Posted: Wed Jul 04, 2007 9:39 am Post subject:
Importance of deformable terrain ?
i need to know if deformable terrain is important enough to warrant a lot of effort, i'm not talking pavement-placement or something here, but the modification of terrain shape _________________ Micro TS
Portable, no campaign, movies or music, just the engine and needed resources for skirmish and lan. QUICK_EDIT
Without doubt! But IMO there must be a way (maybe with Pavement) to rise or lower a terrain to extend building space. And IMO Pavament should be destroyable sooner or later but not with any weapon (SW only feature?) _________________ Gangster is a Project Perfect Wuj (c)Aro QUICK_EDIT
Joined: 07 Mar 2006 Location: In ur BIOS, Steeln ur Megahurtz!
Posted: Wed Jul 04, 2007 6:20 pm Post subject:
Jabberwoky aka Gangster wrote:
Without doubt! But IMO there must be a way (maybe with Pavement) to rise or lower a terrain to extend building space. And IMO Pavament should be destroyable sooner or later but not with any weapon (SW only feature?)
QFT _________________ Please, read the signature rules of the forum. QUICK_EDIT
Would be a good feature - but, I would leave it to last. I would really (REALLY^99) like it, but there is so much more to do - AI and pathfinding, networking, basic physics engine, interfaces, sound etc. Unless you have an algorithm / method all ready in mind? Also, if there is terrain deformation, we might have to code some simple fluid simulation for water / rivers if the water is at different heights - won't be static as well. QUICK_EDIT
I can't see deformable terrain being all that difficult to implement... Since you're using a heightmap, can't you simply modify it and reload it as necessary? I certainly can't see that being very difficult... _________________ QUICK_EDIT
I can't see deformable terrain being all that difficult to implement... Since you're using a heightmap, can't you simply modify it and reload it as necessary? I certainly can't see that being very difficult...
My major concern was to do with terrain optimisation, which as far as I am aware we are implementing. Redundant tri-faces are taken out of flat terrain as to increase performance, so it is not a simple approach of altering the existing terrain vertices, more vertices have to be added and the the existing ones changed as well. Either we add tri's to the list in real-time, or have the whole terrain unoptimised (like a grid) - have evenly spaced tri's and have an unoptimised terrain mesh for it to be done (not recommended). To do it "right" would require some effort in my opinion if performance is a goal. QUICK_EDIT
Fair enough, though really, that's hardly a problem either, since you could (as long as its fast enough) run the same optimisation on the modified heightmap.
Of course if its not fast enough, I still don't think it would be that much of a problem to just use an unoptimised grid. I'm sure most video card could easily handle it. _________________ QUICK_EDIT
yes they could but if they can when theres also some 40 buildings, 80 tanks and some moving water on the screen to render is another question. QUICK_EDIT
No, I'm still fairly sure that modern video cards can handle this. Lots of games have supported and are coming out that do support deformable terrain, plus lots of fancy graphics that this engine will likely never see. _________________ QUICK_EDIT
AFAIK, most old RTS engines(SAGE,WC3) draw every cell like a grid (without stuff like triangle strips!) and some twice (blending). Unless you want Supreme Commander-like maps, or really detailed heightmap,I think deformation should be easy with just grid terrain, and the deformation part should be more expensive than the more vertices part anyway. You can't get much more from poly optimisation imo.
Just move everything to GPU's memory, map the buffer of the deformable mesh into the client's address space to allow easy modification.
The main problem will be textures. Today's hardware can really handle really many vertices really fast. You can optimize the map mesh by, say, cutting it's vertex count to half, but you still get just some 0.5-3% speedup that allows you to maybe put in 20 more small meshes.
I use 3-time the same vertex count in my engine (no strips), and it's not much slower on 1-3 generations old hardware.
By the time you finish this, OGL2.1/DX10 hardware will be in second, or third generation. Even the first allows you to throw FAR more vertices than you'll ever need into the map and use really detailed meshes at the same time. _________________ Time will tell...
Sooner or later...
Time will tell... QUICK_EDIT
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum You cannot attach files in this forum You can download files in this forum
Importance of deformable terrain ?
