RE: [Flightgear-devel] ANN: a new dimension to FlightGear
Gouthas, Themie writes: I dont think the alpha sorting code was ever comitted, so currently I dont beleive PLIB will alpha sort. Really? I thought this was something that Steve had in place even in the early days of plib ... unless it was broke along the way and still needs to be fixed? Regards, Curt. -Original Message- From: Curtis L. Olson [mailto:[EMAIL PROTECTED]] Sent: Tuesday, 16 July 2002 9:10 PM To: [EMAIL PROTECTED] Subject: Re: [Flightgear-devel] ANN: a new dimension to FlightGear David Megginson writes: David Findlay writes: Just noted one problem. The billboard backgrounds are transparent only for the ground, not other objects. This doesn't look great when you're buzzing through the forest. Other than that it looks great. Thanks, Noted. Unfortunately, alpha transparency is a nightmare -- only things drawn *before* the object with transparency will show through. I draw the dynamically-placed objects after the ground, so you can always see the ground through them, but they are drawn before the clouds in the sky (since you will want to see them through the gaps in the clouds). It will be (apparently) random whether any individual tree shows up through the alpha portion of any other individual tree's texture. I know of no appropriate solution for this problem. The current arrangement works best for the normal situation, flying above 100ft AGL, but you're right that it can cause problems when you're buzzing cows. David, What you need to do is set the transparent flag in the ssgSimpleState for these objects. That way plib can sort them and draw them back to front. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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 -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Curtis L. Olson writes: Really? I thought this was something that Steve had in place even in the early days of plib ... unless it was broke along the way and still needs to be fixed? No, unfortunately we've already been through this with the 3D models. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Gouthas, Themie wrote: I dont think the alpha sorting code was ever comitted, so currently I dont beleive PLIB will alpha sort. I'm not sure this is a great idea in any case. There are a *lot* of these objects, and doing an NlogN sort of them (with attendant geometry processing to get the distances, not to mention the cache effects of doing an extra sweep over all of them) every frame is likely to be awfully slow. Hacking around the issue by diddling the rendering order (and maybe double-rendering problem objects like nearby clouds) sounds like the best idea to me. We could also investigate the use of destination alpha, which is available and fast on high end hardware these days. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] ANN: a new dimension to FlightGear
Granted, about the NlogN and cache misses penalties.. but whateve solution you choose will be at a cost. Double rendering cloud layers? There goes your fill rate out the window! If anyone is interested in experimenting, I'll be happy to provide the code that you can toggle alpha sorting. Ive never done a performace comparison, but seeing as you guys have something implemented that can excersise it, it might be worth your while to assess its merits quantitavely. By the way, do you mind elaborating on what you mean by use of destination alpha. regards, Themie -Original Message- From: Andy Ross [mailto:[EMAIL PROTECTED]] Sent: Thursday, 18 July 2002 2:22 AM To: [EMAIL PROTECTED] Subject: Re: [Flightgear-devel] ANN: a new dimension to FlightGear Gouthas, Themie wrote: I dont think the alpha sorting code was ever comitted, so currently I dont beleive PLIB will alpha sort. I'm not sure this is a great idea in any case. There are a *lot* of these objects, and doing an NlogN sort of them (with attendant geometry processing to get the distances, not to mention the cache effects of doing an extra sweep over all of them) every frame is likely to be awfully slow. Hacking around the issue by diddling the rendering order (and maybe double-rendering problem objects like nearby clouds) sounds like the best idea to me. We could also investigate the use of destination alpha, which is available and fast on high end hardware these days. Andy -- Andrew J. RossNextBus Information Systems Senior Software Engineer Emeryville, CA [EMAIL PROTECTED] http://www.nextbus.com Men go crazy in conflagrations. They only get better one by one. - Sting (misquoted) ___ 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
RE: [Flightgear-devel] ANN: a new dimension to FlightGear
Gouthas, Themie writes: If anyone is interested in experimenting, I'll be happy to provide the code that you can toggle alpha sorting. Ive never done a performace comparison, but seeing as you guys have something implemented that can excersise it, it might be worth your while to assess its merits quantitavely. You might see a difference alpha-sorting, say, 5000 billboarded trees that didn't show up for a few simple polys. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] ANN: a new dimension to FlightGear
Well, here is a simple way to actually get a qualitative feel.. perhaps you may be pleasantly surprise, perhaps you can chant I told you so! with conviction and a wry grin. All I'm suggesting is try it before dismissing it. Just replace ssgDlist.cxx with this version, no other changes are necessary. If your billboards have the translucent bit set, they will subsequently, be sorted before each frame. If it works well for 500 trees and provides good visual integrity over 5000 unsorted trees that keep sporadically disappearing behind invisible walls, perhaps a more judiciuos use of LOD and fewer billboards might be the overall best result. Most Visual Simulation packages and DIS Stealth Viewers do this, which is why I'm not convinced that its completely dismissible. I would prefer to test, and report the results myself, but I'm only a lurker on this mailing list for inspiration and ideas, so I am not set up to build FSGear. :-) -Original Message- From: David Megginson [mailto:[EMAIL PROTECTED]] Sent: Thursday, 18 July 2002 11:00 AM To: [EMAIL PROTECTED] Subject: RE: [Flightgear-devel] ANN: a new dimension to FlightGear Gouthas, Themie writes: If anyone is interested in experimenting, I'll be happy to provide the code that you can toggle alpha sorting. Ive never done a performace comparison, but seeing as you guys have something implemented that can excersise it, it might be worth your while to assess its merits quantitavely. You might see a difference alpha-sorting, say, 5000 billboarded trees that didn't show up for a few simple polys. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel ssgDList2.cxx Description: Binary data
