invoke-ai/InvokeAI · error · ValueError

Input videos are {width}x{height}; H.264 encoding requires e

Error message

Input videos are {width}x{height}; H.264 encoding requires even dimensions. Re-encode or crop the sources to even width and height first.

What it means

H.264 (libx264 + yuv420p) requires even width and height. video_concat keeps source dimensions exactly (macro_block_size=1), so it rejects odd-dimension inputs up front with a descriptive ValueError instead of silently resizing.

Source

Thrown at invokeai/app/invocations/video_concat.py:134

    def invoke(self, context: InvocationContext) -> VideoOutput:
        if len(self.videos) < 2:
            raise ValueError("video_concat requires at least two input videos.")

        paths: list[Path] = [context.videos.get_path(v.video_name) for v in self.videos]

        # Probe inputs up front: enforce matching dims and pick the default output fps.
        probes = [probe_video(p) for p in paths]
        widths = {(w, h) for (w, h, _, _) in probes}
        if len(widths) > 1:
            raise ValueError(
                f"All inputs must share the same dimensions. Got: "
                f"{sorted(widths)}. Re-render at a single resolution before concatenating."
            )
        width, height, _, _first_fps = probes[0]
        # libx264 + yuv420p needs even dimensions; we encode with macro_block_size=1 to
        # preserve the source dimensions exactly, so reject odd sources with a clear error.
        if width % 2 or height % 2:
            raise ValueError(
                f"Input videos are {width}x{height}; H.264 encoding requires even dimensions. "
                "Re-encode or crop the sources to even width and height first."
            )
        self._validate_transition_memory(width, height)
        output_fps = self._resolve_output_fps([probe[3] for probe in probes])

        context.util.signal_progress(f"Joining {len(self.videos)} clip(s) ({self.transition}) @ {output_fps:.2f} fps")

        tmp = tempfile.NamedTemporaryFile(prefix="invokeai_video_concat_", suffix=".mp4", delete=False)
        tmp.close()
        tmp_path = Path(tmp.name)
        try:
            # Frames stream from the decoders straight into the encoder; only the
            # transition windows are buffered. See _iter_joined_frames.
            writer = make_mp4_writer(tmp_path, output_fps)
            num_frames = 0
            try:
                clip_iters = [iter_video_frames(p, is_canceled=context.util.is_canceled) for p in paths]

View on GitHub (pinned to 0b6a024f2f)

Solutions

  1. Re-encode or crop sources so both width and height are even (e.g. crop 853x480 to 852x480)
  2. Apply a crop/scale node in the workflow before video_concat
  3. Use ffmpeg: -vf "crop=trunc(iw/2)*2:trunc(ih/2)*2" to make dimensions even

Example fix

// before
videos=[clip_853x480]
// after
# pre-process: ffmpeg -i in.mp4 -vf crop=trunc(iw/2)*2:trunc(ih/2)*2 even.mp4
videos=[clip_852x480]
Defensive patterns

Strategy: validation

Validate before calling

w, h = probe_video(path)[:2]
if w % 2 or h % 2:
    raise ValueError(f"{path} is {w}x{h}; crop/scale to even dimensions before concat")

Type guard

def has_even_dimensions(w: int, h: int) -> bool:
    return w % 2 == 0 and h % 2 == 0

Try / catch

try:
    output = concat.invoke(context)
except ValueError as e:
    if "requires even dimensions" in str(e):
        videos = [reencode_even(v) for v in videos]  # ffmpeg crop=trunc(iw/2)*2:trunc(ih/2)*2
        output = concat.invoke(context)
    else:
        raise

Prevention

When it happens

Trigger: Concatenating videos whose width or height is odd, e.g. 853x480 or 1920x1081.

Common situations: Cropped screen recordings with odd crop sizes; legacy footage at odd dimensions; images-to-video pipelines producing 1-pixel-off sizes.

Related errors

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/10e7db788f2b5c94. Report an issue: GitHub.