Announcement

Collapse
No announcement yet.

few (many) questions

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

  • few (many) questions

    - how to clean those sparse white pixels?

    - what means "single kernel"?

    - does Vray RT GPU use Metropolis light transport method?

    - any plans to implement caustics?

    - any plans to implement diffractions?

    - fog in refraction?

    - sky portals?

    - proxies?

    - procedural noise, ramp and tiles?

    - infinite plane?


    Best regards

  • #2
    as i promised..... same questions again and again... vlado... please make a "sticky" thread where you answer the general questions....
    Dual Xeon E5-2699 v43, ASUS Gforce RTX 2080S, Samsung M.2 SSD,
    www.robostudio.swiss/portfolio
    mr@robostudio.swiss

    Comment


    • #3
      Originally posted by saudade View Post
      - how to clean those sparse white pixels?
      Depends on what is causing them, but normally you just let it render for longer until they become unnoticeable. V-Ray does adaptive sampling so it is able to concentrate more samples in these problem areas.

      - what means "single kernel"?
      It means that the entire renderer is packed into a single OpenCL program for the GPU.

      - does Vray RT GPU use Metropolis light transport method?
      No. While Metropolis light transport is useful is some rare cases, it does not perform well for most of the common lighting scenarios.

      - any plans to implement caustics?
      GI caustics are already implemented and work quite well.

      - any plans to implement diffractions?
      Not in the initial version, but at some point they will be there.

      - fog in refraction?
      Yes, in fact this is almost done now.

      - sky portals?
      Yes.

      - proxies?
      Yes, although initially they will not save any memory like they do in the CPU version.

      - procedural noise, ramp and tiles?
      Not in the first version.

      - infinite plane?
      Not sure about this one.

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

      Comment


      • #4
        many thanks Vlado, precious answers.

        is it possible to deactivate the gpu computing of the master workstation that launches the rt render?
        I ask for that because many of us have a single pci-x motherboard. (in my case I own 7 Intel Vernonia dual xeon)
        Using my Quadro CX on the master workstation I would like to have this graphic board idle and use a networked client with 3 or 4 Tesla or mount them on different renderfarm slaves.

        Best regards

        Comment


        • #5
          Don't know if this would work or not, but have you tried un-checking Auto-start Local Slave in the RT Render Servers dialog? Of course you'd do this after adding the IP of the server with your Teslas. Would this work Vlado?

          -Alan

          Comment


          • #6
            Yes, it will work.

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

            Comment


            • #7
              Originally posted by Alan Iglesias View Post
              Don't know if this would work or not, but have you tried un-checking Auto-start Local Slave in the RT Render Servers dialog? Of course you'd do this after adding the IP of the server with your Teslas. Would this work Vlado?

              -Alan
              hmm, where do we bring up this dialog?

              also, i've just installed RT-GPU n the first thing i noticed was that it is actualy RT GPU+CPU, is this correct?
              if so, i was thinkin' of assigning the RT tasks to just the GPU.. can i do this? (obviously all i can think of is Windows' Set Affinity option)..

              how do i toggle that information about the RT samples in the viewport?

              finally just a couple initial bugs:
              (probably has been posted a million times) when i change an object's color RT wont update, i'd have to restart it..
              also i have a light in my scene that is set to Include only 1 object (sort of a dome i use to wrap the sky around), but regardless of this the light seems to be affecting all the objects in the scene.. also, this pano-image im using for the sky wont show on RT (its a .tif)..

              Comment


              • #8
                Originally posted by vlado View Post
                Yes, it will work.

                Best regards,
                Vlado
                Excellent! Thanks, Vlado.

                Originally posted by FJ View Post
                hmm, where do we bring up this dialog?)..
                After making sure Vray RT is you current Activeshade renderer, and Activeshade is enabled in the Render Setup dialog (lower left-hand corner), go to the the V-Ray RT tab and locate the Rendering section. There you can check Distributed and then click on the Render Servers button to bring up the dialog. In this dialog you can now uncheck Auto-start Local Slave (bottom left), which should take your workstation out of the RT rendering. Now designate the IP number for the machine with your Teslas in it. Make sure you run on that machine the applet that designates these as OpenCL devices (Start/All Programs/ChaosGroup/Vray RT/Select OpenCL Devices for Vray) and you should be good to go.

                Originally posted by FJ View Post
                how do i toggle that information about the RT samples in the viewport?)..
                That checkbox is also in the Vray RT Tab in the Render Setup dialog: Performance/Show Statistics

                -Alan

                Comment


                • #9
                  Oh Yeah...

                  Originally posted by FJ View Post
                  ...also, i've just installed RT-GPU n the first thing i noticed was that it is actualy RT GPU+CPU, is this correct?
                  Well, yes. Generally speaking, Vray RT gives you the option of using either CPU or GPU. There is a way to bring both CPU and GPU together for a rendering but it is very experimental.

                  Originally posted by FJ View Post
                  ...if so, i was thinkin' of assigning the RT tasks to just the GPU.. can i do this?
                  Yes, using either CPU or GPU is currently the default configuration of the RT program.

                  -Alan

                  Comment

                  Working...
                  X