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, 16 Jul 2002 03:33, David Megginson entered data into a CPU register that indicated: Over the weekend, I finished my first take on dynamically-placed Just noted one problem. The billboard backgrounds are transparent only for the ground, not other objects. This doesn't look great when you're buzzing through the forest. Other than that it looks great. Thanks, David -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE9M72JZOfFgbBAbXARAon3AJoDBF0DcCKIXSsSLaavod3pZJLqDQCfQjq4 UBmgF/i9YyZsI6gqi/Gb0rU= =LJMW -END PGP SIGNATURE- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Curtis L. Olson writes: We need a farmer and a tractor for the farm fields here in MN. If the farmer's name was Nic then we could have a Scenic overlook, he's outstanding in his field. Ok, sorry about that, it was painful to say, even for me. :-) Seriously, I am planning on adding the occasional barn, farmhouse, silo, and blue aluminum storage shed. This points to another problem, though -- as with textures, buildings will look very different in different parts of the world and even in different parts of the same country, province, state, or county. What we need is conditions in materials.xml; for example, within a certain lat/lon rectangle, use a tumbled-down stone building for a barn; within another, use a North-American wooden barn; and so on. We already have XML conditions, so it's doable, but it will be a bit of work all the same. Instead of allowing full conditions, we might allow only a few criteria (maybe time of year, lat/lon, elevation, slope). All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Findlay writes: Very nice. I tried to also do this ages ago but couldn't figure it out. :-) What is the framework for creating the objects like? If a good framework API could be developed, it would be easy for scenery designers to add new objects that are locally appropiate. For instance, kangaroos on runways in australia :-). Thanks, See my last posting. For my part of North America, it would be groundhogs on the runway, with the occasional deer or moose on smaller strips. Actually, you might want to animate these; it would be annoying if the kangaroo always blocked the same runway. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Findlay writes: Just noted one problem. The billboard backgrounds are transparent only for the ground, not other objects. This doesn't look great when you're buzzing through the forest. Other than that it looks great. Thanks, Noted. Unfortunately, alpha transparency is a nightmare -- only things drawn *before* the object with transparency will show through. I draw the dynamically-placed objects after the ground, so you can always see the ground through them, but they are drawn before the clouds in the sky (since you will want to see them through the gaps in the clouds). It will be (apparently) random whether any individual tree shows up through the alpha portion of any other individual tree's texture. I know of no appropriate solution for this problem. The current arrangement works best for the normal situation, flying above 100ft AGL, but you're right that it can cause problems when you're buzzing cows. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] ANN: a new dimension to FlightGear
David, What we need is conditions in materials.xml; for example, within a certain lat/lon rectangle, use a tumbled-down stone building for a barn; within another, use a North-American wooden barn; and so on. We Splended idea. Besides, it would set us ahead of MS FS, which (AFAIK) basically does lack such a mechanism. Regards, Michael -- Michael Basler, Jena, Germany [EMAIL PROTECTED] http://www.geocities.com/pmb.geo/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson writes: David Findlay writes: Just noted one problem. The billboard backgrounds are transparent only for the ground, not other objects. This doesn't look great when you're buzzing through the forest. Other than that it looks great. Thanks, Noted. Unfortunately, alpha transparency is a nightmare -- only things drawn *before* the object with transparency will show through. I draw the dynamically-placed objects after the ground, so you can always see the ground through them, but they are drawn before the clouds in the sky (since you will want to see them through the gaps in the clouds). It will be (apparently) random whether any individual tree shows up through the alpha portion of any other individual tree's texture. I know of no appropriate solution for this problem. The current arrangement works best for the normal situation, flying above 100ft AGL, but you're right that it can cause problems when you're buzzing cows. David, What you need to do is set the transparent flag in the ssgSimpleState for these objects. That way plib can sort them and draw them back to front. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
David Megginson writes: Seriously, I am planning on adding the occasional barn, farmhouse, silo, and blue aluminum storage shed. This points to another problem, though -- as with textures, buildings will look very different in different parts of the world and even in different parts of the same country, province, state, or county. What we need is conditions in materials.xml; for example, within a certain lat/lon rectangle, use a tumbled-down stone building for a barn; within another, use a North-American wooden barn; and so on. We already have XML conditions, so it's doable, but it will be a bit of work all the same. Instead of allowing full conditions, we might allow only a few criteria (maybe time of year, lat/lon, elevation, slope). I could see that turning the materials.xml file into a complicated nightmare. However, it would be a nice feature to have. Perhaps the top level materials.xml could include a .xml for each material that has a long list of conditionals. That might keep things a little cleaner ... (?) Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
