Re: [Flightgear-devel] Instant replay system
Wow. This is *great*! And after nobody responded to my request that it wasn't even on bottom of the priority list Thanks! -Matt Curtis L. Olson wrote: I have implimented a first stab at an instant replay system and have just commited it to CVS. The system continuously records your flight data and allows you to play back your flight. Saving flight data at full resolution can quickly burn up a lot of RAM, so at the moment, only the most recent 60 seconds is stored at full resolution. The next most recent 10 minutes is stored at a resolution of one snap shot every 0.5 seconds. The most recent hour of data is stored at a resolution of one snap shot every 5 seconds. (We might need/want to tune these values a bit as we use the system more.) However, this gives the most resolution to the most recent portion of the flight, and can save up to an hour's worth of flight data without using a huge amount of memory. Each snapshot is time stamped with the simulation time. During replay, the system finds the two snapshots that straddle the replay time and linearly interpolates between them. This way, we can record the data as best as is possible given the current rendering speed, even with varying frame rates with possible lost frames and aren't forced to take extraordinary measures to get a consistent sampling rate. Then when the data is replayed, this interpolation scheme gives us smooth playback even if the recorded frame rate is wildly (or slightly) different from the playback frame rate. This also gives us smooth interpolation for older data that is recorded at a slower rate. And we can play back the recorded flight at a rate that has a 1 to 1 match with real time. This scheme would also allow us to smoothly fast forward the replay, or replay in super smooth Slo-Mo-Gear (patent pending) :-) We could probably also rewind as well as fast forward if we really wanted to. For the moment, I have bound the r key to the instant replay function. This will replay whatever is in the buffer. Upon completion, you are left in pause mode at the point where you activated the instant replay. Just unpause and you can continue the flight. This whole thing is just crying for a gui/dialog box to control it all. Any volunteers? I'm guessing that all the kinks are not yet quite worked out. You'll likely run into one or two ... The sim is put into pause mode while the buffer data is replayed. This has some side effects because many subsystems do not run when the simulator is paused. This needs to be looked at a bit more. Specifically some of the viewer code isn't run. You can switch views with the v series of keys, but you can't rotate a particular view with shift-number pad, and the chase views don't track quite right in replay mode. Also, I don't and can't and won't record everything. This means things like AI or multiplayer traffic, weather conditions, and other things are beyond the scope of this system at this time. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Citiescurt 'at' me.umn.edu curt 'at' flightgear.org Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] Instant replay system
I have implimented a first stab at an instant replay system and have just commited it to CVS. The system continuously records your flight data and allows you to play back your flight. Saving flight data at full resolution can quickly burn up a lot of RAM, so at the moment, only the most recent 60 seconds is stored at full resolution. The next most recent 10 minutes is stored at a resolution of one snap shot every 0.5 seconds. The most recent hour of data is stored at a resolution of one snap shot every 5 seconds. (We might need/want to tune these values a bit as we use the system more.) However, this gives the most resolution to the most recent portion of the flight, and can save up to an hour's worth of flight data without using a huge amount of memory. Each snapshot is time stamped with the simulation time. During replay, the system finds the two snapshots that straddle the replay time and linearly interpolates between them. This way, we can record the data as best as is possible given the current rendering speed, even with varying frame rates with possible lost frames and aren't forced to take extraordinary measures to get a consistent sampling rate. Then when the data is replayed, this interpolation scheme gives us smooth playback even if the recorded frame rate is wildly (or slightly) different from the playback frame rate. This also gives us smooth interpolation for older data that is recorded at a slower rate. And we can play back the recorded flight at a rate that has a 1 to 1 match with real time. This scheme would also allow us to smoothly fast forward the replay, or replay in super smooth Slo-Mo-Gear (patent pending) :-) We could probably also rewind as well as fast forward if we really wanted to. For the moment, I have bound the r key to the instant replay function. This will replay whatever is in the buffer. Upon completion, you are left in pause mode at the point where you activated the instant replay. Just unpause and you can continue the flight. This whole thing is just crying for a gui/dialog box to control it all. Any volunteers? I'm guessing that all the kinks are not yet quite worked out. You'll likely run into one or two ... The sim is put into pause mode while the buffer data is replayed. This has some side effects because many subsystems do not run when the simulator is paused. This needs to be looked at a bit more. Specifically some of the viewer code isn't run. You can switch views with the v series of keys, but you can't rotate a particular view with shift-number pad, and the chase views don't track quite right in replay mode. Also, I don't and can't and won't record everything. This means things like AI or multiplayer traffic, weather conditions, and other things are beyond the scope of this system at this time. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Citiescurt 'at' me.umn.edu curt 'at' flightgear.org Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Instant replay system
