3D animation ques.

  • Started
  • Last post
  • 5 Responses
  • harlequino

    Creating this scene in which the camera passes through a narrow corridor of sorts, makes a couple twists and turns, and then gets spit out into a much larger open space, like a big empty warehouse.

    The geometry involved is not very complicated and I am doing a lot of textural details with displacement mapping and such, but am I asking for a big hassle if I do this whole thing in a single scene? Am I better off splitting up the 3D scenes and cutting together later on?

    Just looking for any guidance when it comes to scenes in which your action or camera covers a lot of 'physical' space.

  • M_C_P0

    is it one long take? consider what might change later down the line and see if breaking it up is justified. then look for obvious points to break up the camera move (ie panning behind a wall can wipe on a new scene/render in post, etc)

    if you're gonna split it up one thing you might look in to is placing dummy geometry that will stand in for your "hi-res" ones. plot out your camera path. then break up the scene into separate parts (hallway, warehouse, etc) and copy paste the camera path into each that way each scene is lighter but still continuous when you composite all the pieces together.

  • harlequino0

    ^Thanks and cheers.

    Yeah, very much along the lines of where I am at. I originally envisioned it as one continuous take. But I'm not necessarily married to that. If it gets too much, I'll find some cutting points.

    The final set (the warehouse-ish one) is blowing up a bit in terms of scale and some geometry details. I am thinking of maybe having nothing (or fpo geometry as MCP said) for the final set, and rendering with an alpha channel. That way, if I can accurately duplicate a starting camera position as it enters the final set, in compositing, layering one over the other should be fairly seamless.

    I think that makes sense...

  • CyBrain0

    Displacement mapping is a huge processor killer. Maybe you could invest some time upfront by pre-renderiing those parts of your movie and import those parts as quicktime?

  • harlequino0

    Haven't had any issues so far with previewing still frames or short segments, but I'll give that a shot.

  • M_C_P0

    if the camera isn't gonna be super up close on the textural details of the walls and/or the cam is moving pretty quickly, you might shave off more render time if you substitute bump maps for displacement maps. i guess it depends on how much more subdividing you're gonna do to the base geometry vs how much more realism you're aiming for. just another consideration.