David Megginson wrote: Over the weekend, I finished my first take on dynamically-placed scenery objects. I've tried it and it looks quite nice. It seems to have a very low footprint (concerning framerate) so that's excelent as well. The only problem I have is it either leaks memory like crazy (about 4Mb per second) or it ernlarges the FlightGear memory footprint by more than 225 Mb. :-( I've looked at the code but couldn't find a place where memory leaks might happen. Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Erik Hofman writes: David Megginson wrote: Over the weekend, I finished my first take on dynamically-placed scenery objects. I've tried it and it looks quite nice. It seems to have a very low footprint (concerning framerate) so that's excelent as well. The only problem I have is it either leaks memory like crazy (about 4Mb per second) or it ernlarges the FlightGear memory footprint by more than 225 Mb. :-( I've looked at the code but couldn't find a place where memory leaks might happen. I suspected from all the extra disk activity that some sort of memory leaking might go on, but I only had a couple minutes to play last night so I didn't get a chance to take a closer look. Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
We need a farmer and a tractor for the farm fields here in MN. If the farmer's name was Nic then we could have a Scenic overlook, [...] Hehe, we're also having really new sort of trouble with these trees - like having one standing right in front of the runway: http://document.ihg.uni-duisburg.de/bitmap/FGFS/tree-approach.png Fortunately one can fly through these trees ;-) Martin. -- Unix _IS_ user friendly - it's just selective about who its friends are ! -- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson [EMAIL PROTECTED] said: What we need is conditions in materials.xml; for example, within a certain lat/lon rectangle, use a tumbled-down stone building for a barn; within another, use a North-American wooden barn; and so on. We already have XML conditions, so it's doable, but it will be a bit of work all the same. Instead of allowing full conditions, we might allow only a few criteria (maybe time of year, lat/lon, elevation, slope). That would be useful for defining more realistic shoreline textures, rock colors, etc as well. Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, 16 Jul 2002 20:38, David Megginson entered data into a CPU register that indicated: Actually, you might want to animate these; it would be annoying if the kangaroo always blocked the same runway. Of course! Mind you in Africa the the baboons will sit on a runway, fly away as you buzz them, then sit back on the runway while you go around the circuit for landing. They are happy to do that all afternoon, so you just have to land and try and miss them. :-) We should be able to implement this, although you'll have to watch that the baboons don't bite you on the backside as you get out of the plane. :-P David - -- - -BEGIN GEEK CODE BLOCK- Version: 3.12 GCC d- s:-- a--- C++ UL P+++ L+++ E- W++ N++ o- K- w-- O-- M+ V-- PS PE Y+ PGP+++ t+ 5-- X-- R- tv- b++ DI D--- G e h! !r y- - --END GEEK CODE BLOCK-- -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE9NAvbZOfFgbBAbXARAsRtAJ9xNRAW5almaZdRQjCGzX9WXDx3rQCfSGgS g94q3+hAQ15APykdKAhEevk= =n35T -END PGP SIGNATURE- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Findlay [EMAIL PROTECTED] wrote: Of course! Mind you in Africa the the baboons will sit on a runway, fly away as you buzz them, then sit back on the runway while you go around the circuit for landing. They are happy to do that all afternoon, so you just have to land and try and miss them. :-) Yes, and then not even to mention the problems that animal life causes for aviators. ;-) Hey, as a native of Africa I'm allowed to make these non-pc jokes. Besides, where else in the world will someone actually steal a NDB but in Cape Town? :-) We should be able to implement this, although you'll have to watch that the baboons don't bite you on the backside as you get out of the plane. :-P Maybe a bite on the backside is exactly what the US legislators and media need for their treatment of GA in the US. After a solid kick in the nuts. Er.. make that several kicks. ObOnTopic: Dave (the other one :-), are you considering dynamic moving objects? You mentioned yachts in passing... Collision detection eats CPU cycles, but still without that dynamic objects IMO are relegated to mainly eye candy. -- Billy ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Curtis L. Olson writes: I could see that turning the materials.xml file into a complicated nightmare. However, it would be a nice feature to have. Perhaps the top level materials.xml could include a .xml for each material that has a long list of conditionals. That might keep things a little cleaner ... (?) Sure. We can do that already, if someone wants to go to the trouble; just create an $FG_ROOT/Materials/ directory, put one file for each material in it, then make materials.xml look like this: PropertyList material include=Materials/grass_rwy.xml/ material include=Materials/dirt_rwy.xml/ material include=Materials/lakebed_taxiway.xml/ ... /PropertyList It will look exactly the same to FlightGear, so no C++ modifications will be required. