How White the Whites???

This thread started in Avian Wild & Free but I have copied most of it here as it turned into a tremendous learning opportunity for me.

later and love to all, artie

While leading my IPTs I go to great lengths to ensure folks wonderful opportunities. On the recently concluded SW FLA Spring IPT, the last ever large (14 folks) group, we carried a bucket and a cast net out to Estero Lagoon on our afternoon trek. I was able to catch a variety of small fish that the Snowy Egrets enjoyed. Then I caught a mullet about 8 inches long. The snowies investigated but decided that it was too big for them. Out of nowhere a gorgeous Great Egret in full breeding plumage flew in and gobbled down the mullet before any of us knew what had happened... Fortunately, it stuck around and posed and posed for the group.

Canon 500mm f/4L IS lens with the EOS-1D MIII. ISO 200. Evaluative metering +2/3 stop: 1/1250 at f/4.

Don't be shy: all comments welcome. Later and love and thanks to Alfred and Fabs and Axel for their help on the IPT

Great-looking specimen, details, and BG. It could go a tad brighter for my taste.

This is what the image looks like with an embedded BPN monitor Cal strip. the whites are too dark, really they are a light gray. This was post was requested by Artie after I brought the dark image tone to Artie's attention. FYI the last square on the strip is not to reference a white tone in an image, the level is 255 so if any white in your image matches that tone it is totally blown
Robert

This is a repost with a more natural white tone and lightened (masked) BG.

Looks better to me, comments welcome.

Reposted with Vignetting correction of +23, +34 in CS3.

Robert

Great repost of Artie's beautiful image. That is one gorgeous great egret!

Two thumbs up for the re-post!

Steve

That sure was a beautiful bird! Great work on the re-post Robert. Thanks Artie for a wonderful IPT, that was full of opportunities like this one. I love the ones I got of him too.

The repost is a nice improvement but it looks a little washed out with the lighter background. How about this version? Levels/curves adjustments with some selective sharpening on the bird's shoulder/neck through an edge mask.

The image itself has a nice painterly feel to it. The subtle color changes in the background are especially nice.

Thanks to Robert and Leroy. Below is my re-post incorporating changes suggested by each. I have been in a "too-dark whites" rut lately. Thanks to all for opening my eyes to that.

I do prefer the background a bit darker than in Robert's version. The brightest whites are now R; 252, G; 240, B: 226.

The major question now is, "Is there detail in the whitest feathers?"

In an e-mail Robert, wondering about the vignetting, asked if the image was made with the 1Ds MIII where the vignetting would be expected. I double-checked: it was created with the 1DIII.

later and love to all, artie

I like Artie's repost best since the plumage doesn't have a blueish cast anymore and there is plenty of details.

looks really nice now to my eye Artie, still seeing details in the whites.
Only found one teeny tiny wee itsy bitsy area in red channel at 255(guess that is on my monitor).....I was laughing and going nuts trying to find after looking at histogram, was a challenge. Definitely the dark bg looks good IMO. This has been a bit of learning experience

Here is an e-mail exchange with Robert O'Toole that many might find educational:

ROT: 255 is pure white, with no detail. Since RGB is additive that means there is zero info there.


AM: Can you explain that. I understand that 255 is pure white = no detail but am confused about the additive part...

ROT: With RGB all the colors add up to make white. With LAB, there is a separate Luminosity, or black-gray-white tone channel. The color info separate. In CMYK is subtractive so no color at all, 0,0,0,0 is pure white.

AM: Should the whites be at 254?

ROT: Not usually. Best would be if they are one square over to the left of the whitest (255) box on the monitor calibration strip, that is, the tone next to pure white. Or, whenever there is detail in the whites.

You can, however, have the whites in the 230-240 range and not have any detail.

AM: That is what always bugs me; is there a way to explain that?

ROT: Yes, sometimes a white area on a bird looks blown, but when you measure the RGB value it is 241,245,242. The whites are not at all technically blown (overexposed) but the white lacks detail.

This has a lot to do with the surface and texture of an object. The softer and flatter the lighit technically is light-light-light gray and not pure white.

AM: That makes sense but I sometimes have whites in the sun, even in nice light, where the whites are at 245 but seem to lack detail... Is there an explanation for that?

Thanks for your help here Roberto!

And later and love, artie

This thread is one of the best I've read on BPN. Thanks, Artie, for showing us that everyone can benefit from the knowledge here. And Robert, I ordered your APTATs CD a couple of days ago, and now I'm more anxious than ever to get it!

Arthur Morris said:

Here is an e-mail exchange with Robert O'Toole that many might find educational:


This has a lot to do with the surface and texture of an object. The softer and flatter the lighit technically is light-light-light gray and not pure white.

AM: That makes sense but I sometimes have whites in the sun, even in nice light, where the whites are at 245 but seem to lack detail... Is there an explanation for that?

Click to expand...

I think in many cases it can be the quality of the light, diffused or direct, and the angle of the light, straight on and flat or angled to accentuate texture and details.
Also the camera itself can be a big factor (in terms or dynamic range), and the between the same cameras the settings can have a big impact (contrast and color space).
The RAW convertor used can have a huge impact. Breezebrowsers is much easier to use and faster in many cases but I dont like the way it handles white tones. I have found ACR (Photoshop's Raw Convertor) handles whites much much better with great detail in the whites.


