UniWB Link -- (and what about Monochrome??)

[Andrea B.]

UniWB, a unitary white balance setting for your camera which forces it to use unitary colour multipliers, came up in another topic. As I recently found a very nice summary of a method for setting UniWB, I thought I would make a separate UniWB thread to post the link. I'm hoping some folks will try this UniWB method and report back to this thread.

To review, using UniWB produces a reasonably accurate raw histogram which helps us set exposures properly or which better enables ETTR for highlight preservation in cameras with smaller dynamic range. Everyone is probably somewhat familiar with UniWB's green cast and the need to shoot a white/grey card to restore "correct" white balance when converting fotos made with UniWB.

There are a multitude of UniWB articles out there on the web. So everyone should be able to find scads of additional info about this topic.
Here is the link to an ETTR and UniWB write-up by photographer (and electrical engineer) Jim Kasson in his Last Word blog:  Using In-Camera Histograms for ETTR

If you want to cut to the chase, go directly to Section 7: Preparing for Monitor-Based UniWB and Section 8: A One-Step UniWB Procedure
You need Excel for some spreadsheet work unless you want to use a calculator.

**********************

Andrea's Monochrome ETTR Experiment

Using an in-camera Monochrome setting seems to be useful for ETTR. Monochrome's R, G, B and L histograms all look the same, of course. What I am currently trying to figure out is how accurate the histogram is when viewed in Monochrome and whether there are any supporting settings which might be useful. For Nikon cameras, this would be such settings as Picture Control's Brightness or Contrast and in-camera ADL. Other cams no doubt have similar settings.

I tripped onto using Monochrome to better deal with the difficult exposures encountered in my UV work in which all colour is false and blown channels abound. My thinking on using Monochrome ETTR is that if we get the Light right, the Colour will follow (eventually). 

Should that be a motto??? Get the Light Right and the Colour Will Follow. 
I'm certainly not claiming to be the first to ever think this. Eons & ages ago a very helpful photographer once showed me how to convert a colour foto by first dealing with the light in a monochrome version before tackling the colour. I don't think I have ever seen any write-ups of this. But then there is a huge web out there and little time for me to search out links.

To better test the feasibility of Monochrome ETTR, I need to shoot some comparisons with that and UniWB. So I'll try to do that and report back to this thread too.

[Frank Fremerey]

Eons & ages ago a very helpful photographer once showed me how to convert a colour foto by first dealing with the light in a monochrome version before tackling the colour. I don't think I have ever seen any write-ups of this. But then there is a huge web out there and little time for me to search out links.

Dan Margulis and his Photoshop Professional book series explain how to treat exposure and contrast apart
from color. Margulis is a very demanding read ... a physicists approach to color contrast and exposure and a
new way to see color for editing in 10 channels ... very enlighting and helpful ...the chapter about how to use the
green channel to sharpen hair and eyes independently of the skin .. wow.

http://www.moderncolorworkflow.com/dan-margulis

[Shane]

What I am currently trying to figure out is how accurate the histogram is when viewed in Monochrome and whether there are any supporting settings which might be useful.

In-camera conversion to monochrome uses a typical luminance formula
RGB Luminance value   =  0.3R + 0.59G + 0.11B

Note that this is heavily weighted towards the G channel.

Here's the problem with a luminance histogram.
It doesn't indicate when one or two channels may be saturated for strong colours e.g. red or blue flowers.
Example: If the RAW B channel (and/or R channel) data is at saturation point and the other channels are below saturation then the luminance histogram will under report a saturated channel.

Saturated Red & Blue Pixel:      Y = R255x0.30 + G250x0.59 + B255x0.11   = 252

Using UWB the histogram will "correctly" report the B channel is saturated and that reducing the exposure might be in order.

[Andrea B.]

Yes, I'm aware of that, of course. But, one (or two) oversaturated channel(s) is easy to deal with. I don't usually really care if I've blown one (or two) channel. Well, blown, but not too far, that is. There is always a point of no return.

Recently for example, I reduced blown red in a NEF by using an NX2 color point and decreasing the contrast in the blown red by a mere -2 points without affecting other colours/contrasts. [Other converters have other methods for dealing with this. Just wanted to present one example.]

**********

Shane, do you know of Kasson's UniWB tutorial which I linked above? I'm just setting up now to try it out. Last time I attempted to produce a UniWB, I got into some kind of crazy non-convergent loop. So I am wanting to try again with a new tutorial.

[Shane]

There is always a point of no return.

And that is when the original RAW data is saturated. Either RAW data is "blown" or it isn't, there is no in between. Once it is saturated then any "recovery" is an apparent recovery and still results in loss of detail. If you can recover a "blown" channel then it was an apparent recovery or, it wasn't really blown to begin with but just a result of WB, colour space application etc.