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Erik Hofman writes: I've tried it and it looks quite nice. It seems to have a very low footprint (concerning framerate) so that's excelent as well. The only problem I have is it either leaks memory like crazy (about 4Mb per second) or it ernlarges the FlightGear memory footprint by more than 225 Mb. :-( I've looked at the code but couldn't find a place where memory leaks might happen. Do you see this sitting still or flying across tile boundaries? I tried running FlightGear with and without dynamic objects, and sitting still at the default location, RSS stabilized quickly at 79MB without dynamic objects and 106MB with, and stayed there for over a minute in each case. I'm using G++ 3.0. The 27MB inflation is, of course, still unacceptable. I can think of some ways to decrease it by quite a bit when I have time: 1. Make sure that objects used by multiple materials are loaded only once. 2. Do some callback trickery to cull the SSG nodes for dynamic objects when they are out of range (i.e. LOD keeps them from being rendered, but we don't even want them to stay in memory). Otherwise, the branch nodes for objects are kept in memory until the tile leaves the cache. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Martin Spott writes: We need a farmer and a tractor for the farm fields here in MN. If the farmer's name was Nic then we could have a Scenic overlook, [...] Hehe, we're also having really new sort of trouble with these trees - like having one standing right in front of the runway: http://document.ihg.uni-duisburg.de/bitmap/FGFS/tree-approach.png Fortunately one can fly through these trees ;-) This is where we need to make TerraGear smarter. We've always had forest right up to the threshold of many runways, but as long as it was a flat texture, nobody minded much. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Findlay writes: Of course! Mind you in Africa the the baboons will sit on a runway, fly away as you buzz them, then sit back on the runway while you go around the circuit for landing. They are happy to do that all afternoon, so you just have to land and try and miss them. :-) We should be able to implement this, although you'll have to watch that the baboons don't bite you on the backside as you get out of the plane. :-P In Canada, a bull moose will charge at a car's headlights or, presumably, a small plane's landing lights. Moose have very large bodies on high legs and can do a lot of damage, and a few drivers die every year from encounters. Standard driving practice when you see a moose (in time) is to turn off your headlights, stop, and wait very quietly. Natural selection tends to take care of the idiots who honk their horns and flash their lights to try to get a moose off the road. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Billy Verreynne writes: Dave (the other one :-), are you considering dynamic moving objects? You mentioned yachts in passing... Collision detection eats CPU cycles, but still without that dynamic objects IMO are relegated to mainly eye candy. I think that those will need to be handled differently. Static objects are attached to a tile, and will be freed when the tile falls out of the cache; moving objects can cross tile boundaries. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson wrote: Seriously, I am planning on adding the occasional barn, farmhouse, silo, and blue aluminum storage shed. This points to another problem, though -- as with textures, buildings will look very different in different parts of the world and even in different parts of the same country, province, state, or county. What we need is conditions in materials.xml; for example, within a certain lat/lon rectangle, use a tumbled-down stone building for a barn; within another, use a North-American wooden barn; and so on. We already have XML conditions, so it's doable, but it will be a bit of work all the same. Instead of allowing full conditions, we might allow only a few criteria (maybe time of year, lat/lon, elevation, slope). IMO it's better to use a lat/lon and a weighting value instead of a rectangles. So you can e.g. place a all farms look like Object1 at lat/lon, standard weighting in the middle of the US and place a all farms look like Object2 at lat2/lon2, strong weighting at a special place where the farms look a bit different. The result would be, that all farms are Object1, exept in a circle around lat2/lon2 where they are Object2. The decicion about which object should be generated is easy, just look which weight/distance_to_lat_lon is gratest... CU, Christian PS: This will automatically generate voronoii diagrams... -- The idea is to die young as late as possible.-- Ashley Montague ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson wrote: Erik Hofman writes: I've tried it and it looks quite nice. It seems to have a very low footprint (concerning framerate) so that's excelent as well. The only problem I have is it either leaks memory like crazy (about 4Mb per second) or it ernlarges the FlightGear memory footprint by more than 225 Mb. :-( I've looked at the code but couldn't find a place where memory leaks might happen. Do you see this sitting still or flying across tile boundaries? I Both. I even renamed the Textures.high directory to give me some more memory (the O2 processes the textures from main memory because it doesn't have any video memory). This gives the same result. It looks like it stabilizes around 50Mb free memory but after about ten seconds it starts to count down again, until all the (180 Mb) swap is used. tried running FlightGear with and without dynamic objects, and sitting still at the default location, RSS stabilized quickly at 79MB without dynamic objects and 106MB with, and stayed there for over a minute in each case. I'm using G++ 3.0. Irix 6.5.14m and MipsPro 7.2.1.3 The 27MB inflation is, of course, still unacceptable. I can think of some ways to decrease it by quite a bit when I have time: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson writes: Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). David, Your cows are hideous but humorous. On my computer they are missing the head. :-) In real life you'd likely see clustered groups of cows rather than randomly spread out over a field. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
