Focal Plane Resolution misreported for Canon 6d mk2

[tomaszg7]

Describe the bug

Values of FocalPlaneXResolution and FocalPlaneYResolution for images taken winth Canon 6d mk2 are incorrect. I suspect it is a bug in hardware and exiv2 just reports what is sitting in the file, however it leads to problems down the line as software miscomputes e.g. crop factor based on this data. I don't know if exiv2 practice is to add a workaround to this issue.

To Reproduce

Steps to reproduce the behavior:

1. On a clean file from the camera I get:

exiv2 --print a img_20240309_155531.cr2 |grep FocalPlane
Exif.Photo.FocalPlaneXResolution             Rational    1  6673.8
Exif.Photo.FocalPlaneYResolution             Rational    1  6720.52
Exif.Photo.FocalPlaneResolutionUnit          Short       1  inch

Expected behavior

Photos have width 6240 which would make sensor width approx. 1 inch. It is incorrect, since it is a full-frame camera with sensor 35.9 mm wide. Either sensor resolution should be lower, or resolution unit is misreported.

Desktop (please complete the following information):

• OS and version: Linux (Gentoo)
• Exiv2 version and source: exiv2 0.28.3 (from distro)
• Compiler and version: Gcc 14
• Compilation mode and/or compiler flags: -O2

Additional context

• The problem is also present in exiftool.
• It causes e.g. darktable to compute crop factor as 1.5 and apply incorrect algorithms to process image.
• If necessary I can supply a sample file.
• I wanted to test it on other Canon, but raw.pixls.us is regretably down, so I checked only the results for 5d, 7d and 30d. All looked correct.

[boardhead]

This is likely a Canon bug. While I haven't checked this specific model, Canon is known to write an incorrect FocalPlaneResolution for some models.

• Phil

[tomaszg7]

I suppose so. What would be a proper way to handle it? Ignore it, since it is Canon bug, or add a workaround to Exiv2?