DonnyBrook MMOFPS technique

Tap In

Legend
DonnyBrook MMOFPS technique

view video for clear info on how the techniques work that Microsoft Research is testing and making available for future online games that will allow 100's or 1000's of participants in an online match in "real" time.

Unlike a Wow type MMO where you have a small group of the thousands actually playable with you, this allows predictive AI bots to simulate all the excess players (not in your immediate vicinity) to be calculated and rendered on your screen with reasonable precision. they do this by using infrequent data updates of their actions and locations therefore preserving precious bandwidth yet allowing those participants to become actually updated in real time as they become a focus or are closer.

interesting stuff IMO.

mods: Tech forum or here?
 
the heuristics of this technique are different. be sure to watch the vid for more details :)

also this will be made available as a tool to be used by devs on any game.
 
How is PlanetSide doing these days ?
(http://en.wikipedia.org/wiki/PlanetSide)

It's always great to see companies carry the torch further until we have a breakthrough...
http://www.hoise.com/primeur/02/articles/weekly/AE-PR-06-02-23.html
http://www.hoise.com/primeur/03/articles/weekly/AE-PR-04-03-15.html

It was hard to tell from your links, but this seems different than what your links are talking about.

The MSR technology is a client-side issue to preserve bandwidth at the expensive of requiring more client-side processing. Your hoise.com links are talking more about a server-side technology to intelligently allocate resources for MMOs.

MSR's technology is viable even in a P2P system.
 
It was hard to tell from your links, but this seems different than what your links are talking about.

The MSR technology is a client-side issue to preserve bandwidth at the expensive of requiring more client-side processing. Your hoise.com links are talking more about a server-side technology to intelligently allocate resources for MMOs.

MSR's technology is viable even in a P2P system.

Here's the original paper: http://research.microsoft.com/workshops/IPTPS2007/papers/PangUyedaLorch.pdf

I remember a conversation with Shifty, DemoCoder (or was someone else ?) about something similar: Saving bandwidth in P2P games using proximity grouping and automatic approximation in player movement (when they are "too far/slow"). The paper also uses bots as standins, and has a working implementation :cool:

They tested the techniques using Q3 (To simulate a large game, 2 players on 2 machines + 30 bots on 1 virtual server -- The bots are always in sync). It's a good effort. Hope they make some more progress for a PlanetSide scale.
 
Last edited by a moderator:
I wonder how one would reconcile the difference between the guide AI and the actual player when the actions taken by the bot are irreversible: falling off a cliff, walking onto a jumping platform, and so forth.
 
For irreversible events like this, they can block/loop the bot (e.g., fire blanks, standard evade/dodge) and try to get an update.

There is still work to do since their test bots are all in-sync. When they become out of sync (which should be common), that's where things become interesting.
 
Back
Top