home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!emory!ogicse!news.u.washington.edu!stein.u.washington.edu!hlab
- From: cdshaw@cs.ualberta.ca (Chris Shaw)
- Newsgroups: sci.virtual-worlds
- Subject: Re: TECH: DataGlove packet rate
- Message-ID: <1992Nov18.045717.22837@u.washington.edu>
- Date: 17 Nov 92 06:00:43 GMT
- Article-I.D.: u.1992Nov18.045717.22837
- References: <1992Nov14.011645.11929@u.washington.edu>
- Sender: news@u.washington.edu (USENET News System)
- Organization: University of Alberta
- Lines: 60
- Approved: cyberoid@milton.u.washington.edu
- Originator: hlab@stein.u.washington.edu
-
-
- In article <1992Nov14.011645.11929@u.washington.edu> dejohnso@asylum.cs.utah.edu
- (David Johnson) writes:
- >I have a little distributed VR system set up - SGI's, DataGlove and
- >Polhemus. I have only been getting about 7 frames/sec so I have been
- >doing some careful timing to see where the bottleneck is - and it's
- >mostly in trying to read the Polhemus 3Space. Right now it is configured for
- >2 sensors, binary, non-continous mode, 19200 baud serial port and I
- >can't get close to the 30 data packets/sec that they claim. Is this
- >experience common? What kind of data rate do you get using the MR
- >Toolkit? Any suggestions?
-
- It all depends on your polling structure. If you're polling the Isotrak once
- at the start of each update loop, you're going to guarantee a wait of 1/60
- second until the data packet is returned. Do that once per Isotrak, and you've
- lost 33 milliseconds already. Sometimes, the timing of the situation can
- be such that you'll get HALF the update rate when polling -- 15 Hz on your
- case. Running in continuous mode eliminates this polling lag.
-
- Our study of this problem, published in the User Interface Software and
- Technology 1991 Conference (ACM UIST'91) indicates that using
- continuous mode can reduce lag by up to 50 milliseconds.
-
- Using ascii polling mode on our slowest machine, the MR Toolkit can get at
- least 13 samples per second on a 3-way sync'd Isotrak using a SGI 3130 (20MHz
- 68020). Back in the old days, we used a 3130 and a SGI 2400 (15MHz or 20MHz
- 68010 box) as our stereo HMD drivers, and could get 15 updates per second
- on dirt simple models (hand plus 3 polygons). Considering that the maximum
- packet throughput on these machines is about 60 TCP/IP packets per second,
- this ain't too shabby. On more modern machines, we get 60 updates per
- second on our Isotrak and glove drivers, using binary continuous mode.
-
- The key point to note here is that you should always try to eliminate
- needless synchronization of the visual update upon other updates, such as
- tracker update. Polling the tracker requires you to hang around waiting for
- the latest data to show up. You can't always guarantee that you can fill
- this time with something useful.
-
- >Also - my VR thumb looks goofy - can anyone describe well how to model
- >its motion?
-
- Depends on your glove. We use a "finger" rotated 90 degrees so that the
- hinge-like joints' axes are perpendicular to the palm, then rotate out at
- the base of the thumb by 20 degrees or so. If you have more sensors, you
- can do more, of course.
-
- |||| <- fingertips | <- fingertips
- / |||| |
- | thumb tip-> \ |
- \ \|
- Back of hand view. Hand as seen while "karate chopping"
- Thumb on left. Thumb joint Rotate thumb counter-clockwise by 20 degrees
- axes come out of page. Rotation axis comes out of page.
-
- >David Johnson
- >dejohnso@cs.utah.edu
-
- --
- Chris Shaw University of Alberta
- cdshaw@cs.UAlberta.ca CatchPhrase: Bogus as HELL !
-