Announcement

Collapse
No announcement yet.

Option for double resolution in ZDepth Render Element

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

  • #16
    If you use Nuke to composite, don't use each pass' individual alpha. Use the alpha pass for it in the "end" of the nodes.
    This has worked for me for getting rid of that bad and jaggy edge in animations

    Tried all sorts of defringing before that with bad results

    Comment


    • #17
      Did this one get lost in the pile?

      I just bought RPManager, so it shouldn't be too much of a hassle for me any longer. But it would still be a nice option to have directly in VRay if possible. Basically just reading the x and y resolution and doubling them - as a checkbox under the VRay Render Element.
      http://henrikbclausen.com

      Comment


      • #18
        It is not as straightforward to implement this; I have noted it down and we will think about it, but it will take some time.

        Best regards,
        Vlado
        I only act like I know everything, Rogers.

        Comment


        • #19
          Good to know.

          How you even manage to get anything done on top of reading + replying to posts and complaints in here is beyond me, but I'm glad it seems to be working out.
          http://henrikbclausen.com

          Comment


          • #20
            I'm pretty sure this could be solved by using an erode filter on your depth element, well in nuke anyhow. Not sure what the equivalent AF filter would be. You objects in the FG that are floating in space need to be contained within the depth map, it won't work every time, but I have had a lot of success using this method.
            Drew Wood-Davies
            Head Of Lighting & Look-Development | Method Studios | Melbourne, Australia

            Comment


            • #21
              +1 for this option too !!! congrat HenrikBC for your short !!!

              Comment


              • #22
                I've been experimenting with dof in post and came across this thread. I understand rendering it at double resolution, then applying the dof in post, then scaling it 50%. With antialiasing though... I know to turn off the antialiasing filter, but should I also set the image sampler to 'Fixed' = 1? When I do that, thin lines get broken and jump around in the animation. Anyone have any advice?

                Comment


                • #23
                  titane: Thanks!

                  rfellers:
                  You will generally have trouble with very fine details. In my experience post dof works best when you have objects placed gradually into the distance - opposed to a shot of some grass in the fg and telephone poles in the far bg (horror!). In that silly example I would render two seperate passes out and apply an out of focus effect to each instead.

                  Does the Image Sampler change anything when all you're rendering out is a z-pass? I've used the scanline renderer for most of my z-passes actually, to save speed.
                  http://henrikbclausen.com

                  Comment


                  • #24
                    So you do set the image sampler to 'Fixed' = 1?

                    Comment


                    • #25
                      I don't even touch it - just apply a global override material which is white (or whatever) and disable shadows before rendering out the z-depth pass.
                      http://henrikbclausen.com

                      Comment


                      • #26
                        Originally posted by rfellers View Post
                        So you do set the image sampler to 'Fixed' = 1?
                        Yes....especially if you have opacity maps such as leaves otherwise it won't cut it out properly

                        NOTE* Turn off anti-aliasing FILTERs as well as turning OFF Filtering for your bitmap that has opacity [Your opacity map should also ONLY BE BLACK & WHITE & not anti-aliased & turn OFF bitmap filter on those types of materials with opacity

                        Hope this helps

                        Comment


                        • #27
                          Originally posted by HenrikBC View Post
                          I don't even touch it - just apply a global override material which is white (or whatever) and disable shadows before rendering out the z-depth pass.
                          Don't even touch it, meaning you leave it at what? Also, I guess I'm not clear on your workflow. If the z-depth and the color pass are suppose to line up exactly, why not just render it once at double res w/ z-depth channel? It sounds like you render the color pass at 100% res antialiased (then scale it up in post), then render the z-depth at 200% res non-antialiased. If that is what you are doing, it doesn't sound like the edges are going to line up exactly.

                          Originally posted by 3DMK View Post
                          Yes....especially if you have opacity maps such as leaves otherwise it won't cut it out properly

                          NOTE* Turn off anti-aliasing FILTERs as well as turning OFF Filtering for your bitmap that has opacity [Your opacity map should also ONLY BE BLACK & WHITE & not anti-aliased & turn OFF bitmap filter on those types of materials with opacity

                          Hope this helps
                          Thanks - that does help.

                          Comment


                          • #28
                            Originally posted by rfellers View Post
                            It sounds like you render the color pass at 100% res antialiased (then scale it up in post), then render the z-depth at 200% res non-antialiased. If that is what you are doing, it doesn't sound like the edges are going to line up exactly.
                            That's exactly what I do. The reason for not rendering everything out at 200% is speed. And even if you did that, everything would still not line up _perfectly_ because the z-buffer lacks AA. On all the projects where I've applied post dof to a 3d rendered shot - this is the way I've always done it, and been quite content with the result so far.

                            I have never had a shot where I needed opacity mapped features and no AA - so maybe that's why I haven't worried too much about the bitmap filtering.

                            Furthermore; On most of my projects, I have made the z-pass by creating an omni light, setting it to ambient, adjusted the far attenuation of the light, linked it to my camera (usually), turned off all other lights in the scene, made a 255 white material and applied it to everything in the scene/global material override using vray and then rendered it out without AA.
                            http://henrikbclausen.com

                            Comment

                            Working...
                            X