Brendan Duncan

I was able to add 32-bit and 16-bit float format to the TIFF decoder without too much fuss. I'll hold off on pushing a new version until I can test...

I went ahead and pushed the new version with the float tiff to pub. I can always push a new version if I need to fix anything, and this way...

toFloatArray() is significantly slower than getFloat, since it allocates a Float32List and does a for-loop over all the pixels of the image to fill it. For 32-bit float images, `image.red.data`...

The rounding error is because that tiff is a 16-bit half-float, and that conversion is getting rounded. Maybe the tiff 16-bit float spec is different than the one I'm using,...

I tried another halfToFloat conversion function and got the same -1000 value, so my guess is the error is in the 16-bit tiff.

I don't mind the questions at all, happy to help. From what I've seen in the TIFF specs, https://www.adobe.io/open/standards/TIFF.html, I couldn't even find 16-bit float mentioned. TIFF is a kitchen...

I pushed 64-bit float tiff support to github, but I haven't tested it. I don't actually know how to create a 64-bit float tiff for testing it with.

I also noticed in the same way I was lazily always creating 4 channels, I was also always lazily making the HdrImage 16-bit half-float. I suspect that's where the rounding...

I pushed the fix for the floating-point precision, so the HdrImage will not be converted to 16-bit floats for every image. This fixes the -9999.0 => -10000 problem. I also...

The updates are in 2.1.18