David Megginson wrote: Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). Ouch, that's too much for me. I've 192 Mb internal memory which is shared with the video adaptor (and textures). Erik (Does anybody have an Octane/MXE to give away?) :-) ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson writes: Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). David, Right now you have the individual object LOD node attached to the model so there is only one copy of it in memory. It might be interesting to make a per instance LOD node above each instance of the object so that we could vary the range for each instance of the object. This would allow us to see just a few of the random objects way out in the distance, but not too many to kill frame rates. lunch passes Ok, I think I have something working that I like: I moved the code that creates the object LOD node over to obj.cxx where the object is actually positioned. Now every instance of every 'random' object has it's own unique LOD node. The LOD range is calculated as range/(random(0,1)*range) + range. So if the specified object range was 2000m you would get a distribution that follows the form of 1/x. The minimum range would be 2000m, but a few would be much higher than that. The visual result is that you'd get a few objects way out in the distance and then others would fill in between more and more as you got closer in. It looks very similar to what you had originally but with a bit more variety and a few interesting objects in the distance, and there isn't a 'hard line' where all the objects pop in. You still see popping but it is maybe a slight fraction less noticable. I think this change makes the random objects look really good. Do you mind if I commit them (you can always back them out if you don't like them, or think of a better way do do this.) Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Erik Hofman writes: David Megginson wrote: Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). Ouch, that's too much for me. I've 192 Mb internal memory which is shared with the video adaptor (and textures). Erik (Does anybody have an Octane/MXE to give away?) Perhaps we could add a property that specifies a percentage of the random objects to create. Those with lower end systems could turn down the amount of random objects. Those with insane amounts of power, could turn up the number ... Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Curtis L. Olson wrote: Erik Hofman writes: Ouch, that's too much for me. I've 192 Mb internal memory which is shared with the video adaptor (and textures). Erik (Does anybody have an Octane/MXE to give away?) Perhaps we could add a property that specifies a percentage of the random objects to create. Those with lower end systems could turn down the amount of random objects. Those with insane amounts of power, could turn up the number ... That might be a good idea. In the mean time I'll see if I can handle 2 dynamic objects (Just to be sure). Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Curtis L. Olson writes: Your cows are hideous but humorous. On my computer they are missing the head. :-) If you're flying close enough to notice, you're flying too low (I've been waiting almost 24 hours to make that comment). In real life you'd likely see clustered groups of cows rather than randomly spread out over a field. Ditto for trees -- you'll tend to see stands of trees of the same type grouped together rather than, say, pine and maple mixed randomly. Any suggestions? I suppose I should make some sheep for our friends in the UK and Eire -- as I recall, sheep account for about 75% of the scenery over there. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson writes: Curtis L. Olson writes: Your cows are hideous but humorous. On my computer they are missing the head. :-) If you're flying close enough to notice, you're flying too low (I've been waiting almost 24 hours to make that comment). I want to see the fear in their eyes as they scatter. But seriously, I really like cows, I would never try to scare them like that. :-) In real life you'd likely see clustered groups of cows rather than randomly spread out over a field. Ditto for trees -- you'll tend to see stands of trees of the same type grouped together rather than, say, pine and maple mixed randomly. Any suggestions? Tough problem. I think it will be likely that if we go beyond something simple, we'd be better off letting a human place the trees (or come up with some sort of image processing util that can pick off every tree position from a aerial/satellite image.) I suppose I should make some sheep for our friends in the UK and Eire -- as I recall, sheep account for about 75% of the scenery over there. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Erik Hofman writes:
Ouch, that's too much for me. I've 192 Mb internal memory which is
shared with the video adaptor (and textures).
I've figured out the solution, but will need some time to implement
it.
Right now, I have top level transformation and range nodes for groups
of objects on every surface -- that way, if a tri-fan (for example) is
more than, say, 10km away, all of the objects on it will be skipped,
saving hundreds or thousands of individual LOD tests.
My plan is to extend the ssgRangeSelector to have an additional member
for over the maximum distance, so the range array would look something
like this:
float ranges = {0, 1f, f};
Under the ssgRangeSelector, I will put two empty ssgBranch nodes, one
for within the range (i.e. =1m) and one for outside the range
(i.e. 1m). Each of these will have a reference to the same user
data structure containing information about the objects should should
be generated and a flag indicating whether the objects currently
exist.
I will then add a pre-traversal callback to each branch that examines
the user data. The callback for the in-range branch will check
whether the dynamic objects currently exist; if not, it will generate
them, add them to itself, and set the flag in the user data. The
callback for the out-of-range branch will check whether the dynamic
objects currently exist; if so, it will remove them from the in-range
branch (so that their memory is freed) and clear the flag in the user
data. As surfaces move out of the user-specified range for dynamic
objects, their memory will be freed automatically, and then
reallocated later if needed.
Any comments or suggestions?