On Friday 18 July 2003 04:06, Curtis L. Olson wrote: I have implimented a first stab at an instant replay system and have just commited it to CVS. The system continuously records your flight data and allows you to play back your flight. Saving flight data at full resolution can quickly burn up a lot of RAM, so at the moment, only the most recent 60 seconds is stored at full resolution. The next most recent 10 minutes is stored at a resolution of one snap shot every 0.5 seconds. The most recent hour of data is stored at a resolution of one snap shot every 5 seconds. (We might need/want to tune these values a bit as we use the system more.) However, this gives the most resolution to the most recent portion of the flight, and can save up to an hour's worth of flight data without using a huge amount of memory. Each snapshot is time stamped with the simulation time. During replay, the system finds the two snapshots that straddle the replay time and linearly interpolates between them. This way, we can record the data as best as is possible given the current rendering speed, even with varying frame rates with possible lost frames and aren't forced to take extraordinary measures to get a consistent sampling rate. Then when the data is replayed, this interpolation scheme gives us smooth playback even if the recorded frame rate is wildly (or slightly) different from the playback frame rate. This also gives us smooth interpolation for older data that is recorded at a slower rate. And we can play back the recorded flight at a rate that has a 1 to 1 match with real time. This scheme would also allow us to smoothly fast forward the replay, or replay in super smooth Slo-Mo-Gear (patent pending) :-) We could probably also rewind as well as fast forward if we really wanted to. For the moment, I have bound the r key to the instant replay function. This will replay whatever is in the buffer. Upon completion, you are left in pause mode at the point where you activated the instant replay. Just unpause and you can continue the flight. This whole thing is just crying for a gui/dialog box to control it all. Any volunteers? I'm guessing that all the kinks are not yet quite worked out. You'll likely run into one or two ... The sim is put into pause mode while the buffer data is replayed. This has some side effects because many subsystems do not run when the simulator is paused. This needs to be looked at a bit more. Specifically some of the viewer code isn't run. You can switch views with the v series of keys, but you can't rotate a particular view with shift-number pad, and the chase views don't track quite right in replay mode. Also, I don't and can't and won't record everything. This means things like AI or multiplayer traffic, weather conditions, and other things are beyond the scope of this system at this time. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Citiescurt 'at' me.umn.edu curt 'at' flightgear.org Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org That's a nice addition - sounds neat:) I was just about to turn in for the night, and I did an update earlier this evening, but I think I'll get this now:) Hmm... I've been using the 'r' key for reheat but that's not a problem - was a consensus reached on a scheme for allocating keys in the recent discussions? I remember some ideas being thrown around but don't recall any decision being made. LeeE ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Instant replay system
Lee Elliott writes: That's a nice addition - sounds neat:) I was just about to turn in for the night, and I did an update earlier this evening, but I think I'll get this now:) Hmm... I've been using the 'r' key for reheat but that's not a problem - was a consensus reached on a scheme for allocating keys in the recent discussions? I remember some ideas being thrown around but don't recall any decision being made. I have absolutely no emotional attachment to the r key. I'm happy to make it something else, or just unbind the replay function from the keyboard and make it a menu option or whatever. (It might make more sense to not have it bound to a key and just have a menu option that brings up some yet to be created instant replay dialog box.) Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Citiescurt 'at' me.umn.edu curt 'at' flightgear.org Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Instant replay system
On Friday 18 July 2003 04:06, Curtis L. Olson wrote: I have implimented a first stab at an instant replay system and have just commited it to CVS. The system continuously records your flight data and allows you to play back your flight. Saving flight data at full resolution can quickly burn up a lot of RAM, so at the moment, only the most recent 60 seconds is stored at full resolution. The next most recent 10 minutes is stored at a resolution of one snap shot every 0.5 seconds. The most recent hour of data is stored at a resolution of one snap shot every 5 seconds. (We might need/want to tune these values a bit as we use the system more.) However, this gives the most resolution to the most recent portion of the flight, and can save up to an hour's worth of flight data without using a huge amount of memory. Each snapshot is time stamped with the simulation time. During replay, the system finds the two snapshots that straddle the replay time and linearly interpolates between them. This way, we can record the data as best as is possible given the current rendering speed, even with varying frame rates with possible lost frames and aren't forced to take extraordinary measures to get a consistent sampling rate. Then when the data is replayed, this interpolation scheme gives us smooth playback even if the recorded frame rate is wildly (or slightly) different from the playback frame rate. This also gives us smooth interpolation for older data that is recorded at a slower rate. And we can play back the recorded flight at a rate that has a 1 to 1 match with real time. This scheme would also allow us to smoothly fast forward the replay, or replay in super smooth Slo-Mo-Gear (patent pending) :-) We could probably also rewind as well as fast forward if we really wanted to. For the moment, I have bound the r key to the instant replay function. This will replay whatever is in the buffer. Upon completion, you are left in pause mode at the point where you activated the instant replay. Just unpause and you can continue the flight. This whole thing is just crying for a gui/dialog box to control it all. Any volunteers? I'm guessing that all the kinks are not yet quite worked out. You'll likely run into one or two ... The sim is put into pause mode while the buffer data is replayed. This has some side effects because many subsystems do not run when the simulator is paused. This needs to be looked at a bit more. Specifically some of the viewer code isn't run. You can switch views with the v series of keys, but you can't rotate a particular view with shift-number pad, and the chase views don't track quite right in replay mode. Also, I don't and can't and won't record everything. This means things like AI or multiplayer traffic, weather conditions, and other things are beyond the scope of this system at this time. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Citiescurt 'at' me.umn.edu curt 'at' flightgear.org Minnesota http://www.menet.umn.edu/~curt http://www.flightgear.org The replay seems ok on a quick check here - no immediately obvious problems. The autopilot altitude hold seems a bit messed up though since yesterday - quite bad porpoising - I checked that with the 747, a4, c-172 and a couple of mine. Terrain following still seems ok though. I doubt this is due to the replay stuff. LeeE ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
