I am trying to add attributes to a dataset to follow the image specification as laid out in the docs here.
I see that there is an attribute called IMAGE_ASPECTRATIO, which can either be the pixel aspect ratio, or the image aspect ration (width / height), but the datatype is listed as unsigned int. Not an array, so it cannot be [16,9] or something, just a single unsigned int. I don’t understand how this can be. Does anyone know what this is supposed to represent?
You’re right to be confused. The HDF5 IMAGE_ASPECTRATO attribute exemplifies an unfortunate, incomplete specification. The encoding scheme is completely missing. Precision guidelines have not been specified. There are no examples provided. Additionally, the distinction between pixel aspect ratio and image aspect ratio is unclear. I believe most scientific software overlooks this attribute entirely, and few, if any, implementations that utilize it would need to document their own encoding scheme. We apologize for this oversight.
Thank you for the reply. Good to know I’m not missing something 
