Announcement

Collapse
No announcement yet.

Huston, we have a problem : Z. deep. EXR

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Huston, we have a problem : Z. deep. EXR

    Hi there Vray guru`s!
    I stuck with childish (at first glance) question - how to get Z-pass ?

    But the devils as always in details .. I need depth pass stored in exr.with deep data.
    So, I`ve setup two passes: default vray ZDepth and my custom Z via "sampler info" .
    Even more I`ve changed default vray ZDepth pas so, that in theory it should be identical to "sampler info" z-pass..

    But as a result I got some really-really-really strange looking depth-pass.

    Actually I did pretty much many different tests.. and cant find slution.. so decide to write here and hope for yur help guys.
    While my investigation I fund that vray shows me correct Z values in his frame buffer dialog! and if I save exr from framebuffer window, I got gorrect Z data.
    So, after this i thought that something wrong is going on the final stage while saving\converting vrayimg to exr file.... just thoughts...

    And one more thing - If I use deepToImage tool in Nuke to obtain depth from this exr, I got z-deph pass without these alien-circles.. but, the values it shows me is incorrect, I don`t understand what they actually are?... In that case you can use zDepth from sampler Info (blue channel) as a reference.. and there you can see that correct values for closest point on my floor-object is something like 27+-
    but depth extracted from deep shows me values 0.038+-

    You can find my resulted exr and Maya scene in links below:
    1) Maya test scene
    2) Result deep EXR image



    Thank you for your attention.

    -serge.



    PS: and yes, I did it all in 32 bit float

  • #2
    Hi.

    As far I can see from your pic you are trying to get the Z-depth from an exr deep format file. You should get the depth Z pass from the exr multichannel. As for the deep workflow, you should try to separate your scene one for traditional AOVs ( using exr multichannel ) and another for the deep information ( exr deep )and test it on Nuke or Fusion. I hope this help.

    Alvaro

    Comment


    • #3
      Originally posted by Alvaro Letelier View Post
      Hi.

      As far I can see from your pic you are trying to get the Z-depth from an exr deep format file. You should get the depth Z pass from the exr multichannel. As for the deep workflow, you should try to separate your scene one for traditional AOVs ( using exr multichannel ) and another for the deep information ( exr deep )and test it on Nuke or Fusion. I hope this help.

      Alvaro
      Thank you for reply Alvaro.

      Speaking your assumptions - yes, I did this already.. and in last part of my message I am wrote exact about this situation - the ZDepth extracted from deep data (via DeepToImage tool in Nuke) have some meaningless (for me at least) values compared to real, correct values which yu can find in my custom Z pass constructed by samplerInfo.. you can find it inside my exr example in attachment.

      If we will find a way to convert deep ZDepth to ZDepth from SamplerInfo.blue channel (which show me real distance from camera to points in Maya units) - that would be enough for me.
      Last edited by namaissur; 13-11-2018, 09:34 AM.

      Comment


      • #4
        Hey,

        I think you need to set deep fragment method by render ID, I couldn't get any deep with fragment method by zdepth. After that the zdepth look correct to me in nuke.
        Attached Files
        Dmitry Vinnik
        Silhouette Images Inc.
        ShowReel:
        https://www.youtube.com/watch?v=qxSJlvSwAhA
        https://www.linkedin.com/in/dmitry-v...-identity-name

        Comment


        • #5
          Originally posted by Morbid Angel View Post
          Hey,

          I think you need to set deep fragment method by render ID, I couldn't get any deep with fragment method by zdepth. After that the zdepth look correct to me in nuke.
          Wow! I definitely should try this one!
          Thank you Dmitry!

          Comment

          Working...
          X