[drougge]

Image Size: 4064x3000 (what are those 64 more colums ? Cannot be for SR as there are no additional lines)

They are probably for calibrating something (black level and similar). Masked pixels. It's the true size of the raw data, but bigger than the ActiveArea. They look like patterned noise when you look at them. Here are the first 128 rows of yours:



You can extract them yourself with dd if=IMGP2562.DNG of=rawdata bs=1 skip=182620 count=18288000, skip is from the StripeOffset tag. This gives you 12 bpp raw image data (whole sensor area), converting that to something more useful is left as an exercise for the reader.

[skyclad666]

They are probably for calibrating something (black level and similar). Masked pixels. It's the true size of the raw data, but bigger than the ActiveArea. They look like patterned noise when you look at them. Here are the first 128 rows of yours:
You can extract them yourself with dd if=IMGP2562.DNG of=rawdata bs=1 skip=182620 count=18288000, skip is from the StripeOffset tag. This gives you 12 bpp raw image data (whole sensor area), converting that to something more useful is left as an exercise for the reader.

Interesting proposition, Drougge! Your practical work is also quite informative. However, can you detail your process to obtain the PNG sample picture of the 128 non visible rows ? Thanks