Hello,
The DIS-VRML-Java people probably have a solution. It may be time based,
sort of "OK, I'm caught up to time x, so send more positions and don't
advance your simulation time till I say so..."
Of course, this could slow down all the participants to the least common
denominator speed wise...
DIS-VRML-JAVA may also have a Dead reconning feature, so slower machines
can interpolate positions and then correct when they can accept an official
position update. The slow machine can just toss any EAI messages until it
needs to correct.
John F. Richardson
At 10:05 AM 6/18/99 -0500, you wrote:
>Hi Carlos;
>
>First, look at http://www.crc.ca/FreeWRL/ for a linux based browser. If
>you use it with VNET, make sure you select "tinman" as your avatar. Some
>of
>the built in avatars have problems, and FreeWRL doesn't handle avatar
>problems as nicely as Cosmoplayer does.
>
>
>My problem was with the rate of eventouts for position changes. Lets
>say,
>on my really fast computer, with hardware accelleration, that I can make
>70 position moves per second, which corresponds to 140 EAI messages per
>second. (position and orientation - two different EAI messages)
>
>So, then, we have 140 messages being sent to a browser on another
>machine
>that can, (for example) either through network delays or machine speed,
>handle 10 per second.
>
>That means that one second of my move, takes 14 seconds to make the same
>move
>on the other machine.
>
>If we could throttle down the rate of sending EAI messages to, say, 10
>per
>second, that makes up to 20 messages (pos and orient again) per second;
>the
>other machine would now take 2 seconds to show this move, not 14.
>
>Big difference!
>
>When testing my implementation, I would go to another lab, do some moves
>with one browser, walk back to my lab, and see the tail end of my moves.
>
>I have managed to speed things up considerably, btw, by implementing two
>things in my eai code, so now my browser is able to handle things well.
>
>But, the network problem still exists.
>
>Cheers;
>
>John Stewart
>[EMAIL PROTECTED]
>
>> I'm not sure to understand your problem. I understnad that every eventOut
>> has to been sent to another machine in the network to do some work with it.
>> If this is true, may be you could make some of that work in the local
>> machine before sending to remote.
>> I hope I can help you.
>> By the words, would you tell me where can I find a Linux based VRML
Browser.
>> Thank you very much.
>>
>>
>
