Question for Pekka

[Les Swartz]

You stated when converting RAW to TIFF you custom select white point and choose contrast for each image. Does this method offer an advantage over converting and THEN doing the adjustments in PS or PSP, etc? Also, it sounds like you are using Zoombrowser for the above action. Am I correct?

As a "newbie" here, thank you for the forum and especially for the help you give.

Thanks
Les

[Pekka]

Hi Les,

Technical advantage is that in TWAIN dialog (same as Zoombrowser custom settings dialog) you work in 10-bits per channel (30-bit image). This means your adjustments in that mode do not pose a threat to final image quality (in theory). If you convert to 8-bits per channel (24-bit image), you loose the 'hidden headroom' you have in original CRW.

One D30 in dpreview.com user made some tests between 24-bit and 48 bit images and the concusion is that the advantage can not be seen if the image does not contain full dynamic scale. So in many images you don't see any difference. See http://www.dpreview.com/forums/read.asp?forum=1019&page=3&message=1008420

But in my experience doing custom RAW conversion is much quicker and more comfortable way to seek for the right settings than in PS, so I do it by habit just to see if it helps getting the right WB the easy way. But note that custom RAW->TIFF WB picker and settings can not only be used for seeking the right colors (sometimes even this is hard), but also in making the colors look better.

See the fruit shots: http://photography-on-the.net/forum/showthread.php?t=129
The colors were ruined by a TV nearby, and I tried to restore colors in PS, but in Zoombrowser it took one mouseclick to get the colors back to correct.

Try it and you'll see if it works for you.

PS. I've noticed that the gray picker in Photoshop is NOT the same as the WB picker in TWAIN dialog. It seems that the WB picker does something undocumented :)

Pekka

[ega1]

Pekka wrote:
PS. I've noticed that the gray picker in Photoshop is NOT the same as the WB picker in TWAIN dialog. It seems that the WB picker does something undocumented :)

Pekka


I think you're right. Recently I took a photo under fairly bad conditions and it came out with a nasty greenish hue. I tried fixing it with my image editor, but could never quite get the shadows right. The Canon TWAIN WB picker fixed it right up with almost no effort.

[nadim]

Pekka,

there is another theory.. if Canon engineers are smart, (and I believe they are), this is how they could have implemented the Custom WB in CRW before the conversion.

The Raw CRW images contains the value of the pixels before interpolation, i.e. as the mosaic Bayer pattern. Every pixel contain 10 bits of information (i.e. levels between 0 to 1023, 4 times more accuracy than 8 bits) of a monochrome color (Red, Green or Blue only).

The Custom WB could effectively work if implemented right on these pixels before interpolation, and therefore, the color correction is just an amplification factor applied separately to the G, the B and the R pixels. The interpolation to create the 3.3Mpixels image is done then on color corrected pixels. This would be far superior to correcting colors on already mixed/interpolated pixels.

We'll never know if this is the way Canon implemented their algorithm. nadim

[Geert]

This is most likely what they do, given that the software has to be smart enough to do the color interpolation anyway. A bit related to this, I wonder if there is a way to get an uncompressed RAW image, which basically is just a series of 10-bit (possibly padded to 16-bit) values representing the exact contents of the CCD.

I would be very interested in this in order to develop a customized calibrated conversion tool. The idea is to take a series of test shots of a few known targets, such as the Kodak Q-60 one, in order to measure the response characteristics of the camera sensors. For the basic idea see http://www.aim-dtp.net. My goals are a bit different though: I don't want absolute color, but rather optimal compensation for known CCD characteristics or flaws.

As we know, not all 3.3 million pixels are equally good, "hot" pixels being the extreme example. It would be nice to use a set of benchmark pictures in a controlled environment (esp. lighting) to characterize all sensors and have the conversion software produce a final image based on that.

I realize that this has not so much to do with the creative part of photography as well an urge to really understand the entire process of digital photography.

-Geert

[nadim]

Geert,

Dan Foody is trying to uncompress the G1 CRW data... I know that Eric Hyman has decoded the D30 CRW data for Bible (www.biblelabs.com).

You may want to contact Dan directly, and check with him... if you are more lucky that we have been so far, I'd love to have access to these data too.

Nadim

[kennethwoodruff]

i'm reasonably certain that i read in the documentation for the g1 that it does not divulge > 8 bit per channel by any of the methods of acquisition.

did i misread something, or was this misstatement amended?