RAW VNG interpolation and '4 colours' (-f in dcraw)

[tomczak]

Is there a way of treating G RAW channels as separate colours when developing RAW using VNG interpolation (or perhaps it is already done this way)? Cheers!

[den]

Maciej...

I am not really sure what you are asking but in any case I came across RAWHide Converter (RHc) at http://www.my-spot.com/rhc/. If you click on the left side link listing "Demosaicing", there will be a description of six methods, one of which is Advanced Chroma Corrective demosaicing (ACC) which might be of interest...

...while I have downloaded the program, I have not yet experimented with it and may not... ...as for my Canon EOS-350D CR2 files, AHD demosaicing seems to be sufficient...

Note: Fuji cameras are not listed among those that are officially supported: http://www.my-spot.com/rhc/RHC_Cameras.htm

Advise if you find "RAWHide Converter" useful should you try it...

...den...

[tomczak]

Hi Den,

This is pretty uncanny - I just discovered the RawHide a few days ago! I more or less know what it can do now: can't address colour bleeding at edges, which what I was after, but it's quite capable and non-nonsense otherwise.

The explicit choice of several demosaicking algorithms that can be compared is worth the price of admission alone.

I especially like its histogram style: x-axis is scaled in exposure stops, it shows the histogram of RAW data (after WB and clipped channels recovery, shows the clipping R,G,B limits and the clipping limit of the output file. I almost understand how it works - if I got it right, this single histogram shows the only two things that I want to know every time I work with a RAW image: the distribution of RGB values in INPUT RAW (and which ones are actually clipped), and the distribution of RGB values of the OUTPUT image (with clipping warning) once the demosaicking, WB scaling, highlight recovery, exposure scaling and curves are all done. Refreshing!

The histogram is linked to the image (similar to PWP Curves histograms): clicking on the preview image, shows the corresponding brightness of this area on the histogram. It doesn't go the other way though: i.e. there is no mapping of clipped/near-clipped areas.

One problem that I already noticed is that Saturation slider can easily blow a channel or two in the output image, but the histogram doesn't reflect it.

p.s. as to the original question: I understand that standard VNG interpolation uses three RAW channels: RGB. A variant of it can use 4 channels RGBG (i.e. treat green as two separate channels), which may help in reducing some artifacts. I wondered what PWP VNG implementation does.

http://www.pages.drexel.edu/~par24/rawh ... facts.html