/Christer, thinking they wouldn't have any chance at all against the
GPL hotshots
That would be great!! Maybe some kind of app could be written to deliver
the races in real-time to a browser, like AMA Superbike is said to be
able to do when it's released.
--
Cheers!
Graeme Nash
(Calvin & Hobbes)
http://www.karisma1.demon.co.uk
ICQ# 11257824
___________________________________________________________________________
"There's never enough time to do all the nothing you want"
You can download it from here, if my ISP hasn't deleted the file...
http://www.karisma1.demon.co.uk/gpl/races/kya_race.htm
--
Cheers!
Graeme Nash
(Calvin & Hobbes)
http://www.karisma1.demon.co.uk
ICQ# 11257824
___________________________________________________________________________
"There's never enough time to do all the nothing you want"
Mark
> /Christer, thinking they wouldn't have any chance at all against the
> GPL hotshots
> > What about an online race between Jackie Stewart and Michael Schumaker,
> > with some GPL hotshots thrown in?
> --
> http://home.swipnet.se/~w-41236/ (Read all about the "Global online
> racing"-proposal under "For developers". Read it a couple of times,
> cause noone has understood it the first time they've read it yet :o)).
Better than watching TV, as you can use the replay facility to switch to
any driver in any view, and backtrack if you see something interesting.
Then I realised that this could actually work - the same code that
allows up to 20 cars to take part in a single race could be modified to
allow thousands (or even millions) of people to watch a race in real
time, without affecting the performance of the race server itself.
All that needs to happen is for one computer to join the race as a
spectator (ie. it doesn't have a car of its own, but it receives the
position information for the other cars), and then anyone else joining
the race as a spectator will actually join this machine rather than the
actual race machine.
This is a bit of a simplification: in fact it would have to work as a
tree structure: the first machine joins the actual race, the next 20 (or
maybe less depending on performance) join this machine, the next 400
join the second-level machines, etc. etc.
Even if you assume that each node in the tree can only support 4 other
machines, you'd only need 10 levels of indirection for a crowd of well
over 1 million (that gives over 1000,000 'leaf' nodes, plus the ones
higher up the tree), which is not really a problem as there is no need
for information to flow back up the tree, so there are no
synchronisation problems and no degradation of performance. (It's
possible to put in a slight time delay to give delayed packets time to
arrive).
This approach would need the cooperation of the central VROC server to
store the IP addresses of all nodes in the tree, so that new people
joining would automatically be assigned places at the 'branch' end of
the tree, and it could also allow automatic reassignment if one of the
machines in the tree were to break down (or be thrown out of the window
by the spouse of the ***ed GPL fan, say).
All you'd see in GSB would be a single race, but when you join as a
spectator the GSB client asks the VROC server for which node it should
actually join, and in return the client machine's IP address is added to
the tree.
It's really much simpler than the stuff that GPL already does in order
to support online racing!
All we need then is built-in voice support to allow one or more
commentators to broadcast to everyone in their part of the tree
(localised for different countries of course) and we have a global media
event of staggering proportions!
Assuming enough people are actually interested in watching, of course...
> >Assuming enough people are actually interested in watching, of course...
> Heh, sounds like fun to me. I had a ball just watching a replay of Grahme Nash
> winning a race at Kyalami by a few feet against an AI car, so I can't image how
> much fun it'd be to see live online racing! :)
> > Then I realised that this could actually work - the same code that
> > allows up to 20 cars to take part in a single race could be modified to
> > allow thousands (or even millions) of people to watch a race in real
> > time, without affecting the performance of the race server itself.
> for the NROS, since it's an official NASCAR sanctioned series.... this has
> been though for years.
> The main problem is that even for today the connections aren't enough fast.
> Just think about how much download you would need to see all the cars on the
> track. Because even in GPL when you race online, you see what 4-5cars
> forward on fast connections and 1 in your mirrors, no ?
> Now just imagine if you had ALL the cars. And on the NROS it means 24cars.
> It's the current limitation. Even with ADSL and Cable-modem, it's barely
> acceptable. Especially with GPL and it's physics. But for the future, there
> is a great market for such a thing, and incredible possibilities.
Additionally, since information flow is one way only, it's possible to
almost completely eliminate latency problems - all you need to do is to
get each machine to buffer car positions in the background so that it
doesn't need to display the next frame until (say) 10 seconds after it
should have received the next set of car positions (by which time it
should already have the next several seconds' worth of positions
buffered up as well).
This way, if there is a delay between packets of anything less than 10
seconds (or whatever the time delay is set to), the user notices
absolutely nothing.
This means that the only jitter the spectators see is exactly that which
the race server sees - ie. the latency in the drivers' machines sending
their positions to the host.
The only reason there is a problem with latency when racing is that
information flow has to be two-way, so it's not possible to buffer up
positions and display them later.
