Near UniWB for D3
The thread started by Julia Borg, following a translation from a Spanish paper on a novel method of creating UniWB, seems to have been lost in the disc failures.
My personal exoperience was that the method did lead to a color which, when used to set the WB (like a gray card) gave coefficient that were closer to unity but I had trouble getting an exact match. In the end I went back to my engineering roots and did it empirically (i.e. I did a lot of trial and error). In the end I came up with an RGB color of 96/64/72 which gave RGB cooefficients of 1.000, 1.000, 1.012 or as close and can be to UniWB.
Iliah Borg has kindly agreed to host a jpg taken with this WB so everyone can access it. The file is at
Downloading it to a CF card, putting it in your D3 and setting the WB from that image should be all you need to do. If my understanding is correct you also may want to set up a Picture Control using a custom curve which is linear to make best use of it.
Downloading it to a CF card, putting it in your D3 and setting the WB
from that image should be all you need to do. If my understanding is
correct you also may want to set up a Picture Control using a custom
curve which is linear to make best use of it.
I tried doing this and my WB was way off. In an incandescent lighted room everything was overly yellowish but when I used 2700K everything looked good. Possibly I did something wrong?
“Be kind whenever possible...It is always possible.”
It does look very green when taking images and reviewing them on the D3's screen, you have to correct the WB in NX or some other RAW converter as part of the development process.
What it does, is effectively make the histogram on the D3 more accurate with RAW capture, it removes the effects of WB and, if you also use a linear tone curve it ensures that the blinkies etc. are as close to being accurate for the RAW data as they can be. This in turn enables ETTR to be used more effectively and with less "fingers crossed" that highlights have not been blown in any channel.
The original article referred to by Julia was in Spanish, there is a translation on Luminous Landscapes here for those interested:
Jim Keye wrote:
are those RGB values in?
How exactly the numbers are displayed depends on your monitor. For some monitors they will not be even a good starting point (no, I do not mean monochrome monitors).
The real reason I like UniWB is because with it I can use colour correction filters in a predictable manner, and use hand-held colormeter to find out what are the filters I need for the scene.
It is also very useful when evaluating the histogram of strong-coloured objects, like red or blue flowers.
For those who practice ETTR coloured filters on the lens or on the light sources allow to expose all channels up to the maximum.
Here is an old example
Both left and right images are captured with D2X using UniWB,linear custom curve; SB28 flash was used as a light source. For the right image flash is filtered through magenta filter, CC40m. That allowed to increase the exposure by 2/3 eV. Both images originated from direct raw data dump, demosaicing is done by binning (4 pixels of RGBG Bayer pattern substituted with one RGB pixel, greens are averaged G=(G1+G2) 2 ), processing was done in the following manner: assigning linear gamma space to the binned image, conversion of the image to gamma=2.2 colour space, and cropping. No additional colour adjustment or white balance was applied.
You can see that on the left image red and blue channels are underexposed while green channels are essentially the same on both.
agreed, the use of the monitor itself as a setting tool only works when the original target and the derived color are displayed and shot with the same camera from the same monitor so errors cancel out. The jpg file, however, should get you pretty close!
thanks philip. i've been using uniWB on my D200 and D2H, and have been waiting for one for the D3.
Before uploading this file (D3_NearUniWB.JPG) it needs to be renamed, e.g. DSC0001.JPG then placed in the current shooting folder on the CF card.
Then, when you wish to use the 'near UniWB' for preset WB, use the method given on page 159 of the (English) D3 manual to load from the image file into a preset.
Hope that helps?
J Kuek wrote:
I tried copying the file to my CF card , but the D3 cannot see the
file... what am I doing wrong? Thanks!
Jim Keye wrote:
if we're talking about calibrated monitors, what would be the
difference between color numbers and a jpg picture of color numbers?
It depends on the calibration of both monitors. In my case difference between two monitors manufactured by different companies is about deltaE = 5, depending on the brightness of colour and colour itself.
Thank you very much for posting this information.
Can you provide more information on how to "set up a Picture Control using a custom curve which is linear"?
I've looked at the "custom curve" controls in the Picture Control application, but have no idea how to build a "linear" custom curve. Have you done this, and if so can you talk us through how to do it?
when you select the custom curve option, you need to make sure the curve is a straight line from bottom left to top right (I think that is the default)
|Steamin' Mad by ahrensjt|
from Angered Subjects (Street Photography)
|Smile by Olymguy|
from Ultra Asian Indian Female Faces
|Space Shuttle Cockpit- by vbuhay|
from Aircraft Control Stick