Robert

Great edit Artie. One of the problems with bright whites (and dark grays for that matter) is that it becomes difficult for our eyes to see subtle differences between shades that are similar in the extreme ends of the color gamut. In the squares above the upper left square is (255,255,255) and the right is (245,245,245). It's easy to see the difference between them with a straight edge. In the bottom rectangle the white forms a gradient from over the same color range. Each block is shown against a dark and light background. In a high contrast scene our eyes have more trouble with fine shade differences. If you look closely you should be able to see vertical bands of gray in the lower rectangle (posterization effect). To my eyes its easier to see the shade differences with a light background (lower dynamic range) than with the dark background.

This optical effect shows up in editing images as well. For example the background colors in lightroom are fairly dark. If you get the image looking good then post it against a bright white background it will look too dark, even though it looked fine with a dark background.

Hi Leroy, I am a bit confused. Can you please clarify the diagram above?

Thanks and later and love, artie

Maybe this is easier to see. The various color blocks are marked. On my monitor I can see vertical bands in the gradient region. You can see this numerically in photoshop if you scroll your pointer through the gradient region. If you can't see where the bands are then open the image in photoshop and move the midpoint slider towards the right in the levels menu.

I can see the bands fairly easily with the light background but it's harder for me to see against the dark background.

I hope that helps explain things better. It's complicated by the fact that all of our monitors or not the same - even when calibrated.

this has been excellent Leroy....thanks for your time.
And thanks to you too Artie

Excellent thread. Thanks to all involved!

Leroy's demonstration is a very good one - our eyes are easily fooled. It also shows a problem with the concept of whites - as Robert alludes to: we consider 255, 255, 255 as "blown" and don't want the "whites" to reach that value - actually none of the RGB values should be 255. Trouble is - there are no "whites" - on the additive RGB scale there is only ONE white = 255, 255, 255. If you change the value to 254, 254, 254 you technically have gray - a very very light gray though. As long as you keep all the value identical, you will always have gray - exactly up to the point 0,0,0 which is black. So if your subject (egret) really is white you as a photographer would exclude his correct color from your consideration but not allowing it to be "white".

Again, technically speaking - change only one of the RGB values away from 255, 255, 255 - so 254, 255, 255, or 255, 254, 255, or 255, 255, 254 - then you are not talking white anymore but very light green, violet, or brown, respectively. This, of course, is a totally technical exercise - as Leroy pointed out, our eyes can not easily distinguish those small differences but will perceive a whole range of values as "whites" though they technically are light color "casts".

All the above does not even account for "white balance" - unless your light is exactly of one color temperature - let's say 5500K - any deviation from it will technically also register as "color cast".

This technical discussion is moot, however. As photographers, we are talking about detail or lack thereof - an area that is within a certain range of RGB values looks uniform to the eye - and we call this lacking detail. Whether or not this is because the subject really has no detail in that area or whether we already exposed the image incorrectly and "blew" it (the highlights that is).

We all know that to expose a mostly "white" subject correctly, we need to use some positive exposure compensation (Editor's note to avoid confusion: this is always true when you are spot metering but only sometimes true when using Evaluative or Matrix metering... Be careful here...) - and the reason for that is that your camera's exposure meter is calibrated for some "medium" gray, i.e. it assumes that your subject is reflecting somewhere between 12 and 18% of the light that hits it - without that exposure compensation, the egret would look very much gray. (Editor's note again: the situation is much more complex than Dieter is making it--he is here again referring only to spot metering. I would recommend that everyone check out my e-Zine article on Exposure as well as my additional comments in this Forum.)

If the egret was pure white - then we would have 100% reflection and hence a value of 255, 255, 255.

At this point, it is up to the photographer's experience to choose an exposure that attempts to match the percentage of light reflected from the subject - which could be measured correctly with an incident light meter (and in this day and age of digital cameras also get the "white" balance right) . Trouble is, for this to function properly, one would need to point the meter in the direction of the camera at the exact position of the subject - so hand that meter to the egret or make sure the egret doesn't mind that you get very very close to it.

(While Dieter is raising many excellent points here he is treading in dangerous waters; if you took a good incident meter reading while photographing a brilliant white bird you would need to stop down one full stop to maintain detail in the whites. This is John Shaw's Sunny f/22 for Whites....

The histogram is of great help here - but there are problems there as well. First off, it works of an 8-bit JPEG with all your camera settings applied - but you are shooting 12-bit or even 14-bit RAW (where white is either 4096, 4096, 4096 or 16384, 16384, 16384, respectively). So what we really try to achieve is not to have our exposure chosen in a way that the values "bunch up" against the right side of the histogram - which unless the subject reflects that much light would be interpreted as an overexposure = blown out highlights.

To sum this up, there is nothing inherently wrong with a value of 255, 255, 255 in your image - IF you are certain that there is an area in the subject that reflects all the light that hits it. Under those circumstances, there would be nothing wrong with the values "bunching up" on the right side of the histogram either. In practice though, it is "safer", to avoid that scenario, as it will - correct or not - generally be interpreted as an overexposure.

As Leroy pointed out - the eye has trouble distinguishing those fine changes in the "whites" and as Robert pointed out - detail or lack thereof has much more to do with quality of light and the light incident angle on the subject.