movie recording details? - WorldViz User Forum

halley · #1 05-30-2006, 11:29 AM

I read discussions involving the plans for AVI and/or snapshot recording from within Vizard. If these are script-accessible features, a demo script would be helpful. (If they're not script accessible yet, maybe it's not too late to expose some rudimentary start/stop/snap routines.)

halley · #2 06-01-2006, 07:50 AM

I found the viz.screenshot() routine for single frame snaps. I still haven't figured out if there is programmatic or hotkey support for recording AVIs.

tobin · #3 06-06-2006, 12:58 PM

There are two new recording methods that will be supported under R3.

The first is a raster recorder that makes uncompessed AVIs. It's built-in the R3 B1 engine already and can be toggled on/off by pressing CTRL-F12. The screen size will be forced to 640x480 and the avi file will be stored in the current project folder. I believe the frame-rate is currently capped at 30Hz. Be ware that this is an uncompressed avi and will grow to quite a large file on the hd quite quickly. Window's MovieMaker is good for quickly converting that to a much smaller wmv file, or use virtualdub for doing more serious stuff.

The second is a vector recorder. This records amost everything that occurs in a simulation to a text file that can later be played back for viewing. The difference is that the "playback" is a live simulation that can be viewed from novel vantage points. See the attached sample files for an explanation on using the vector recorder.

halley · #4 06-06-2006, 01:14 PM

Thanks, tobin. These specs are in line with the info which farshizzo had mentioned to me separately. He didn't share the keyboard switch with me, but did explain the two methods (raster versus action recording) that would be offered. Both will be highly useful in our work here.

Also it sounds AWESOME that the action recorder is just dumping an ASCII file... this may mean that I can programmatically crop or tweak (clean up) the recorded sequence as well, by massaging this text file.

I'll have to dig into that example and try a few things, but an ultimate concept we have in mind would be to record one person doing something in the immersive space, then take that recording and MIX its geometry and motion into another simulation session that involves other objects. Think oompa loompas. (The big question would be conflicts with internally chosen object identifiers, between recorded and playback session geometry, I think.)