I wondered whether a 'better rawnalyse' would make a good sample application for the libraw library? For those who want to examine what use they are making of the available raw data count range.
The article White Balance in Digital Cameras: Problems mentions the use of Rawnalyze to investigate the raw channel data, but also notes some issues:
- the data is scaled
- unknown black compensation
Given that libraw splits off the fuctionality of dcraw into a library usable from multiple programs, I wondered whether a 'better rawnalyse' would make a good sample application for the library. It would report back on the unscaled, uncorrected, raw channel data (perhaps as histograms?) and give statistics on how many sensor cells had blown (assuming that a value of 4095 means 'most likely blown' rather than 'exactly one more than 4094'. Such statistics are hard to see from a histogram and would be better presented as a number.
I also think this would help people see how much channel scaling happens when the raw data is converted to an RGB image with a given white point.
(I downloaded and ran rawanalyze, but it turns out the Nikon D40 is an unsupported camera. I did read the documentation however).