If the RAW data is truly saturated and you recover it then, you will lose any fine detail that may have been present in the subject. This of course may be acceptable in some cases.


I have used Kasson's method for a while now for creating UWB for a D800, D800IR830 and my D810. It works really well and is pretty quick providing you set it up by carefully following the directions.

[Peter Forsell]

And that is when the original RAW data is saturated. Either RAW data is "blown" or it isn't, there is no in between.

This is an important distinction. Capture NX-D (similarly NX2 before that) shows data as blown sometimes about 0.5 stops too early. This is frustrating, since Capture cannot "recover" even this unblown data but insists it is blown.

Lightroom seems to go wrong to the opposite direction by doing behind-the-scenes highlight recovery when opening a file, even when no recovery has been asked -- and LR represents blown data as not-blown.

For these reasons I gave up on Nikon Capture a decade ago but have been doing a test with every new release. All in vain ... looks like it is incurable. Lightroom's workflow on the other hand would fit my work style, but the undisclosed data massaging bothers me.

Rawdigger and/or FRV show the true raw data but regrettably both 1) cost extra and 2) are difficult to integrate seamlessly into my workflow.

[Andrea B.]

I highly recommend Raw Digger. It is well worth the cost to learn how your camera actually records.

*******

Meanwhile, on with my UniWB experiment......
I've created my red, green and blue TIFs and have the Excel spreadsheet set up.

Equipment: 
Nikon D810 + 70-200/4G AFS VR
f=8.
Focal length=100mm.

Settings:
ADL=Off.
WB=Auto1. (Shooting raw so WB can be auto.)
Space=sRGB. (Keep everything in gamut. Well, R/G/B should always be in gamut anyway.)
ISO=64. (Base setting to minimize noise.)
Picture Control = Flat[0]. (Reduced contrast improves the accuracy of the in-camera histogram.)

White Balance
Average values from the R, G and B shots were entered in the Excel spreadsheet.
In PS Elements 11 a Magenta (225,111,193) file was created for monitor display.
I now made the in-camera WB against the big Magenta screen.
Given that I was using a Nikon, this did not happen on the first try.
But eventually I got the D810 to declare a WB against the Magenta screen as "good".

However --- The Magenta UniWB shot shows RGB coefficients of .92, 1.0, .63.
Red is in the ballpark but slightly low. Blue is very much off.

So I did not quite attain a UniWB on this first attempt.
I know the methodology is sound. (I've hand calculated this stuff in the past.)
Possible errors:

• I'm thinking that the difficulty I had with the D810 in-camra WB likely contributed to the problem?
• And/or maybe there was too much ambient light when I performed the WB against the magenta screen?
• I probably overexposed the R, G and B shots? Not supposed to use ETTR until after setting up UniWB. "-)

[1,2,3] Raw Digger Screen Shots
I used a sample from the brightest area.

[4] Magenta (225,111,193) File for One-step WB
Created in PSElements.

[5] Excel Spreadsheet
Note the Coefficients column has a bad Blue value.

[6] UniWB Shot #1 with Flat[0] pic control.
Although I did not produce unitary RGB coefficients in my first attempt this shot looks very green. :-)

[7] UniWB Shot #2 with Std[4,3,1,0,2,0] for [sharpening,clarity,contrast,brightness,saturation,hue]
Still looking very green.

[Shane]

Possible errors:

    I'm thinking that the difficulty I had with the D810 in-camra WB likely contributed to the problem?
    And/or maybe there was too much ambient light when I performed the WB against the magenta screen?
    I probably overexposed the R, G and B shots? Not supposed to use ETTR until after setting up UniWB. "-)

I found it best to use a fixed F/L lens (note your shots are at 100/78/100mm f/l)
Define your manual exposure characteristics and don't change them between shots.
Watch the ambient light issue.
None of your exposures are O/exp (check your Rawdigger results) if anything they are a little on the low side.
Your Blue value is probably off due to underexposure of the B channel (too much noise).
The first time I used this method my exposures were a little low (similar to yours), increasing the exposure but without saturation will probably solve your problem.

I have notes on the issues somewhere but, as usual, put them somewhere safe! I will see if I can find them.

[Andrea B.]

Thanks, Shane. I did not even see that I shot green at 78mm instead of 100mm. Not sure how I managed to do that!!! "-)

I'll try again later when there is no grey light coming in thru the windows. (We are having horrible hurricaney weather.)

[Shane]

When I build a custom UWB using the monitor I usually do this in the evening with the room lights off. No glare and no reflected colour casts.