4 Replies Latest reply on Aug 20, 2013 1:51 PM by mncrnich_sarofsky

    Reading ProRes4444 Quicktime between AE CS5.5 & Nuke 7.0v8 produces different color results

    mncrnich_sarofsky

      Please read this first.  THIS IS NOT A GAMMA ISSUE!  I am well aware of the difference between RGB v YUV & 0-255 v 16-235.  It is not a RGB v YUV problem or a 0-255 v 16-235.

       

      I am experiencing a noticeable color shift within highly saturated files.  I am specifically noticing a difference in REDs.  At least witin my current project.

       

      Files are Alexa C-Log.  Edited within FCP.  Graded by BaseLight within FCP.  Comped within NUKE.  Titles added within AE.  Sent back to FCP timeline. 

       

      Here is what I cannot explain.  I can roundtrip the file within Nuke.  I can Read a ProRes4444 file which I can then render out as multiple formats, i.e.  QT ProRes4444, EXR, TIFF, DPX.  I can bring those files back in and create a difference matte which produces pure black (well close enough, small differences in gamma when elevated astronomically).  So apples to apples, color space matches, gamma matches.

       

      Now when I bring these rendered files into AE the image sequences display a different color than the ProRes4444 QTs.  ALL the image sequences match and ALL the ProRes4444 match.  I have tried creating a project with an unmanaged color space as well as a traditional rec709 project color space.  I have also tried intrepreting the PR4444 files different color profiles.  NO amount of variation will allow the PR4444 files to match the image sequences.  I am partial to image sequences, BUT, ProRes files have become standard file types and I would like to resolve this issue.

       

      Any thoughts?

       

      I have tried this:

       

      http://blogs.adobe.com/aftereffects/2009/12/prores-4444-colors-and-gamma-s.html

       

      and this:

       

      http://blogs.adobe.com/aftereffects/2010/05/prores-4444-and-prores-422-in.html

       

      Please observe . . . .

       

      Sh510_ProRes4444_QT_Color.jpgSh510_ImageSequence_Color.jpg

       

      TIA

       

      -M

        • 1. Re: Reading ProRes4444 Quicktime between AE CS5.5 & Nuke 7.0v8 produces different color results
          mncrnich_sarofsky Level 1

          Okay, so after some exhaustive testing, I have determined the issue is stemming from Quicktimes delightful "Enable Final Cut Studio Color Compatibility".  If I toggle this switch, it changes the color exactly the same as the differential I am seeing between AE & Nuke.  Clearly one application is recognizing this flag whereas the other application is not.  I HATE APPLE & QUICKTIMES!  Arg.

           

          It is not a gamma issue, this is something to do with Apple's ColorSync.  With that said, does anyone have additional information on ColorSync and whether AE uses this toggle switch?  This is maddening.

           

          Hey Adobe guys you will be happy to know this further pushes me into switching our facility to Premiere Pro. 

           

          Any information or links would be greatly appreciated.

           

          TIA

           

          M

          • 2. Re: Reading ProRes4444 Quicktime between AE CS5.5 & Nuke 7.0v8 produces different color results
            mncrnich_sarofsky Level 1

            I have more results after further testing. 

             

            Running OS X 10.8.4

            AE 10.5.0.253

            K5000 GPU

             

            My AE project is setup as a 16bpc, HDTV (rec709), Blend Colors Using 1.0 Gamma, Match Legacy After Effects QuickTime Gamma Adjustments, Compensate for Scene-referred Profiles.

             

            I import a FCP 7 generated QT ProRes4444 file.  The file is rendered as YUV Full Precision with Super-Whites.  In an attempt to round trip the file, I rendered the singular file with no effects back out as a ProRes4444 file with rec709, sRGB or Preserve RGB color profiles.  Once imported back into AE, each of these 3 files matches the original identically.  Quite well actually, even when I turn on difference mode and crank the gamma, they hold up really well.  Nothing perceptually different here.

             

            NOW, when I render out as DPX, EXR or TIFF, I get a completely different result.  EACH image sequence imports identically, so DPX matches EXR matches TIFF.  BUT they are all identically different than the original.  Once again there is a color shift.  NOT a gamma shift, a HUE shift.

             

            I am rendering each file as a rec709 and importing as a rec709. 

             

            What gives?

             

            Here is the kicker, when I switch to my windows box (WIN7 SP1), then open my AEP file.  The colors match identically across all files & formats!!!  I literally set my OS X AE comp to solo 2 layers on difference mode.  There is a clear difference.  When I open the same file within WIN7, pure BLACK.  NO DIFFERENCE.  Holy crap this is alarming.  Side note, still have the hue shift within NUKE across platforms but it is consistent.

             

            WIN7 64bit SP1

            AE 10.5.0.253

            GTX680 GPU

             

            I am not going to blame Adobe, since I am almost certain this is an APPLE issue.  I would like to ask the Adobe AE Team, what value is AE reading that Nuke is not?  Is there different QT importers?

             

            Thanks,

             

            Matt

            • 3. Re: Reading ProRes4444 Quicktime between AE CS5.5 & Nuke 7.0v8 produces different color results
              fnordware Level 3

              My advice is to avoid QuickTime, especially with Nuke. Pick a way to convert QuickTime to something like a DPX sequence and then use that for all your production, converting it back to whatever format you need at the end.

               

              Yes, this is sort of a drag and will take up a lot more disk space, but it's the only way to be sure that all programs will read your source material the same way.

              • 4. Re: Reading ProRes4444 Quicktime between AE CS5.5 & Nuke 7.0v8 produces different color results
                mncrnich_sarofsky Level 1

                Thanks for the input.  I am not overly concerned about the storage required for image sequences v. QTs.  I do agree an image sequence based workflow allows better control.

                 

                With that said, I am surprised someone else hasn't had this issue?  Rendering a QT file to a image sequence within AE causes this color shift.  SO even attempting to begin an image based workflow breaks at least using AE to generated the image sequences. 

                 

                Anyway, I think in an ideal situation I would completely remove FCP from our pipeline.  Going to start testing out Premeire Pro.  Do I dare go back to Smoke?

                 

                I am going to run a few more tests.

                 

                -M