Need help to inference image for upscaling

[md-rifatkhan]

I've tried a upscale s simple red image, but output is weired,

    Future<void> loadModel() async {
    print(ortEnv.availableProviders());
    final byteData =
        await rootBundle.load('assets/realesr-general-x4v3-fp32.onnx');
    final buffer = byteData.buffer;

    OrtSessionOptions options = OrtSessionOptions();
    session = OrtSession.fromBuffer(buffer.asUint8List(), options);
  }

  Future<void> runInference() async {
    OrtRunOptions options = OrtRunOptions();

    img.Image image = img.Image(256, 256);

    // Sample image with red color
    int redColor = img.getColor(255, 255, 0);
    for (int y = 0; y < image.height; y++) {
      for (int x = 0; x < image.width; x++) {
        image.setPixel(x, y, redColor);
      }
    }
    orginalImage = image;

    List<int> imageBytesList = image.getBytes();
    Float32List normalizedInput = Float32List.fromList(imageBytesList
        .map((b) => b / 255.0)
        .toList());
    Float32List inputTensor = Float32List(image.height * image.width * 3);
    int i = 0;
    for (int h = 0; h < image.height; h++) {
      for (int w = 0; w < image.width; w++) {
        inputTensor[i++] = normalizedInput[h * image.width * 3 + w * 3 + 0];
        inputTensor[i++] = normalizedInput[h * image.width * 3 + w * 3 + 1];
        inputTensor[i++] = normalizedInput[h * image.width * 3 + w * 3 + 2];

      }
    }
    final createInput = OrtValueTensor.createTensorWithDataList(
        inputTensor, [1, 3, image.height, image.width]);

    final result = session.runAsync(options, {'input': createInput});
    var outputResult = await result!;
    List outputTensor = outputResult[0]?.value as List;

    if (outputTensor is List<List<List<List<double>>>>) {

      List<List<List<List<double>>>> tensorData = outputTensor;

      int height = tensorData[0][0].length;
      int width = tensorData[0][0][0].length;
      int channel = tensorData[0].length;

      Uint8List outputImageData = Uint8List(width * height * 4);
      int i = 0;
      for (int h = 0; h < height; h++) {
        for (int w = 0; w < width; w++) {
          double red = (tensorData[0][0][h][w] * 255).clamp(0, 255);
          double green = (tensorData[0][1][h][w] * 255).clamp(0, 255);
          double blue = (tensorData[0][2][h][w] * 255).clamp(0, 255);
          outputImageData[i++] = red.toInt();
          outputImageData[i++] = green.toInt();
          outputImageData[i++] = blue.toInt();
          outputImageData[i++] = 255;
        }
      }

      setState(() {
        outputImage = img.Image.fromBytes(
            width, height, outputImageData);
        ScaffoldMessenger.of(context).showSnackBar(
          SnackBar(
            content: Text('Output Image Dimensions: $width x $height'),
            duration: const Duration(seconds: 2),
          ),
        );
      });
    }
  }