mudler/LocalAI · error

3

3

Error message

[sam3-cpp] Failed to encode image\n

What it means

Emitted by the sam3-cpp backend's C shim when sam3_encode_image() fails after an image was successfully loaded from disk. It means the SAM 3 image encoder could not preprocess/embed the loaded pixels (e.g. unsupported dimensions, corrupt pixel data, or an encoder/model state mismatch). The function returns exit code 3 so the Go caller can distinguish it from 'model not loaded' (1) and 'image load failure' (2).

Source

Thrown at backend/go/sam3-cpp/cpp/gosam3.cpp:79

    fprintf(stderr, "[sam3-cpp] Model loaded: %s (threads=%d)\n", model_path, threads);
    return 0;
}

int sam3_cpp_encode_image(const char *image_path) {
    if (!g_model || !g_state) {
        fprintf(stderr, "[sam3-cpp] Model not loaded\n");
        return 1;
    }

    sam3_image img = sam3_load_image(image_path);
    if (img.data.empty()) {
        fprintf(stderr, "[sam3-cpp] Failed to load image: %s\n", image_path);
        return 2;
    }

    if (!sam3_encode_image(*g_state, *g_model, img)) {
        fprintf(stderr, "[sam3-cpp] Failed to encode image\n");
        return 3;
    }

    return 0;
}

int sam3_cpp_segment_pvs(float *points, int n_point_triples,
                         float *boxes, int n_box_quads,
                         float threshold) {
    if (!g_model || !g_state) {
        return -1;
    }

    sam3_pvs_params pvs_params;

    // Parse points: each triple is [x, y, label]
    for (int i = 0; i < n_point_triples; i++) {
        float x = points[i * 3];

View on GitHub (pinned to 44413a9d06)

Solutions

1. Verify the image loads and has sane dimensions (open it in an image tool; re-export as PNG/JPEG) and retry
2. Confirm the model was fully loaded first: sam3_cpp_load returned 0 and no 'Model not loaded' (code 1) was printed
3. If it persists after a successful reload, rebuild/inspect the underlying sam3 library's encode step for the exact failure (dimensions, normalization, memory)

Example fix

// before: encode immediately after load
sam3_cpp_load(model_path, "cpu");
sam3_cpp_encode_image("weird.tiff"); // returns 3

// after: re-export to a well-supported format and check the return code
if (sam3_cpp_encode_image("input.png") != 0) {
    // handle: image load (2) vs encode (3) failure
}

// Go caller: ensure model is loaded and the image decodes before encoding
if sam3Loaded != 0 {
    return fmt.Errorf("sam3 model not loaded")
}
f, err := os.Open(imagePath)
if err != nil {
    return fmt.Errorf("image unreadable: %w", err)
}
_, format, err := image.DecodeConfig(bufio.NewReader(f))
f.Close()
if err != nil || (format != "png" && format != "jpeg") {
    return fmt.Errorf("image is not png/jpeg")
}

When it happens

Trigger: Calling sam3_cpp_encode_image(image_path) when g_model/g_state are set but the loaded image cannot be encoded: wrong image dimensions for the model, a truncated file that stb_image still decodes partially, or a model/state pair left inconsistent by a failed prior operation.

Common situations: Running segmentation (PVS/PCS) on an image the model was not prepared for, reusing a stale global state after a model reload, or feeding an exotic/corrupt image format that decodes but has invalid metadata.

Related errors

• -1
• Web Audio API not available
• Web Audio API not available
• engine_args must be a JSON object, got {type(extra).__name__
• unknown engine_args key {key!r}.{hint}

AI-assisted analysis of mudler/LocalAI@44413a9d06 (2026-08-15). Data as JSON: /api/errors/f3f244263de2c19d. Report an issue: GitHub.