invoke-ai/InvokeAI · error · ValueError

Wan VAE decode produced zero frames.

Error message

Wan VAE decode produced zero frames.

What it means

After decoding, invoke() compares the number of decoded frames to zero and raises if the Wan VAE decode produced none. This catches degenerate decode results (e.g. empty/invalid latent input that slipped through, or a decode returning an empty tensor) before video encoding proceeds.

Source

Thrown at invokeai/app/invocations/wan_latents_to_video.py:222

                                    _write_video_frames(writer, _iter_decoded_frames(chunk), context.util.is_canceled)
                            finally:
                                writer.close()
                        else:
                            # [C=3, T_pixel, H, W] in [-1, 1] (roughly), on CPU.
                            decoded = vae.decode(latents, return_dict=False)[0][0].cpu()
                            num_frames = decoded.shape[1]
                        del latents, latents_mean, latents_std
                finally:
                    # The VAE instance is cached and shared; don't leak tiling into other nodes.
                    if use_tiling:
                        vae.disable_tiling()

            TorchDevice.empty_cache()

            if context.util.is_canceled():
                raise CanceledException
            if num_frames == 0:
                raise ValueError("Wan VAE decode produced zero frames.")
            if num_frames != t_pixel:
                raise ValueError(f"Wan VAE decode produced {num_frames} frames; expected {t_pixel}.")

            height, width = h_pixel, w_pixel
            duration = num_frames / float(self.fps)
            if decoded is not None:
                context.logger.info(
                    f"Encoding MP4: {num_frames} frames @ {self.fps} fps "
                    f"({duration:.2f}s) at {width}x{height} via libx264"
                )
                context.util.signal_progress(f"Encoding MP4 ({num_frames} frames @ {self.fps} fps)")
                writer = make_mp4_writer(tmp_path, self.fps)
                try:
                    _write_video_frames(writer, _iter_decoded_frames(decoded), context.util.is_canceled)
                finally:
                    writer.close()
                del decoded
                TorchDevice.empty_cache()

View on GitHub (pinned to 0b6a024f2f)

Solutions

  1. Ensure latents have a valid temporal size and were produced by a Wan denoiser.
  2. Check the installed diffusers AutoencoderKLWan version for decode-output changes; upgrade/downgrade as needed.
  3. Inspect num_frames vs t_latent temporal scaling; verify frame_count/stride parameters are sane (>= 1).

Example fix

// before
latents = denoise(num_frames=1)  # decode yields 0 frames
video = wan_latents_to_video(latents=latents)  # ValueError
// after
latents = denoise(num_frames=21)  # temporal stride-safe frame count
video = wan_latents_to_video(latents=latents)
Defensive patterns

Strategy: validation

Validate before calling

expected = (latents.shape[2] - 1) * 4 + 1  # Wan temporal expansion
if expected <= 0:
    raise ValueError("Latents would decode to zero frames; check temporal dim and frame_count params")

Type guard

def decodes_to_frames(latents) -> bool:
    return latents.ndim == 5 and latents.shape[2] > 0 and latents.shape[3] > 0 and latents.shape[4] > 0

Try / catch

try:
    video = node.invoke(context)
except ValueError as e:
    if "produced zero frames" in str(e):
        latents = regenerate_latents(valid_frame_count=True)
        video = node.invoke(context)
    else:
        raise

Prevention

When it happens

Trigger: VAE decode returning a tensor with zero temporal frames — typically after decoding empty or all-invalid latents, or a decode implementation quirk (e.g. t=1 latents producing 0 frames under a slicing scheme).

Common situations: Edge-case frame counts (single-frame videos) interacting badly with temporal slicing; corrupted latents; a VAE version whose decode output layout changed.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/741ede5bc2a6682f. Report an issue: GitHub.