7 Replies Latest reply on Jun 4, 2014 5:15 AM by D Fosse

    64/16/double bit Adjustment layers?

    kentconf

      Hi, I have a 64 bit imported RAW, CR2. I have recently got in the habit of using multiple curves adjustment layers.

      Indeed, when I apply a Curves adjustment by Ctrl-M, the histogram stays nice and smooth (as 16 bit implies), but if I use even two adjustment layers, stratification occurs- there is a reduced number of colours, resulting in zeros in the histogram followed by instant peaks.

       

      Is there a way to enable high bitdepth adjustment layers? I used to have CC, and I think I noticed that it wasn't even fixed there... I went back to CS6, thought I'd save my money...

        • 1. Re: 64/16/double bit Adjustment layers?
          Mylenium Most Valuable Participant

          Not sure what you are getting at. Even 16bit color space isn't unlimited and depending on how discrete values overlap/ coincide quantization can occur. Only working in 32bit can further expand that range. That aside it's really not clear how you work and what adjustments you apply. You would have to provide screenshots. Furthermore the histogram may not necessarily be an exact representation since it's calculated on the fly. Plus of course we really don't know anything about your RAW settings, color management, hardware acceleration settings and so on. This is really much more complex than just saying that the histogram looks crooked.

           

          Mylenium

          • 2. Re: 64/16/double bit Adjustment layers?
            kentconf Level 1

            Perhaps someone helpful can answer it. By the way, your display picture is crooked.

            (I mean all due respect, but after giving a 'get out of here' line, you demonstrated that you understood the question. Is that the kind of thing you get a + next to your name for around here?)

             

            If you don't know the answer, just refrain from posting. In the interests of defending myself I will deconstruct your post:

             

            Not sure what you are getting at. Even 16bit color space isn't unlimited (correct) and depending on how discrete values overlap/ coincide quantization can occur (correct). Only working in 32bit can further expand that range (to the point where most of the time it is imperceptible). That aside it's really not clear how you work and what adjustments you apply (sigh, nor is it relevant). You would have to provide screenshots (because you say so). Furthermore the histogram may not necessarily be an exact representation since it's calculated on the fly(interesting, but beside the point). Plus of course we really don't know anything about your RAW settings, color management, hardware acceleration settings and so on. This is really much more complex(no, and I should know, I have an Engineering degree, plus more than a decade of coding experience) than just saying that the histogram looks crooked. (nice analogy =) )


            I've highlighted some words which people trip over online. It's good manners to assume that your readers are intelligent, so if you say something like "depending" you are basically conceding the point. As I noted, that sentence is correct. But (and this should have been obvious) I'm not talking about one of those 'extreme' cases.


            Sorry for any offense. Maybe you were just having a bad day, or needed a coffee...

            • 3. Re: 64/16/double bit Adjustment layers?
              Mylenium Most Valuable Participant

              Well, but you still haven't told us anything that would help us to explain the matter and advise, didn't you? We can stand here all day and try to be smart-(assy) about one another, but that's rather unproductive, don't you think? You could still have some weird interaction of different adjustments, an issue with hardware acceleration, a messy color profile and whatnot - even if this is some weird bug.... The magic sauce is all in the details, you know. As a programmer you should know that...

               

              Mylenium

              • 4. Re: 64/16/double bit Adjustment layers?
                D Fosse Adobe Community Professional & MVP

                It's the histogram caching. Click on the refresh symbol or the warning triangle (either one), and you'll get an updated accurate histogram:

                 

                histogram.png

                • 5. Re: 64/16/double bit Adjustment layers?
                  kentconf Level 1

                  Yeah but I'm not some rogue bit of code,  I'm a typical user, using a tool as  it was meant to be used, even an experienced user not trying anything fancy. "Standard assumptions hold."

                   

                  So anyway you realise that I am talking about a bit depth problem, and it can be seen when the values are stratified, giving zeros in the histogram, after you apply a single Curves adjustment layer. With a 64 bit image this should not be the case- indeed, if I use the proper "Curves..." tool from the menu, my histogram stays smooth. For any perceived number of times I apply it! You see, for stratification to occur in an (assumed perfect) histogram now would require it to be more than 65536 pixels wide. So even if I apply it multiple times, quantisation will not be visible in a  small histogram.

                   

                  But the real issue is what you can see. You will probably get away with a single 32bit adjustment layer producing a histogram with zeros... Maybe not for some printed images. But if you have a few adjustment layers with different layer masks, then you see my problem. Instead of skipping 1 value, you are working with the rounded/floored result a number of times, and you might skip as many single values as you have adjustment layers, or more!

                   

                  thanks for the help

                  • 6. Re: 64/16/double bit Adjustment layers?
                    kentconf Level 1

                    Thanks twenty_one! That works great! So as Mylenium got at, it's because the histogram isn't technically correct. Although I don't know why I didn't see that ! before...

                    So that's great (and this typing interface that Adobe uses is good, but it doesn't work on Chrome Android)

                    • 7. Re: 64/16/double bit Adjustment layers?
                      D Fosse Adobe Community Professional & MVP

                      Well, it is technically correct, but only when it's refreshed. Until then it's a low-precision cached version.

                       

                      There's probably a technical explanation for why it doesn't auto-refresh.