TestLAB: PE v0.3 – Day 1 to 3

Post by: Joris Weijdom Add comments
Performance engine v03

P1020592On Wednesday morning we drove the PE hardware to Huis aan de Werf. Our system now officially fits in the back of a Citroën Berlingo, which has about 3 cubic meter of loading space. Our current hardware setup roughly consists of the following hardware:

  • 3 Dell XPS PCs with Windows XP
  • 2 Optoma EP 774 and DLP 1 BENQ 720p beamers
  • 1 ACER S2000 short-throw DLP beamer
  • 1 Cisco Linksys WRT320N 1Gb Wireless Router
  • 1 3Com 8 port switch
  • 3 Logitech Quickcam Pro 9000
  • Vemux SM4X8 15 v LCD Video Matrix Switch
  • 3 WII-motes with 1 Plus adapter
  • A lot of cables (320V, VGA, UTP and 6 powered 5m USB! )

P1020594Based on our experience of the testing days we did in may, we decided to turn the opeator workstations towards the wall. This gave us more floorspace and, more importantly, gives visitors and participants a view on the operator monitors! Setting up the computers and network was done in half a day. Then we started with hanging the camera’s to the ceiling grid. 2 camera’s are used now for the optical color (or movement) tracking. These two camera’s are stitched together to get a larger floor-space that we can track. The third camera is used for the marker-based tracking with the Reactivion software.

On thursday we started with hanging the short-throw beamer for the back-wall projection. A short-throw beamer has a special lens that makes the projected image much larger on shorter distances than a ordinary beamer. Needless to say that this is a great advantage in small spaces and also makes it less likely that the performer is creating shadows.

Secondly we set up a new variation on the projection mapping tests we had already done before at the faculty. By projecting a perspectively correct image of a virtual 3D copy of the actual space and objects we create the possibility to play with collisions of virtual objects with physical objects. Also virtual lights can create (or actually simulate) virtual shadows coming from the physical objects. Have a look at the video below to get an impression.

More in depth information will be put online soon!

[media id=12 width=560 height=304]

At the end of the day we tried to get the stereoscopic projection to work. We hang our silverscreen, made already in a earlier TestLAB and setup two beamers with linear polarizing filters. But when running the test application within the NVIDIA stereoscopic driver setup we unfortunately encountered the horrible blue-screen of death. For non-pc users: serious crashes. So we let it go for the moment.

On Friday Joris went for three days to Ars Electronica and the rest of the team worked on concepts or used the time to further bug-fix and stress-test the system. We already noticed on Thursday that the system would really become slow after activating a few engines. After some tests and deduction we figured out several problems:

  1. Reactivision TUIO client takes a lot of (local) resources
  2. Our region patch was running through the network and taking way too much bandwidth
  3. The 2D engine was conflicting with Unity 3D in fullscreen OpenGL on one computer
  4. The Statemachine was sending way too much data over the network

Also we, again, encountered a lot of problems with the color-tracking so we decided to re-consider an earlier dismissed CV.JIT patch based on background substraction and thus detecting motion in any given area. Furthermore our new WII-mote patch (for the PC) needed to be finalized and many more bits and pieces.

Have a look below for some photos.

Comments are closed.

WP Theme & Icons by N.Design Studio
Entries RSS Comments RSS Log in