[kh1234567890]

Idly playing with Rawdigger I noticed that the K-7 RAW files have missing levels. It is every 7th or 8th bin that always has a zero count. I've looked at a number of files and they are all the same. Is it a bug in Rawdigger (same thing shows up in RawAnalyze) or a bug in the K-7 firmware ?

Any ideas anyone ?

Attached Images

 

[hcc]

I checked my K-7 DNG files with Photome and PDCU. The histograms do not miss every 7-8th column.

Hope that the info may help....

EDIT: It is possilbe that I saw only mapped data

[kh1234567890]

I checked my K-7 DNG files with Photome and PDCU. The histograms do not miss every 7-8th column.

Hope that the info may help....

Are you looking at the RAW histograms or at the mapped histograms (which is what PhotoMe or PDCU shows)? The holes (zeros) are only in the RAW files, by the time the data had been mapped etc. they obviously disappear.

[Just1MoreDave]

My guess is that it's because K-7 bits are 12-bit.

[kh1234567890]

My guess is that it's because K-7 bits are 12-bit.

12 bits is 4096 levels (0 to 4095). On a well exposed continuous tone image there should on average be some pixels at all of these levels. It would appear that there is some dodgy scaling being done after the K-7 (Samsung sensor) analog to digital converter, the effect of which is that some counts (recorded intensity levels) can't occur. On my K-7 it is every 7th or 8th count, on earlier K-7s it could be as bad as every 4th or 5th count.

On the K-5 a different approach is used, without such scaling but with an offset from zero.

All this will not make much difference at medium/high intensity levels, once the data had been gamma corrected and demosaiced. However at low intensity levels and at high ISO after scaling and demosaicing the missing counts will result in an apparent increase in image noise.