Announcement

Collapse
No announcement yet.

use camera path.. more details?

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

  • use camera path.. more details?

    ive never really used the "use camera path" options very much, and the details on how it works are a little vague.


    my question is: i have a flythrough with lots of drastic changes in camera speed. this makes using "multiframe incremental" a bit of a pain, as every 20 frames wont cut it, maybe even every 10 is too few, but the more frames i add, the more huge the imap gets due to the grass and trees (theres not much sample overlap even on consecutive frames at the micro detail level, meaning i also get very little speedup from previously calculated samples.... if that makes sense)

    so in this case using something like "use camera path" would seem like a sensible choice as i would not need to test to find the optimum "nth frame" setting, or manually choose frames to do the imap calc on.


    HOWEVER, its not at all clear how one would use the "use camera path" setting to effectively replace "multiframe incremental" ?

    i note in the past vlado has said its fine to use on "short camera moves" but what constitutes a short camera move? this is a 1000 frame sequence passing through a lot of scenery.

    if i simply use it, i understand i need to increase the hsph subdivs? is this correct? how to judge how far to increase this setting? seems a bit counter intuitive to me, as the hsph subdiv controls the sample quality, not density....?


    *basically* can i use "use cam path" on the imap for a long flythrough, if so how, and if not, why not and what are the limitations / reasons..


    as a side note there are many similar newish features in vray that have never been documented in much detail.. might be worth spending some time doing a proper update of spot3d....?

  • #2
    A guess:
    More Detail = more samples. Two ways to get more samples: Precalculate with a higher image-resolution or use oversampling (not -3/-1 but more like 0/4 or higher). As far as I understood this thing, every GI-sample is calculated at a random-time of the sequence. More samples would mean more accuracy over time for more detail... but this is just a guess!
    If the camera animation is from 0-300 but the timeline is to 1000, it messes up the sampling as well as 700 of 1000 possible timesamples are showing the last frame.. Check the vrmap if you like what you see.
    Kind regards, Wolf S./K.
    www.faber-courtial.de

    Comment


    • #3
      sounds like a good guess! id have imagined something similar.. really hoping someone at chaos can chime in with some concrete descriptions of how to control this method, for longer camera moves.

      Comment


      • #4
        To set a fly-through animation with the "use camera path" option you need to use IRR map in a single mode. Set its settings so you get a good render quality for a frame. Then just enable the "use camera path" option. After IRR map calculation is done you need to save it and load it "form file".
        As for the LC calumniation make sure its "use camera path" option is enabled and you use World scale. In contrast to the IRR map you need to increase the subdivs of the LC double to the one you get a good image quality for a frame.
        Best regards,
        Zdravko Keremidchiev
        Technical Support Representative

        Comment


        • #5
          How does "use camera path" determine where to put new samples on each frame? Does it have some kind of rule about the level of sample density it needs to reach on each part of the scene it flies through?

          Comment


          • #6
            V-Ray sets all the samples on the visible part of the scene throughout the animation path. There is no specific rule there.
            Best regards,
            Zdravko Keremidchiev
            Technical Support Representative

            Comment


            • #7
              thanks for the official reply Zdravko. so i dont need to increase any quality settings to compensate for the length of the camera path? thats easy then.. i shall do some tests.

              also regarding the lightcache, i generally use "flythrough mode" and screen scale, just because im dealing with massive landscapes that cover many km. world scale is a killer in these cases. i assume its ok to use flythrough mode and "screen scale? or is there some reason to specifically use world scale in this case...

              Comment


              • #8
                Generally we would not advise using LC when rendering trees. That might be very time consuming due to the Light cache countless bouncing capability.
                In this cases its best to use Brute force instead.
                Best regards,
                Zdravko Keremidchiev
                Technical Support Representative

                Comment


                • #9
                  ahh the lightcache works fine in my case.. renderings look great and go pretty quick. just the issue of having to choose frames to calculate when using multiframe incremental that im trying to resolve. and im hoping that "use camera path" for the imap will let me avoid this issue.

                  Comment

                  Working...
                  X