Comfy-Org/ComfyUI · error · RuntimeError

Only {len(urls)} of {len(response.data)} images were generat

Error message

Only {len(urls)} of {len(response.data)} images were generated before error.

What it means

Raised by ByteDanceSeedreamNode (Seedream 4.5/5.0 unified image node) when a sequential/batch generation request finished but only some of the requested images succeeded. The API returns a list of result items in response.data and some items lack a 'url' key, meaning the model stopped or errored partway. With fail_on_partial=True (the default for this node) the node refuses to return a partial batch.

Source

Thrown at comfy_api_nodes/nodes_bytedance.py:760

            ApiEndpoint(path=BYTEPLUS_IMAGE_ENDPOINT, method="POST"),
            response_model=ImageTaskCreationResponse,
            data=Seedream4TaskCreationRequest(
                model=model,
                prompt=prompt,
                image=reference_images_urls,
                size=f"{w}x{h}",
                seed=seed,
                sequential_image_generation=sequential_image_generation,
                sequential_image_generation_options=Seedream4Options(max_images=max_images),
                watermark=watermark,
                output_format="png" if model == "seedream-5-0-260128" else None,
            ),
        )
        if len(response.data) == 1:
            return IO.NodeOutput(await download_url_to_image_tensor(get_image_url_from_response(response)))
        urls = [str(d["url"]) for d in response.data if isinstance(d, dict) and "url" in d]
        if fail_on_partial and len(urls) < len(response.data):
            raise RuntimeError(f"Only {len(urls)} of {len(response.data)} images were generated before error.")
        return IO.NodeOutput(torch.cat([await download_url_to_image_tensor(i) for i in urls]))


def _seedream_model_inputs(
    *,
    max_ref_images: int,
    presets: list,
    max_width: int = 6240,
    max_height: int = 4992,
    supports_batch: bool = True,
):
    inputs = [
        IO.Combo.Input(
            "size_preset",
            options=[label for label, _, _ in presets],
            tooltip="Pick a recommended size. Select Custom to use the width and height below.",
        ),
        IO.Int.Input(

View on GitHub (pinned to 1c6d8d45b3)

Solutions

Set fail_on_partial=False if partial results are acceptable; the node then downloads whatever URLs came back instead of raising.
Reduce max_images (e.g. to 2-3) so the sequential run is less likely to fail partway.
Simplify the prompt to avoid content that causes individual images in the batch to be dropped.
Retry with a different seed; a partial failure is often specific to the sampled sequence.

Example fix

// before (workflow / node params)
sequential_image_generation = "auto"
max_images = 6
fail_on_partial = True  // raises when only 4 of 6 URLs come back

// after
sequential_image_generation = "auto"
max_images = 3
fail_on_partial = False // returns the images that succeeded

Defensive patterns

Strategy: fallback

Try / catch

# partial batch results are expected behavior, not a crash
try:
    out = await ByteDanceSeedreamNode.execute(...)  # or run node in workflow
except RuntimeError as e:
    if 'images were generated before error' in str(e):
        logger.warning('partial Seedream batch: %s', e)
        out = await rerun_with(fail_on_partial=False)
    else:
        raise

Prevention

Set fail_on_partial=False unless downstream code requires an exact image count.
Keep max_images small (2-3) for long sequential runs.
Wrap batch outputs with a length check before indexing a fixed count.

When it happens

Trigger: sequential_image_generation='auto' with max_images>1 on Seedream 4.5/5.0; the BytePlus API completes the task but returns fewer URL-bearing entries than response.data entries (e.g. content-filtered or failed generations), and fail_on_partial is left at its default True.

Common situations: Batch story-scene generation where one image in the sequence trips moderation; long multi-image runs that time out server-side mid-batch; users who need all N images downstream (e.g. torch.cat of a fixed count) and therefore keep fail_on_partial on.

Related errors

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/7890c693e328f574. Report an issue: GitHub.