ATH-MaaS/Pixelle-Video · error · Exception

The workflow did not return a video. Please check the workfl

Error message

The workflow did not return a video. Please check the workflow configuration.

What it means

After running the image-to-video workflow, the code looks through node outputs for a 'videos' list and picks the first URL; if none is found it raises this Exception pointing at the workflow configuration. It prevents a null-URL download attempt of the final video.

Source

Thrown at web/pipelines/i2v.py:341

                            workflow_input = workflow_config["workflow_id"]
                        else:
                            workflow_input = str(workflow_path)

                        video_result = await kit.execute(workflow_input, workflow_params)

                        generated_video_url = None
                        if hasattr(video_result, 'videos') and video_result.videos:
                            generated_video_url = video_result.videos[0]
                        elif hasattr(video_result, 'outputs') and video_result.outputs:
                            for node_id, node_output in video_result.outputs.items():
                                if isinstance(node_output, dict) and 'videos' in node_output:
                                    videos = node_output['videos']
                                    if videos and len(videos) > 0:
                                        generated_video_url = videos[0]
                                        break

                        if not generated_video_url:
                            raise Exception("The workflow did not return a video. Please check the workflow configuration.")

                        timeout = httpx.Timeout(300.0)
                        async with httpx.AsyncClient(timeout=timeout) as client:
                            response = await client.get(generated_video_url)
                            response.raise_for_status()
                            with open(final_video_path, 'wb') as f:
                                f.write(response.content)
                        progress_bar.progress(100)
                        status_text.text(tr("status.success"))
                        await save_web_generation_history(
                            pixelle_video,
                            task_id=task_id,
                            video_path=final_video_path,
                            pipeline="image_to_video",
                            title="图生视频" if get_language() == "zh_CN" else "Image to Video",
                            input_params={
                                "text": prompt,
                                "prompt_text": prompt,

View on GitHub (pinned to 848b054e4f)

Solutions

Inspect the workflow JSON and ensure it ends with a proper video output node (VHS_VideoCombine / SaveVideo etc.)
Check ComfyUI server logs for node failures (OOM, missing custom nodes) during execution
Verify the 'image' param placeholder name matches the workflow's LoadImage node input
Broaden the result-parsing to check alternate keys ('gifs', 'images') if your output node classifies video differently

Example fix

// before
if not generated_video_url:
    raise Exception("The workflow did not return a video. Please check the workflow configuration.")
// after
if not generated_video_url:
    raise RuntimeError(
        f"i2v workflow returned no video. workflow={workflow_path}, "
        f"outputs={node_outputs!r} — verify the video output node exists and ran."
    )

Defensive patterns

Strategy: validation

Validate before calling

node_outputs = result.get("node_outputs", [])
video_urls = [v for out in node_outputs if isinstance(out, dict) for v in out.get("videos") or out.get("gifs") or []]
if not video_urls:
    raise RuntimeError(f"i2v workflow returned no video: {node_outputs!r}")

Type guard

def extracted_video_url(node_outputs):
    for o in (node_outputs or []):
        if isinstance(o, dict):
            vids = o.get("videos") or []
            if vids:
                return vids[0]
    return None

Try / catch

try:
    video = await generate_audio_visual_video(...)
except Exception as e:
    if "did not return a video" in str(e):
        log.error("i2v workflow output had no videos; check workflow config and ComfyUI logs")
    raise

Prevention

Verify the exported workflow JSON ends in a video-producing node before deploying
Monitor ComfyUI server logs for silent node failures (OOM, missing nodes)
Match the 'image' param key to the workflow's LoadImage placeholder exactly
Add a smoke test that runs each i2v workflow and asserts a video URL is returned

When it happens

Trigger: The i2v workflow completes but no node emits a non-empty 'videos' array — the workflow lacks a video output node, the video node failed or was muted, outputs are returned under 'images'/'gifs' instead, or input param 'image' was not bound so the video node never produced output.

Common situations: Using a workflow JSON exported from a different ComfyUI/custom-node version where the output node name/key changed; replacing a SaveAnimatedWEBP/VHS node with a SaveImage node; ComfyUI queue error swallowed so downstream output missing; insufficient VRAM causing the video encode node to fail silently in the parsed results.

Related errors

AI-assisted analysis of ATH-MaaS/Pixelle-Video@848b054e4f (2026-08-30). Data as JSON: /api/errors/774debfa65c5af19. Report an issue: GitHub.