DNG compression use faulty method

[rolkar]

The compression of 16 bit DNG data uses method 8, ZIP.

But, this method shall, according to the 1.4 spec, only be used for 32 bit integer and 64 bit float data.

16 bit data shall use method 7, which is lossless JPEG Huffman encoding.

The reasons for the choice of non compliant compression are

• The method 7 is not obvious how to implement
• Method 8 is supported by Adobe products (even though it is not in the spec)

This is not a good choice, and should be fixed.

[rolkar]

Some more info from a personal message on DPReview from 'geni1105'.

geni1105

BTW, some warning messages given out by Adobe DNG SDK, which I now use for my prototype QuickLook plugin:

iMac:DNG-testfiles georgniedrist$ qlmanage -p DP2M1372.X3F.dng Testing Quick Look preview with files: DP2M1372.X3F.dng Warning: Missing or invalid UniqueCameraModel Error: ZIP compression is limited to floating point and 32-bit integer and transparency masks (SubIFD 1)

Readily understandable, but does not harm ...

KR Georg