Note that if they were going to do this seriously the race organisers
would probably set up the top-node spectator machine themselves on a LAN
connected to the race server, and possibly even have language-specific
servers at the next level as well, since the machines at the top of the
tree are the most critical to the working of the system - if any machine
goes down, all machines that were connected to it (and any others below
them) will have to wait while the machines that were directly connected
to the failed one renegotiate with VROC to get the IP addresses of
alternative hosts. Of course, if you set the display time delay to 10
minutes or so, there will nearly always be time to reconnect before the
user notices anything.
[*] In case you haven't heard of them (!) Amway is the company that
pioneered network marketing in the states, whereby each person recruits
several other people and sells stuff to them, which they then send to
more people that they recruit, etc. etc.
> http://WWW.Resounding.com Here you can pick up a "voice" communications
> application called "Roger Wilco". With this ap, You log onto an IP address
> and voice chat with other people with that app. My self(Barlow_Racing),
> Ross MacGreggor(MacWun), Mark seery(Mark_Seery), have been rigorously
> testing this app while online in GSB. We have even went as far as chatting
> while racing against each other in GPL. Even when someone got booted from
> the race, we were still able to voice chat with them and give a play-by-play
> of the action. This app is free ware. The demo has a 15 minute limit
> between uses. However, it's FREE to register it. When registering it,
> you're asked some basic questions and you get a registration key set to get
> rid of that 15 minute time limit.
> [SNIPPED]
I also heard that Microsoft has bought them out so that they can put
this into DirectX 8, so they obviously think it's worth using!
Would it be worth trying out this system instead, or is that just too
much hassle now that you've already got the RW stuff going?
> You are not alone in this thought Neil. I have been thinking about this for
> a long time now. All it would take would be a server to receive data
> streamed from the race feed. A function that would not be too hard to
> incorporate. This in turn could be fed out in the manner you describe. Cut
> down client viewing software - IE the replay part of most programs
> (freeware), could be distributed on mag covers and be available to download
> on the web.
> If we had the co-operation of the software houses and manufacturers of
> hardware on this one, imagine the possibilities. Semi Pro racing might not
> be too far off. Software could be demoed in real time to prospective buyers
> in a much more exiting manner, kudos would be gained by winning hardware
> manufacturers ....hundreds of people watching a race, A driver using a TSW
> wheel wins the race. TSW gets loads of orders in the next week. Sound
> familiar?
The thing is, I really don't think it would take much for Papy to put
this into the game, as a lot of the functionality is pretty similar to
what already happens when racing.
The main extra work would need to go into the VROC server and GSB, as
they would have to maintain the list of the crowd's IP addresses and
negotiate to find suitable hosts.
There's no doubt that this type of system will become a huge commercial
success for someone in the future, but I guess they need to get all the
elements in place at the right time, including sponsorship, media
interest etc. etc.
What about an online race between Jackie Stewart and Michael Schumaker,
with some GPL hotshots thrown in?
Professional sim racing, could be around the corner...
/Christer, would be watching :o)
> I had a mad idea the other day - wouldn't it be great if us ordinary
> mortal GPL fans could watch a world championship race in real time?
> Better than watching TV, as you can use the replay facility to switch to
> any driver in any view, and backtrack if you see something interesting.
> Then I realised that this could actually work - the same code that
> allows up to 20 cars to take part in a single race could be modified to
> allow thousands (or even millions) of people to watch a race in real
> time, without affecting the performance of the race server itself.
> All that needs to happen is for one computer to join the race as a
> spectator (ie. it doesn't have a car of its own, but it receives the
> position information for the other cars), and then anyone else joining
> the race as a spectator will actually join this machine rather than the
> actual race machine.
> This is a bit of a simplification: in fact it would have to work as a
> tree structure: the first machine joins the actual race, the next 20 (or
> maybe less depending on performance) join this machine, the next 400
> join the second-level machines, etc. etc.
> Even if you assume that each node in the tree can only support 4 other
> machines, you'd only need 10 levels of indirection for a crowd of well
> over 1 million (that gives over 1000,000 'leaf' nodes, plus the ones
> higher up the tree), which is not really a problem as there is no need
> for information to flow back up the tree, so there are no
> synchronisation problems and no degradation of performance. (It's
> possible to put in a slight time delay to give delayed packets time to
> arrive).
> This approach would need the cooperation of the central VROC server to
> store the IP addresses of all nodes in the tree, so that new people
> joining would automatically be assigned places at the 'branch' end of
> the tree, and it could also allow automatic reassignment if one of the
> machines in the tree were to break down (or be thrown out of the window
> by the spouse of the ***ed GPL fan, say).
> All you'd see in GSB would be a single race, but when you join as a
> spectator the GSB client asks the VROC server for which node it should
> actually join, and in return the client machine's IP address is added to
> the tree.
> It's really much simpler than the stuff that GPL already does in order
> to support online racing!
