odysseus-dev/odysseus · error · HTTPException

detail[-4000:]

Error message

detail[-4000:]

What it means

This is not a fixed message: _run_bridge raises HTTPException(500, detail[-4000:]) when the MLX Swift bridge subprocess (odysseus-mlx-inpaint / mlx-lama-serve / odysseus-mlx-colorize / mlx-ddcolor-serve) exits non-zero. The detail is the last 4000 chars of the bridge's stderr (or stdout), also logged server-side, so the visible text is the underlying Swift/MLX failure.

Source

Thrown at scripts/mlx_image_server.py:211

        img = Image.open(io.BytesIO(raw))
        if img.mode == "RGBA":
            # OpenAI edits mask convention: transparent = regenerate.
            alpha = img.getchannel("A")
            mask = alpha.point(lambda p: 255 if p < 128 else 0)
        else:
            mask = img.convert("L")
        mask.save(out_path, format="PNG")
    except Exception as e:
        raise HTTPException(400, f"Invalid mask image: {e}") from e


def _run_bridge(cmd: list[str]) -> None:
    env = os.environ.copy()
    proc = subprocess.run(cmd, env=env, text=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    if proc.returncode != 0:
        detail = (proc.stderr or proc.stdout or "MLX Swift bridge failed").strip()
        logger.error("MLX Swift bridge failed (%s): %s\n%s", proc.returncode, " ".join(cmd), detail[-4000:])
        raise HTTPException(500, detail[-4000:])


def _run_ddcolor_bridge(model: str, image_raw: bytes, out_path: Path) -> None:
    bridge = _resolve_bridge(["odysseus-mlx-colorize", "mlx-ddcolor-serve"])
    if not bridge:
        raise _unsupported_swift_mlx_runtime(model)
    with tempfile.TemporaryDirectory(prefix="odysseus-ddcolor-") as td:
        inp = Path(td) / "input.png"
        _write_bridge_input_image(image_raw, inp)
        weights = _weights_path(model)
        tier = "tiny" if "tiny" in model.lower() else "large"
        _run_bridge([
            bridge,
            "--model", str(weights),
            "--image", str(inp),
            "--output", str(out_path),
            "--tier", tier,
        ])

View on GitHub (pinned to f9235ebbf1)

Solutions

1. Read the embedded stderr tail (and server log line 'MLX Swift bridge failed') to identify the real cause
2. Re-run with a smaller image (resize to e.g. 1024px max) if the tail shows a memory error
3. Verify the bridge executable runs standalone with --help, and reinstall the MLX Swift package if it doesn't
4. Ensure model weights resolved correctly (see errors 941/942) and are compatible with the bridge tier

import shutil
bridge = shutil.which('odysseus-mlx-inpaint') or shutil.which('mlx-lama-serve')
assert bridge, 'MLX Swift bridge not installed'

except HTTPException as e:
    tail = e.detail[-500:]
    if 'out of memory' in tail.lower():
        shrink_image_to(1024); retry_once()
    else:
        surface_bridge_stderr(tail)

When it happens

Trigger: Any /v1/images/edits or /v1/images/harmonize call on LaMa/MI-GAN/DDColor models where the bridge process fails: missing weights, out-of-memory, incompatible Swift toolchain, bad weights path, or unsupported image dimensions.

Common situations: First run before model weights are downloaded; macOS or Xcode/Swift version mismatch for the MLX Swift package; large input image exhausting unified memory; passing a safetensors file the bridge doesn't support; bridge binary half-installed after a partial pip/brew install.

Related errors

• MLX Swift bridge completed but did not write {out_path}
• No safetensors weights found for {model} in {snap}
• LaMa / MI-GAN inpainting requires an image mask. Use the edi
• HiDream generator script not found in snapshot: {script}
• Boogu MLX generation failed: {e}

AI-assisted analysis of odysseus-dev/odysseus@f9235ebbf1 (2026-08-14). Data as JSON: /api/errors/063d4435b64de5f1. Report an issue: GitHub.