All the best,
David
--
David Megginson, [EMAIL PROTECTED], http://www.megginson.com/
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Curtis L. Olson writes: Right now you have the individual object LOD node attached to the model so there is only one copy of it in memory. Right -- so the memory problem is only the transform nodes. It might be interesting to make a per instance LOD node above each instance of the object so that we could vary the range for each instance of the object. This would allow us to see just a few of the random objects way out in the distance, but not too many to kill frame rates. This is a great idea, but it will make the memory problem even worse, at least until I have a chance to implement the suggestion in my last posting. I moved the code that creates the object LOD node over to obj.cxx where the object is actually positioned. Now every instance of every 'random' object has it's own unique LOD node. The LOD range is calculated as range/(random(0,1)*range) + range. So if the specified object range was 2000m you would get a distribution that follows the form of 1/x. The minimum range would be 2000m, but a few would be much higher than that. That sounds like an excellent idea. The visual result is that you'd get a few objects way out in the distance and then others would fill in between more and more as you got closer in. It looks very similar to what you had originally but with a bit more variety and a few interesting objects in the distance, and there isn't a 'hard line' where all the objects pop in. You still see popping but it is maybe a slight fraction less noticable. I think this change makes the random objects look really good. Do you mind if I commit them (you can always back them out if you don't like them, or think of a better way do do this.) Sure, you can either commit it now, before I start working on the memory problem, or wait a day or two (or more) for my restructuring. Thanks, and all the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David Megginson writes: This is a great idea, but it will make the memory problem even worse, at least until I have a chance to implement the suggestion in my last posting. I moved the code that creates the object LOD node over to obj.cxx where the object is actually positioned. Now every instance of every 'random' object has it's own unique LOD node. The LOD range is calculated as range/(random(0,1)*range) + range. So if the specified object range was 2000m you would get a distribution that follows the form of 1/x. The minimum range would be 2000m, but a few would be much higher than that. That sounds like an excellent idea. I tweaked the random distribution a bit, but feel free to play with the formula and coefficients. The visual result is that you'd get a few objects way out in the distance and then others would fill in between more and more as you got closer in. It looks very similar to what you had originally but with a bit more variety and a few interesting objects in the distance, and there isn't a 'hard line' where all the objects pop in. You still see popping but it is maybe a slight fraction less noticable. I think this change makes the random objects look really good. Do you mind if I commit them (you can always back them out if you don't like them, or think of a better way do do this.) Sure, you can either commit it now, before I start working on the memory problem, or wait a day or two (or more) for my restructuring. Ok, before I forget all about the changes, I have committed them. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
David, The randomized objects are very cool, but I'm seeing one problem on longer flights. After flying maybe 100 miles or so (?) the tile loader locks up and the entire program hangs. I haven't had a chance to debug this, but you might try upping the acceleration factor and go for a quick, but longer flight and see if you can replicate this. It doesn't appear to be a problem with RAM, but appears as if the program goes into an infinite loop somewhere ... This does not happen with the random objects turned off. Regards, Curt. David Megginson writes: Over the weekend, I finished my first take on dynamically-placed scenery objects. I'm attaching two screenshots: 1. Approaching a city over some woods, with a pasture in-between. http://www.megginson.com/flightsim/dynamic-01.png 2. Sitting on a lake, pretending to be a floatplane. http://www.megginson.com/flightsim/dynamic-02.png This is not, primarily, designed to be eye candy (well, only a bit of it is). Dynamic objects like buildings and trees serve several very important purposes during VFR flight: a. They give some indication of altitude AGL and groundspeed. Even at low altitude Trees get very small very fast; if they're big and there's not a runway in your windshield, add power and pull up. b. They provide aiming points for turns and straight flight (i.e. to turn 90 degrees, look off your wing, pick and object, and then fly towards it). For example, it is much easier to hold a compass heading if you turn to the heading, pick an object off the nose, then keep that object off the nose than if you constantly chase the compass around. c. (not so nice) They cause very confusing illusions because of drift when flying near the ground. It is easy for a pilot to get into a spin or spiral by banking too steeply because the plane doesn't seem to be turning much. Right now, I've added only a few dynamic object types, so it's a bit monotonous. Over the next week or two, I'll try to introduce a greater variety of buildings and trees, at least, and maybe a few pure eye-candy things like boats out in lakes. All non-billboarded dynamic objects also currently face the same way; I plan to fix that in the next day or two. I've made very aggressive use of LOD to keep the framerates up; for example, trees are visible only from a range of 2000m and large buildings from 1m; I also slap a range on all the objects on each triangle or fan, so that plib won't have to do hundreds or thousands of LOD calculations for objects that are far out of range. Even then, trees cannot be too dense. NOTE: Dynamic objects are off by default. After you update the FlightGear and base-package CVS, you will have to set the property /sim/rendering/dynamic-objects to true to see anything (if you change it during the program run, you'll have to reload the scenery). Depending on the terrain, visibility, and altitude AGL, you may experience slow framerates for a few seconds while FlightGear places dynamic objects on all the tiles; just wait, and everything will speed up again shortly. Special thanks to Curt, whose ground-lighting code I stole^H^H^H^H^Hused as inspiration to calculate object placement. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Curtis L. Olson writes: The randomized objects are very cool, but I'm seeing one problem on longer flights. After flying maybe 100 miles or so (?) the tile loader locks up and the entire program hangs. I haven't had a chance to debug this, but you might try upping the acceleration factor and go for a quick, but longer flight and see if you can replicate this. It doesn't appear to be a problem with RAM, but appears as if the program goes into an infinite loop somewhere ... Ditto. It could be that very small triangles are causing problems, or it could be that we're just creating too many nodes. I'll keep playing around and see what I can find. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
