Luminosity Issues

[orenh88692336]

There are several ways to give all pixels in an image an identical luminosity value, and each method has issues. I found the most precise one, which works perfectly, is doing it in Lab. However, when converting it to sRGB (to create a jpg file), the apparent image looks different, as it creates variations in colors and contrast, as well as some variations in luminosity. So, I get (and found some sources) there are issues in computing and converting saturation and luminosity in such different color spaces.

However, it would have seemed as though the simplest imaginable method of doing it in the sRGB color space itself would be efficient as well. This is by putting a gray layer on top in using the luminosity blend mode. This should have given a fixed luminosity value to all pixels in the image below. That does not happen, however.

You may argue that the way I measure luminosity, by using the Lab’s L, is in a different color space, so I run again into transitions and calculations problems, and that nothing is wrong with the Gray-Luminosity action. This however is incorrect.

After transferring an image from Lab to sRGB, I can assess luminosity variations by creating an RGB mask (i.e., by selecting the RGB channel). If I do this to the Lab-sRGB conversion I do get some variations that I can see in the RGB Mask. However, if I put a Gray-Luminosity layer on top, and Stamp the image, and repeat the creation of an RGB mask, the variations in it actually increase. What is going on? It goes much beyond rounding errors. Is it a Photoshop bug?

[TheDigitalDog]

Best to ask this where Adobe engineers hang out: You should go here and comment after selecting the appropriate product (Photoshop):
https://feedback.photoshop.com/categories/products/5f5f2090785c1f1e6cc40864

[orenh88692336]

Thank you thedigitaldog, First answer by Photoshop help was that there might be issues with my ICM file (i.e. the color management profile). I doubt if this is true, because it means that Photoshop worries about visual appearance on my monitor more than about precision (i.e. numbers and values of RGB or Lab), which does not make much sense, but I'll try that. What it seems to mean, as far as I understand it, which is easy to test, is that  different color profiles will give me different RGB masks, or at least different calculations of Luminosity values. Really? - I'll find out.

[D Fosse]

There is a lot of relativity here. All numbers are relative to color space - sRGB, Adobe RGB etc. The visually same color/lightness will yield different numbers.

In addition, any grayscale rendering, such as channels, masks, luminosity values, are relative to your working gray. In these cases, there will be no embedded profile to determine the numbers, so the working space is used.

Photoshop does "worry" about the visual appearance on-screen, in the sense that all data are converted into your monitor profile for display on-screen. The converted numbers are sent to screen, not the original numbers (the definition of a color managed display pipeline). But that shouldn't have any relevance in this case; that's just for correct screen appearance.

Other than that I must confess I don't really understand what you're trying to achieve.

[orenh88692336]

Thanks D Foss,

It took me a few days to return to that. I first did what PS help suggested, which is building a new color profile with a starting point of sRGB.

It did not matter, of course, and I knew that, because even remote profiles did not change the RGB mask (or the L values), which means that PS does use numbers to output luminosity, and the values are those of the original file, not those translated to my monitor, i.e., not of appearnce.

I am trying to make a very simple thing - getting an image with no variance what so ever, in luminosity. I.e., that after using a gray layer on top, in Luminosity blending mode, all pixels will have the same gray, namely, that of the gray layer, and will have no variations in L and uniformity in a mask created from slected luminosity (i.e., from RGB channel).

As far as I understand, gray layer in luminosity should have done that, and it doesn't do a precise job.