Writing RGB neuroglancer image with png compression

[LarsKoeppel]

Hi,

I'm trying to write an RGB image using TensorStore with PNG compression, but I'm encountering some issues. The resulting image looks strange—similar to subpixels on a display. I've attached images showing the problem (split into individual channels for clarity):

• Red Channel
  

• Green Channel
  

• Blue Channel
  

For reference, here's how the image is supposed to look:

Here's the code snippet I'm using:

tensorstore::Open({{"driver", "neuroglancer_precomputed"},
     {"kvstore", {
          {"driver", "file"},
          {"path", <path>}
      }},
     {"multiscale_metadata", {
          {"data_type", "uint8"},   
          {"num_channels", 3},
          {"type", "image"},
      }},
     {"scale_metadata", {
          {"resolution", {resolution[0], resolution[1], resolution[2]}},
          {"encoding", "png"},
          {"png_level", 6},
          {"chunk_size", {1024, 1024, 1}},
          {"size", {sizeX, sizeY, 1}},
      }},
     {"scale_index", 0},
    },
    context,
    tensorstore::OpenMode::create,
    tensorstore::RecheckCached{false},
    tensorstore::ReadWriteMode::write).value();

When I use the same code with a single-channel image, everything works fine.

Am I doing something wrong?
Any help or guidance would be greatly appreciated!