Hello, first I would say that dynamic objects are very cool. But I can testify that it eats a lot of memory and it leaks it. I am under Win2k, compile with MSVC a debug version. On startup it takes 586 Mb (read Mega) but stay still while sitting on the runway (LFLK is surrounded by forest of SomeSort material ). Then I take off and engage AP, flying A4. It then leaks 80Mb from time to time, saturating the 1Gb of ram and begins to take swap memory. I stop it before all my hardisk eaten by the Windows PageFile. CPU is 100% of my Athlon XP1800+ before beginning swaping. By comparision, startup consumes 220Mb without dynamic objects and no more memory is reclaimed after. CPU is about 50-60%. Cheers, -Fred - Original Message - From: Erik Hofman [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Tuesday, July 16, 2002 7:29 PM Subject: Re: [Flightgear-devel] ANN: a new dimension to FlightGear David Megginson wrote: Erik Hofman writes: If it would be 27Mb only, there wouldn't be a problem for me because FlightGear without dynamic objects leaves 39Mb spare memory. The problem seems to be that there are a lot of extra SSG nodes attached to every tile in the cache (one ssgTransform and one ssgRange for every dynamic object, whether currently visible or not). The solution, I think, will involve using callbacks under the group LOD node to manage things so that SSG nodes are created only when needed. During a high-speed cross-country magic carpet ride, the memory usage got as high as 180MB before stabilizing (vs. 80MB without dynamically-placed objects). Ouch, that's too much for me. I've 192 Mb internal memory which is shared with the video adaptor (and textures). Erik (Does anybody have an Octane/MXE to give away?) :-) ___ 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
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
At 01:36 PM 16/07/2002 -0400, you wrote: Curtis L. Olson writes: Your cows are hideous but humorous. On my computer they are missing the head. :-) If you're flying close enough to notice, you're flying too low (I've been waiting almost 24 hours to make that comment). In real life you'd likely see clustered groups of cows rather than randomly spread out over a field. Ditto for trees -- you'll tend to see stands of trees of the same type grouped together rather than, say, pine and maple mixed randomly. Any suggestions? FS2K2 solves this problem with using the placement info in the textures. All you need is a 2bit texture where bit 1 means place an object here and bit 0 means no object here. You'd need these masks for each ground texture you have. Placing the objects manually on top of each texture is not that much work, unless you really have a lot of textures. It also has the advantage that 3D trees are above the trees in the image... I suppose I should make some sheep for our friends in the UK and Eire -- as I recall, sheep account for about 75% of the scenery over there. Please don't forget New Zealand (probably about 90% :) )!!! Cheers, Christian ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] ANN: a new dimension to FlightGear
I dont think the alpha sorting code was ever comitted, so currently I dont beleive PLIB will alpha sort. -Original Message- From: Curtis L. Olson [mailto:[EMAIL PROTECTED]] Sent: Tuesday, 16 July 2002 9:10 PM To: [EMAIL PROTECTED] Subject: Re: [Flightgear-devel] ANN: a new dimension to FlightGear David Megginson writes: David Findlay writes: Just noted one problem. The billboard backgrounds are transparent only for the ground, not other objects. This doesn't look great when you're buzzing through the forest. Other than that it looks great. Thanks, Noted. Unfortunately, alpha transparency is a nightmare -- only things drawn *before* the object with transparency will show through. I draw the dynamically-placed objects after the ground, so you can always see the ground through them, but they are drawn before the clouds in the sky (since you will want to see them through the gaps in the clouds). It will be (apparently) random whether any individual tree shows up through the alpha portion of any other individual tree's texture. I know of no appropriate solution for this problem. The current arrangement works best for the normal situation, flying above 100ft AGL, but you're right that it can cause problems when you're buzzing cows. David, What you need to do is set the transparent flag in the ssgSimpleState for these objects. That way plib can sort them and draw them back to front. Regards, Curt. -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
