In an earlier post I described the structure of quantization tables in a JPEG file. Closer inspection reveals that there are two different ways to store the information.
Now, here is a sample image:
As it can easily be seen, the file comprises of a single DQT segment. This segment in turn contains two quantization tables.
The JPEG standard allows to define multiple tables in a single segment. Also it is perfectly ok to include multiple DQT segments in a file. Once again the same sample image, with a different ordering of the quantization tables:
And here is the structure of the 2nd file's header:
There is no obvious difference to be seen between these two images. However, due to the additional segment the second file is 4 bytes larger than the first one.
Whether the quantization table is defined in one or multiple segments could give a weak indication of a camera or image processing software. However, as it can be seen from the example above it is very easy to tamper with this property.
By the way: Multiple DQT segments are not allowed in the preview image which is a part of the EXIF data structure. In version 2.2 of the EXIF standard (JEITA CP-3451) they say:
All quantization tables shall be defined in one DQT marker segment; recording of multiple DQT markers is not allowed.
