RE: [Flightgear-devel] Billboard animation
David Megginson writes: > > Feel free to play with the sample tree and to compare it to the > non-billboarded tree. I don't know which approach would be better for > an entire forest: > 1. The non-billboarded tree has 18 vertices (or so), but doesn't require any special dynamic transformation. > >2. The billboarded tree has only 4 vertices, but requires a > transformation based on the view matrix. > >Norm: which do you think would be more efficient for a forest of, say, >500 trees? David One of the ideas behind the 'impostor' is that the transform is allready applied to the texture so that unless the 'impostor' needs updating the 'impostor' only needs the 4 vertex calls. Because it is relatively expensive to rerender the 'impostor' only those trees beyond 'some distance' are considered as impostor candidates. IMHO This occasionally recreating the impostor texture and then applying ONLY a texture mapped quad will be significantly faster then either of your 'possibilities' above when using an appropriate 'impostor distance' Note that Impostor textures can be a shared resource too :-) Norman ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Billboard animation
Erik Hofman writes: > Particles (in OpenGL) are a large number of polygons which have the > same characteristics. By updating the particles in one shot you > should be able to get the best perfomance out of the hardware. Do you mean that all of the billboarded trees would be under the same transformation? That wouldn't work, because each tree has to turn to a slightly different angle to face the camera, and each has to rotate about its own centrepoint. 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] Billboard animation
David Megginson wrote: > Erik Hofman writes: > > > > Norm: which do you think would be more efficient for a forest of, say, > > > 500 trees? > > > > I would say, billboarded spherical trees handled as particles. > > I don't understand 'particles'. Particles (in OpenGL) are a large number of polygons which have the same characteristics. By updating the particles in one shot you should be able to get the best perfomance out of the hardware. Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Billboard animation
Erik Hofman writes: > > Norm: which do you think would be more efficient for a forest of, say, > > 500 trees? > > I would say, billboarded spherical trees handled as particles. I don't understand 'particles'. 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] Billboard animation
Use billboarded trees, especially when they rotate around z only, very careful. The funnniest sight I ever had in a flight sim was when I flew directly over a forrest of billboarded trees and (in outside view) looked straight down. You get concentric tree rings that move along at the speed of the airplane. I should have done an animation of that. Bye bye, Wolfram. ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
Re: [Flightgear-devel] Billboard animation
David Megginson wrote: > Norm: which do you think would be more efficient for a forest of, say, > 500 trees? I would say, billboarded spherical trees handled as particles. Erik ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
RE: [Flightgear-devel] Billboard animation
David Megginson writes: >Feel free to play with the sample tree and to compare it to the >non-billboarded tree. I don't know which approach would be better for >an entire forest: > >1. The non-billboarded tree has 18 vertices (or so), but doesn't > require any special dynamic transformation. > >2. The billboarded tree has only 4 vertices, but requires a > transformation based on the view matrix. > >Norm: which do you think would be more efficient for a forest of, say, >500 trees? It probably depends . Hardware assisted OpenGL can be hard to predict But I am guessing you already know the answer, on your machine at least :-) Norman ___ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
[Flightgear-devel] Billboard animation
I've just checked in changes to add a billboard animation to FlightGear 3D models. The billboard animation causes an object (or entire model) to rotate towards the camera about its z-axis and, optionally, its x-axis as well. Here's the XML wrapper for a sample billboarded tree (only 4 vertices) in $FG_ROOT/Models/Trees/billboard-tree.xml: billboard-tree.ac billboard false The 'spherical' property is false (the default) for objects with only cylindrical symmetry, like trees, and true for objects with spherical symmetry, like simplistic clouds. Feel free to play with the sample tree and to compare it to the non-billboarded tree. I don't know which approach would be better for an entire forest: 1. The non-billboarded tree has 18 vertices (or so), but doesn't require any special dynamic transformation. 2. The billboarded tree has only 4 vertices, but requires a transformation based on the view matrix. Norm: which do you think would be more efficient for a forest of, say, 500 trees? 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