iliah_borg

<p>Raw before being processed has some technical characteristics that can be checked; like the subject falls into noise-free zone, and it is not blown out. One can also check that the subject is in focus. It is possible to measure noise, density, linearity, flare, vignetting, etc. But the "look" of raw data is not very telling, as raw is open to interpretation.<br /> BTW, neither RawDigger nor FastRawViewer apply any arbitrary processing. Everything is totally under user control, and documented in manuals.</p>

<p>In some of the recent models WB pre-scaling is in analogue domain.<br> --<br> Best regards,<br> Iliah Borg</p>

<p>> I notice that Nikon scale the raw file contents according to the capture white balance<br> White balance pre-scaling (or preconditioning) are calibration constants, they are very close to "1", and they do not depend on the actual white balance the camera is set to.<br> We got several requests to "undo" WB pre-conditioning in LibRaw. So I decided to check what this pre-conditioning is.<br> From NEF analysis it seems Nikon are not recording various white balances presets into the NEFs, instead they have just one, "as shot". Canon and many others do have a portion of Makernotes dedicated to white balance presets, listing coefficients for each one, and numbers vary even for the same camera, correlated with the data for camera/sensor temperature). So Nikon must be going by a fixed white balance preset table in their raw conversion software/firmware.<br> Nikon way makes sense if there is no need to account for camera sample variation and temperature factors in raw conversion. But that means accounting for that should take place before the raw is written out. That is what WB pre-conditioning is for.<br> I've made several experiments with different camera bodies similar to this one: took my 2 D4 bodies and shot black frames (the usual, viewfinder closed, 1/8000, f/16, ISO 100, lens cap on). Shots were uncompressed 14-bit NEFs. Next, I brought the shots into RawDigger and examined the histograms. Neither the gaps between the rectangles nor the widths of the rectangles aligned between two different bodies of the same model.<br> --<br> Best regards,<br> Iliah Borg</p>