Announcement

Collapse
No announcement yet.

Render Gurus: validate my test process

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

    Render Gurus: validate my test process

    The shared library of textures that comes with Revit includes three different versions of every texture, at 256 pixels, 512 pixels and 1024 pixels. I believe that Revit only actually uses two of those, but that it will use the less dense version when the camera is further from the surface being rendered, and thus speed rendering.

    My thought to validate this is to make a rather large wall, and place one camera that sits well back from the wall, but just barely misses the edges of the wall. Then create another camera of exactly the same final image size, but that is only 6" from the wall. In effect I have two cameras that show the wall full frame, the only difference being the distance.

    Now I should be able to render both using a material with all three sizes, and also render with a material that has only the largest size. If Revit is rendering like I think, there should be some render time delta in there, likely the distant camera will render faster with a low res texture available.

    The goal is to validate the effort required to make multiple versions of our custom materials. Or perhaps validate that everything should be 512 pixels if the render time difference isn't extreme.

    Anyone see an obvious flaw with this approach? Or done something like this already and can share some results?

    Thanks!
    Gordon
    Pragmatic Praxis

    #2
    Why not just make the large medium and small images different solid colors and see if it renders the different colors instead of timing it?
    .Carl - Architect, BIM Manager, Father, Husband, Coach, Player, Disc Tosser, Driver... not necessarily in that order.

    Free Revit Chat | Cre8iveThings Blog | Live Architecture! | Past Live Architecture!

    Comment


      #3
      Ah, good thought. Then I can ID which sizes Revit uses, and still go back and time to validate if it is worth making multiple versions of real textures. That is my main goal. If the difference is negligible, then it may not be worth the effort to make multiple textures. But if it is a big difference, then it would be worth it. I suspect I will need to time with a very complex pattern, like a multicolored brick, where having look good close up is important, and complex enough to slow a rendering down if anything is going to.

      Thanks Carl!
      Gordon
      Pragmatic Praxis

      Comment


        #4
        Map size in Materials

        I'd stay with the smaller maps in general.

        Like you mentioned, if you are doing a small, close-up shot where the camera is very close to the model, and a detailed map is required, then consider using the larger map size.

        It depends on a lot of things:

        -Size/scope/complexity of the model/scene
        -What is desired output--i.e. size in pixels of the deliverables?
        -Still shot, or animation, or both?
        -Lighting? Daylight or artificial or combination?

        cheers
        Cliff B. Collins
        Registered Architect
        The Lamar Johnson Collaborative Architects, St. Louis, MO
        Autodesk Expert Elite

        Comment


          #5
          Any progress on this? I started doing some tests but can't seem to get revit to use anything but the material I select. I also tried to place some of my own materials in the library structure but man Revit is just uppity about what is placed in that folder! Have you tried editing the images in those folders? Gimp shows the updated image, but windows does not recognize it at all! I'm pretty sure one of those FFBX dlls is a hook that caches the material library folder. Kinda lame if you ask me.
          .Carl - Architect, BIM Manager, Father, Husband, Coach, Player, Disc Tosser, Driver... not necessarily in that order.

          Free Revit Chat | Cre8iveThings Blog | Live Architecture! | Past Live Architecture!

          Comment


            #6
            Well, I did some tests, but I am note sure what it all means.

            So I used a copy of the chip board texture, made it red and saved it at 1024, made another copy that is blue and 512, and a green one at 256. I have folders named 1, 2 and 3, each with a folder called mats, and I have added the root folder for all that to my rendering path. This mimics the OOTB folders. The three images are the same name and dispersed to match OOTB also. The material I created was specifically mapped to the high resolution image, just like OOTB materials.
            I then made a huge wall, such that a camera at 80' from the wall still has wall filling the frame. I then rendered, moved the camera to 40' from the wall and rendered, and again at 3' from the wall. All three renders took the same 1:42 +/- 1 second. And every render showed the red 1024 pixel texture.
            So then I remapped the material to the green 256 pixel version and repeated the rendering. All still 1:42 +/- 1!
            I then changed the wall to use an OOTB brick material, and got times of 2:30, 7:02 & 8:16 as I moved away from the wall.
            Last I copied the three Brick_Herringbone_Red textures to my folders, mapped the material to the 1024 version and got 1:41, 1:43 & 1:52. Then got basically the same times using the 256 pixel texture as well.

            Which means that I am really not sure what is going on yet, and have more testing to do. If nothing else, it looks like some OOTB materials are ridiculously slow compared to others, and just knowing which ones would be helpful. Another thing that wikihelp could maybe be useful for.

            Gordon
            Pragmatic Praxis

            Comment

            Related Topics

            Collapse

            Working...
            X