Over the weekend, I finished my first take on dynamically-placed scenery objects. I'm attaching two screenshots: 1. Approaching a city over some woods, with a pasture in-between. http://www.megginson.com/flightsim/dynamic-01.png 2. Sitting on a lake, pretending to be a floatplane. http://www.megginson.com/flightsim/dynamic-02.png This is not, primarily, designed to be eye candy (well, only a bit of it is). Dynamic objects like buildings and trees serve several very important purposes during VFR flight: a. They give some indication of altitude AGL and groundspeed. Even at low altitude Trees get very small very fast; if they're big and there's not a runway in your windshield, add power and pull up. b. They provide aiming points for turns and straight flight (i.e. to turn 90 degrees, look off your wing, pick and object, and then fly towards it). For example, it is much easier to hold a compass heading if you turn to the heading, pick an object off the nose, then keep that object off the nose than if you constantly chase the compass around. c. (not so nice) They cause very confusing illusions because of drift when flying near the ground. It is easy for a pilot to get into a spin or spiral by banking too steeply because the plane doesn't seem to be turning much. Right now, I've added only a few dynamic object types, so it's a bit monotonous. Over the next week or two, I'll try to introduce a greater variety of buildings and trees, at least, and maybe a few pure eye-candy things like boats out in lakes. All non-billboarded dynamic objects also currently face the same way; I plan to fix that in the next day or two. I've made very aggressive use of LOD to keep the framerates up; for example, trees are visible only from a range of 2000m and large buildings from 1m; I also slap a range on all the objects on each triangle or fan, so that plib won't have to do hundreds or thousands of LOD calculations for objects that are far out of range. Even then, trees cannot be too dense. NOTE: Dynamic objects are off by default. After you update the FlightGear and base-package CVS, you will have to set the property /sim/rendering/dynamic-objects to true to see anything (if you change it during the program run, you'll have to reload the scenery). Depending on the terrain, visibility, and altitude AGL, you may experience slow framerates for a few seconds while FlightGear places dynamic objects on all the tiles; just wait, and everything will speed up again shortly. Special thanks to Curt, whose ground-lighting code I stole^H^H^H^H^Hused as inspiration to calculate object placement. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
David, Very cool ... it would be nice if we could hide the fade in of the objects a bit better, but I'm not sure if there is an easy way to do that. Vega/performer supports something called fade level of detail which uses alpha to fade in the object (or fade between levels of detail.) But even so, very cool. We need a farmer and a tractor for the farm fields here in MN. If the farmer's name was Nic then we could have a Scenic overlook, he's outstanding in his field. Ok, sorry about that, it was painful to say, even for me. :-) Regards, Curt. David Megginson writes: Over the weekend, I finished my first take on dynamically-placed scenery objects. I'm attaching two screenshots: 1. Approaching a city over some woods, with a pasture in-between. http://www.megginson.com/flightsim/dynamic-01.png 2. Sitting on a lake, pretending to be a floatplane. http://www.megginson.com/flightsim/dynamic-02.png This is not, primarily, designed to be eye candy (well, only a bit of it is). Dynamic objects like buildings and trees serve several very important purposes during VFR flight: a. They give some indication of altitude AGL and groundspeed. Even at low altitude Trees get very small very fast; if they're big and there's not a runway in your windshield, add power and pull up. b. They provide aiming points for turns and straight flight (i.e. to turn 90 degrees, look off your wing, pick and object, and then fly towards it). For example, it is much easier to hold a compass heading if you turn to the heading, pick an object off the nose, then keep that object off the nose than if you constantly chase the compass around. c. (not so nice) They cause very confusing illusions because of drift when flying near the ground. It is easy for a pilot to get into a spin or spiral by banking too steeply because the plane doesn't seem to be turning much. Right now, I've added only a few dynamic object types, so it's a bit monotonous. Over the next week or two, I'll try to introduce a greater variety of buildings and trees, at least, and maybe a few pure eye-candy things like boats out in lakes. All non-billboarded dynamic objects also currently face the same way; I plan to fix that in the next day or two. I've made very aggressive use of LOD to keep the framerates up; for example, trees are visible only from a range of 2000m and large buildings from 1m; I also slap a range on all the objects on each triangle or fan, so that plib won't have to do hundreds or thousands of LOD calculations for objects that are far out of range. Even then, trees cannot be too dense. NOTE: Dynamic objects are off by default. After you update the FlightGear and base-package CVS, you will have to set the property /sim/rendering/dynamic-objects to true to see anything (if you change it during the program run, you'll have to reload the scenery). Depending on the terrain, visibility, and altitude AGL, you may experience slow framerates for a few seconds while FlightGear places dynamic objects on all the tiles; just wait, and everything will speed up again shortly. Special thanks to Curt, whose ground-lighting code I stole^H^H^H^H^Hused as inspiration to calculate object placement. All the best, David -- David Megginson, [EMAIL PROTECTED], http://www.megginson.com/ ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel -- Curtis Olson IVLab / HumanFIRST Program FlightGear Project Twin Cities[EMAIL PROTECTED] [EMAIL PROTECTED] 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] ANN: a new dimension to FlightGear
David Megginson [EMAIL PROTECTED] said: Over the weekend, I finished my first take on dynamically-placed scenery objects. Wow...very nice! Best, Jim ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] ANN: a new dimension to FlightGear
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, 16 Jul 2002 03:33, David Megginson wrote: Over the weekend, I finished my first take on dynamically-placed scenery objects. I'm attaching two screenshots: Special thanks to Curt, whose ground-lighting code I stole^H^H^H^H^Hused as inspiration to calculate object placement. Very nice. I tried to also do this ages ago but couldn't figure it out. :-) What is the framework for creating the objects like? If a good framework API could be developed, it would be easy for scenery designers to add new objects that are locally appropiate. For instance, kangaroos on runways in australia :-). Thanks, David - -- - -BEGIN GEEK CODE BLOCK- Version: 3.12 GCC d- s:-- a--- C++ UL P+++ L+++ E- W++ N++ o- K- w-- O-- M+ V-- PS PE Y+ PGP+++ t+ 5-- X-- R- tv- b++ DI D--- G e h! !r y- - --END GEEK CODE BLOCK-- -BEGIN PGP SIGNATURE- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE9M5DuZOfFgbBAbXARAh3+AJ9AMmMQYRelaXuQzwlrJff6/8C0ogCfZ0IT q3O6yBtxDIrgMkDZss0XK/8= =SGI5 -END PGP SIGNATURE- ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
