Heres a post which is less about day to day happenings but more about one of the many little ongoing research experiments I’m constantly tinkering with. Ive thought about this one for a while but haven’t yet tried to implement this until now!
After chatting with the Ai guys at work over the past few days about different content/ video mapping scenarios I’ve decided to try and patch sic a thing in vvvv specifically for playing content back on a “Volumetric type video display”. For a description on what the hell a volumetric display is, its basically a video display screen with depth or extra layers which the viewer can see. In most cases a video screen is a single skin surface from a TV screen, projector scene or a projection mapped building. A volumetric display can have depth as well as width and height, heres a brilliant example by AntiVJ.
Using everyones favourite node based programming software I positioned 4 quads in 3d space at varying depts in the z-plane show in the image above and below. This gave me my basic 3d scene and some reference markers for my camera to look at.
Using the same render information from the initial 3d scene I then too 4 views on the same scene, flat onto the relevant coloured quad. From there i then clipped the far and near plane in the Z-depth so that the view was only seeing a shallow depth of that particular plane. This proves the theory of this to be working but I run in to the issue of the model primitive I’ve been using in my scene is hollow, as are most models. Opening up the clipping planes gives a little pseudo thickness but I need to find a way giving the walls of the model a thickness, or the model it’s self some volume.
Do more research on:
- Volumetric Models
- Voxels
- Shader/.fx to pseudo thicken model walls on the output.
vvvv project available here > Volumetric Using:45beta 31.2 64bit