Limits of Exposure Composition

[Saorsa]

Brian,

This is going way off the track of the OP, but I don't think this is correct. If I am wrong, hopefully someone who knows the math of photoshop will chime in. Yes, you can use the eyedropper tools to select pixels for setting the black and white points, and changing the black point can change colors. However, assuming you are using the default levels tool, then using the eyedropper is just using that location to select a luminance level. It does not tell photoshop to separately adjust that color channel. I just tested this by taking a landscape with a lot of blues (water) and greens (trees) and using the eyedropper to set a fairly aggressive black point, first using a blue pixel and then a green. I then turned on the histogram window and turned the levels adjustment layer on and off. You can see the three histograms for all the three color channels shifting sharply to the left when the adjustment is turned on. And the entire image darkens accordingly.

Dan

In a former life I did software for Digital Signal Processing. Reading this Wikipedia page on JPEG reminded me of old headaches.

Changes really are applied across the image. Here is one paragraph of many.

Color space transformation

First, the image should be converted from RGB into a different color space called Y′CBCR (or, informally, YCbCr). It has three components Y', CB and CR: the Y' component represents the brightness of a pixel, and the CB and CR components represent the chrominance (split into blue and red components). This is basically the same color space as used by digital color television as well as digital video including video DVDs, and is similar to the way color is represented in analog PAL video and MAC (but not by analog NTSC, which uses the YIQ color space). The Y′CBCR color space conversion allows greater compression without a significant effect on perceptual image quality (or greater perceptual image quality for the same compression). The compression is more efficient because the brightness information, which is more important to the eventual perceptual quality of the image, is confined to a single channel. This more closely corresponds to the perception of color in the human visual system. The color transformation also improves compression by statistical decorrelation.

This is a picture I'll use for a demo. It has been processed already but it really doesn't matter. This is the starting jpeg of a local restaurant with a neat paint and lighting scheme.



Capture NX shows me that the three lights and TV screen over the bar all present as 255,255,255 basically pure white. If I make one of those my 'black' spot the entire image turns black. There is no detail anywhere, just black so I won't show that one.

Now,the wall to the right of the TV between the lights presents as orange which would be some mix of Red and Green. If I call that Black look what happens.



The major impact is on the green channel which practically disappears except for a bit on the left side. The orange of the lights to the left of the TV becames very very red.

Now, let's take a primarily Red area. In this case on the left side of the bar is a set of shelves with wine bottles. The center one is empty and the right side presents as a shade of red. If I make that the 'black' spot this is what I get.



There is still a bit of red left but all that orange and yellow has become very green.

The paragraph I quoted above is basically how luminance is covered and the wiki article explains it in more detail than I want to know. The Wiki is a good general description. The real challenge in developing this sort of software is dealing with all the various manifestations of information from internal storage to hologram to hardcopy. I am happy to be retired now long enough that the technology has changed enough that nobody wants me back

[DanK]

Brian,

Thanks. Very interesting. Very different from what I saw in my one test, but that was using photoshop. I'll try a few additional tests. I wonder if the two software packages handle this differently? Or maybe I just misinterpreted the test I did.

Dan

[Saorsa]

Brian,

Thanks. Very interesting. Very different from what I saw in my one test, but that was using photoshop. I'll try a few additional tests. I wonder if the two software packages handle this differently? Or maybe I just misinterpreted the test I did.

Dan

Capture NX2 was based on Nik software when Nikon was invested in the company. I don't know PS at all. My only exposure was using PS Elements 2.0 for spot removal for years. Since Nikon has dropped CNX2 I decided to upgrade Elements and add the free NIK suite.

NikVIVEZA works on picking a location and working on a defined area within the image and it automatically applies gradients to 'feather' the changes into the overall image. In this way it builds graduated masks for a particular image applied in a particular order. In CNX2 the changes were embedded in the NEF file, whatever PS does seems to be kept in a sidecar file.

I have no idea what the internal processing is like and am not particularly interested in learning it for the sake of learning. I don't believe that knowing the details of the construction of a tool are necessary to it's craftsmanlike application.

Not many farmers could tell you how to forge a shovel but they can sure tell you what's wrong with one.

[DanK]

Not many farmers could tell you how to forge a shovel but they can sure tell you what's wrong with one.

Right, but your tests and mine suggest that the tools function very differently in the two pieces of software, and that is something a craftsman would want to know regardless. I do in fact enjoy learning the math up to a point, but even apart from that, I want to know if I am misunderstanding how basic tools work in the software I use.

[Saorsa]

OBTW, here is what I finally did with the bar photo.