> All we need then is built-in voice support to allow one or more
> commentators to broadcast to everyone in their part of the tree
> (localised for different countries of course) and we have a global media
> event of staggering proportions!
> Assuming enough people are actually interested in watching, of course...
I like more to dl the race replay when it is finished ... (in Spain the
connection time is gold)
ie:
GPL Championship
Race 1 Nurbur <-- Click to download
Race 2 Monza
...
...
Race N
easier, isnt? :)
McKafre De La Rosa
And we'd get teamorders on-track, and we'd get drivers very anxious to loose
the prize money, and in the process falling into in-line racing, waiting for
errors and mistakes by the drivers in front. And then we'd get concord
agreements with the winning teams, and we'd get our very own virtual Bernie
Ecclestone, and Max Mosley.
Who'd like to be Schumacher? Anyone? Or Jimmy Clark?
---Asgeir---
CuSeeME from Cornell (the VC software) implemented a reflector
system of satellite machines to enable lurking (ie. watching bu
not really participating). From memory the reflector software
was freely available off the Cornell pages. Unix reflectors
were simple to setup ...
I dropped out of this when CUSeeMe went more commercial, but I
can't believe that the software isn't floating around
somewhere ...
> > You are not alone in this thought Neil. I have been thinking about this for
> > a long time now. All it would take would be a server to receive data
> > streamed from the race feed. A function that would not be too hard to
> > incorporate. This in turn could be fed out in the manner you describe. Cut
> > down client viewing software - IE the replay part of most programs
> > (freeware), could be distributed on mag covers and be available to download
> > on the web.
> > If we had the co-operation of the software houses and manufacturers of
> > hardware on this one, imagine the possibilities. Semi Pro racing might not
> > be too far off. Software could be demoed in real time to prospective buyers
> > in a much more exiting manner, kudos would be gained by winning hardware
> > manufacturers ....hundreds of people watching a race, A driver using a TSW
> > wheel wins the race. TSW gets loads of orders in the next week. Sound
> > familiar?
> Dead right!
> The thing is, I really don't think it would take much for Papy to put
> this into the game, as a lot of the functionality is pretty similar to
> what already happens when racing.
> The main extra work would need to go into the VROC server and GSB, as
> they would have to maintain the list of the crowd's IP addresses and
> negotiate to find suitable hosts.
> There's no doubt that this type of system will become a huge commercial
> success for someone in the future, but I guess they need to get all the
> elements in place at the right time, including sponsorship, media
> interest etc. etc.
> What about an online race between Jackie Stewart and Michael Schumaker,
> with some GPL hotshots thrown in?
Craig Morgan Sun Microsystems Ltd
Training Instructor Enterprise Services
European Systems Training Centre (ESTC) Citygate
Cross Street
Tel: +44 (0)161 905 8155 UK
Fax: +44 (0)161 962 4140 M33 7JF
I make the following assumption:
If you make a digital box capable of giving dead accurate imputs to a F1 car,
and placed it in place of the driver, and let the driver remotely steer the
car with his electronic wheel and pedals, the driver would not be able to
drive the car any faster than if he was sitting in the car himself. Far from
it!
And I find it very hard to believe that the GPL hotshots really are better
drivers than Schumacher and Clark, Prost and Senna...
---Asgeir---
> /Christer, thinking they wouldn't have any chance at all against the
> GPL hotshots
> > What about an online race between Jackie Stewart and Michael Schumaker,
> > with some GPL hotshots thrown in?
> --
> http://home.swipnet.se/~w-41236/ (Read all about the "Global online
> racing"-proposal under "For developers". Read it a couple of times,
> cause noone has understood it the first time they've read it yet :o)).
: I make the following assumption:
: If you make a digital box capable of giving dead accurate imputs to a F1 car,
: and placed it in place of the driver, and let the driver remotely steer the
: car with his electronic wheel and pedals, the driver would not be able to
: drive the car any faster than if he was sitting in the car himself. Far from
: it!
Well, I'm not so sure - didn't one of the drivers say something along
the lines of "it's the danger that makes it competitive - any fool
could drive one of these things fast if it wasn't dangerous" - wish
I could remember the exact quote. Sure, you'd have the disadvantage of
not having the "feel" of the roll of the car - but you'd have the huge
advantage of not risking your life in any crash.
: And I find it very hard to believe that the GPL hotshots really are better
: drivers than Schumacher and Clark, Prost and Senna...
All depends what you mean by "better" - I'm damn sure that they're
better players of GPL than Schumi, Clark, Prost would be.
Sure, playing GPL isn't going to mean that you can just sit in a '67
car and do hotlaps for real better than the pros. I'm also sure that if
the '67 drivers had GPL available they could use it to learn tracks.
--
Richard G. Clegg Only the mind is waving
Dept. of Mathematics (Network Control group) Uni. of York.
www: http://manor.york.ac.uk/top.html
