _Image.Image | str | Path | None
    ) -> str | None:
        """
        Parameters:
            y: image as a numpy array, PIL Image, string/Path filepath, or string URL
        Returns:
            base64 url data
        """
        if y is None:
            return None
        if isinstance(y, np.ndarray):
            return processing_utils.encode_array_to_base64(y)
        elif isinstance(y, _Image.Image):
            return processing_utils.encode_pil_to_base64(y)
        elif isinstance(y, (str, Path)):
            return client_utils.encode_url_or_file_to_base64(y)
        else:
            raise ValueError("Cannot process this value as an Image")

    def set_interpret_parameters(self, segments: int = 16):
        """
        Calculates interpretation score of image subsections by splitting the image into subsections, then using a "leave one out" method to calculate the score of each subsection by whiting out the subsection and measuring the delta of the output value.
        Parameters:
            segments: Number of interpretation segments to split image into.
        """
        self.interpretation_segments = segments
        return self

    def _segment_by_slic(self, x):
        """
        Helper method that segments an image into superpixels using slic.
        Parameters:
            x: base64 representation of an image
        """
        x = processing_utils.decode_base64_to_image(x)
        if self.shape is not None:

View on GitHub (pinned to ae05379cc9)