Comfy-Org/ComfyUI · error · RuntimeError

The model returned no layers. Try a different prompt or inpu

Error message

The model returned no layers. Try a different prompt or input image.

What it means

Raised by ByteDanceSeedreamLayerSeparationNode when the response contained a base image but zero layer items with a 'url' (entries without URLs are already dropped with a warning before this check). The generation technically ran but produced no separable layers for this input.

Source

Thrown at comfy_api_nodes/nodes_bytedance.py:1272

        if not data or "url" not in data[0]:
            raise RuntimeError("Unexpected response: no base image returned.")
        base_item = data[0]
        if base_item.get("bounding_box") is not None:
            logging.warning(
                "ByteDance layer separation: base item unexpectedly carries a bounding_box; ignoring it."
            )
        if z_index_of(base_item) not in (0, 1_000_000):
            raise RuntimeError("Unexpected response: the first item is not the base image.")
        layer_items = [d for d in data[1:] if "url" in d]
        dropped = len(data) - 1 - len(layer_items)
        if dropped > 0:
            logging.warning(
                "ByteDance layer separation: %d of %d returned elements had no 'url' and were dropped.",
                dropped,
                len(data) - 1,
            )
        if not layer_items:
            raise RuntimeError("The model returned no layers. Try a different prompt or input image.")
        layer_items.sort(key=z_index_of)

        base_image = (await download_url_to_image_tensor(str(base_item["url"])))[..., :3].contiguous()
        height, width = base_image.shape[1], base_image.shape[2]

        specs = []
        for item in layer_items:
            flags = []
            bbox = item.get("bounding_box")
            absolute = bbox.get("absolute") if isinstance(bbox, dict) else None
            if (
                isinstance(absolute, (list, tuple))
                and len(absolute) == 4
                and all(isinstance(v, (int, float)) and not isinstance(v, bool) for v in absolute)
            ):
                left, top, right, bottom = (int(round(v)) for v in absolute)
                rect_w, rect_h = right - left, bottom - top  # exclusive right/bottom
                if rect_w > width or rect_h > height:

View on GitHub (pinned to 1c6d8d45b3)

Solutions

Try a different input image with clearly separated objects and a clean background.
Adjust the prompt to describe the layer decomposition you want.
Ensure the input meets the node's constraints (>=512px, aspect ratio within 1:16..16:1).
Note the run was still billed; check api_logs if you believe it is a server fault.

Defensive patterns

Strategy: validation

Validate before calling

# pre-check the cheap local constraints before paying for a run
assert get_number_of_images(image) == 1
assert image.shape[2] >= 512 and image.shape[1] >= 512  # min dims
# then choose an image with clearly separable objects; no code can guarantee layers

Try / catch

try:
    base, layers = await separate_layers(image)
except RuntimeError as e:
    if 'no layers' in str(e):
        base, layers = await separate_layers(enhance_contrast(image))  # different input
    else:
        raise

Prevention

Use images with distinct foreground objects on a clean background.
Expect billing even when zero layers come back; validate inputs first.

When it happens

Trigger: Passing an image the model cannot decompose (flat single-object background, heavily compressed input, abstract noise) or an empty/irrelevant prompt for the layer task.

Common situations: Using low-contrast or texture-only images; images where no distinct foreground objects exist; prompts that describe new content instead of guiding separation.

Related errors

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/def35b45aae9623f. Report an issue: GitHub.