The dynamic range limitation by the camera is the biggest problem I have when shooting. Exposing to the right can often be a solution for raw shooters; however, the question is, how to know where the "right" ends, i.e. how high can one go with the exposure. I often shoot with exposure bracketing, sometimes five or more shots to be sure that one of them is exposed as I would like it - and somtimes the shot is not optimal even so.
The core of the problem is, that the histogram displayed in-camera is based on the JPEG image, which is embedded in the raw file. That image reflects the settings contrast, saturation, sharpness, color tone, and most importantly, the white balance. All of these settings can be neutralized, except the white balance.
Accordingly, the goal is to prod the camera into using WB coefficients 1.0. 1.0. 1.0, so that the resulting RGB histogram represents the raw data as far as possible. It makes no relevant difference, if the camera displays a luminosity histogram only or shows the three color histograms.
I made numerous attempts to find a WB setting in temperature and tone, which yields a neutral WB, but the result was mediocre, although with more patience and many test shots one could achieve a relative good approximation.
Then a fellow poster/photographer/pixel peeper, Guillermo, had a genial idea:
1. create an image with a solid color such, that the camera displays it as grey (with all other settings neutralized),
2. use this image for custom WB setting.
Guillermo introduced his idea on a Luminous Landscape forum,
http://luminous-landscape.com/forum/index.php?showtopic=22250&hl=
and we discussed it quite long (101 posts), though the topic got a bit changed, one does not need to read so much.
Anyway, I extracted the essence and tuned the method a bit; everyone can repeat it (note, that some Nikon and high-end Canon cameras acceot the WB coefficients directly, so this issue is not a problem with those cameras).
The steps are:
1. Create an image with all possible hues and saturations.
Guillermo did that, it can be downloaded from
w w w.cryptobola.com/PhotoBola/WBCalibrationChart_Guillermo.jpg
(I spaced the
http://www
in order to prevent displaying that large image)
2. Display it on a monitor and make a shot. The lighting should be controlled; the simplest is to do it in a dark room. The distance should be enough to prevent so high resolution, that the monitor lines become visible; slight defocusing is advised
3. Find the point in the resulting image, where it is grey. Guillermo used Photoshop for this.
I did it with Rawnalyze, see
http://forums.dpreview.com/forums/read.asp?forum=1019&message=26523531:
a. right clicking anywhere, where it looks somewhat greyish,
b. the selection statistics shows the raw values,
c. change the position according to the pixel stats; for example if the red is too high, click towards the green/blue corners, until the three colors in the stats are very close, let's say within 1%
d. note the position of this spot (from the selection stats)
4. Pass the image to Photoshop w/o any adjustment (this may require to explicitely eliminating some adjustments) and check out the RGB values of that spot, based on the noted co-ordinates,
5. Create an image as large as the monitor (as tall) filled with that color (GIF is the best); save this image as layered TIFF for iterations. The image of this block has to completely fill the circle in the middle of the viewfinder (and then some more) for Canon cameras, as that area will be used for taking custom white balance,
6. Display that image on the same monitor under the same lighting conditions and make a shot of it,
7. Check out the result with Rawnalyze; the raw values of the three colors should be very close (select an area and look at the average values),
8. If the result is not grey, change the fill color accordingly; for example if the red is higher than the oders, reduce it in the filler, then repeat it from 7.
I needed four iterations.
9. Save the last/best shot, make it read only and load it on every memory card (the location and the acceptable names depend on the camera).
10. Use this image for custom white balancing.
If the camera supports custom settings, like the Canon 40D, these settings can be assigned to such a custom group.
The result with my 40D is so good, that I see tiny clippings on the display (blinking area) exactly as I see them on the raw image.
[b]Do not[
b] test this with monochromatic light, like LED or laser, it will not work. We can not eliminate the de-mosaicing and the adjacent color transformation from the camera's color space to sRGB or whatever.
However, in praxis it works excellently.
There are some downsides:
1. the embedded JPEG is useless,
2. raw+JPEG is senseless,
3. the color of the image displayed in-camera is horrendeous,
4. [b]white balancing in the raw processing is a must[
b]; don't forget to shoot something white if there is no suitable subject in the scenery.
Despite the numerous disadvantages, this is now my firm setting in all custom groups.
The process may appear complicated, but it needs to be done only once, and those using Canon 20D or 40D can save all the work except step 10 by using one of these white-balance templates (the first one is for 20D, the second one is for 40D):
http://www.cryptobola.com/PhotoBola/WB_00020.CR2
http://www.cryptobola.com/PhotoBola/WB_00040.CR2
