Does trueform have a future?
- Thread starter Galilee
- Start date
I think it is a smart feature that allows a developer to show more in his art work and still allow compatibility for those who don't have the hardware to handle it. I don't think it is the be-all do-all feature but it does increase IQ noticeably and for displacement mapping it will become a necessity.
This feature is not even a year old some of you are calling it dead? Where new games already use this feature within months of its release
. Well I will promote anything that increases IQ and especially when it is so easy to do.
This feature is not even a year old some of you are calling it dead? Where new games already use this feature within months of its release
. Well I will promote anything that increases IQ and especially when it is so easy to do.I voted no. While it's certainly cool and useable for some application I don't think it'll be widely supported. Displacement mapping will take it's place more or less completely.
Also, most future games will use a large number of polygons anyway, either that or they'll use displacement mapping. And some kinds of data simply can't be interpolated, such as indices, which makes it incompatible with for instance matrix palette skinning.
Also, most future games will use a large number of polygons anyway, either that or they'll use displacement mapping. And some kinds of data simply can't be interpolated, such as indices, which makes it incompatible with for instance matrix palette skinning.
In the long run yes, but for what the typical card owner has as in MX/Radeon level, a many polygon game would be unplayable and a killer for game sells. It is a good solution to be able to get great looking models and yet play on most hardware for the next 3 years. I don't think displacement mapping will take its place due to the utter lack of hardware capable machines in use presently. TRUFORM can be incorporated now and not break anyones graphic card capability. I see it stupid not to use TRUFORM or N-patches in games except maybe where it interferes with unique game engines like DOOMIII.
I don't think think TRUFORM has an future.
TRUFORM is only supported by ATi and even ATi doesn't support it fully, otherwise the R9000-series would have TRUFORM in hardware too. The installed base of R8500 and R9700 is really small then.
This means that this feature will never have the critical mass to justify the additional development work for an game. Maybe a few developers will include it, when they have time left (which nearly never happens).
TRUFORM is only supported by ATi and even ATi doesn't support it fully, otherwise the R9000-series would have TRUFORM in hardware too. The installed base of R8500 and R9700 is really small then.
This means that this feature will never have the critical mass to justify the additional development work for an game. Maybe a few developers will include it, when they have time left (which nearly never happens).
What addtional developer work has to be done? If the models are done right then TRUFORM isn't really an extra cost except supposenly a few extra lines for each model. In addition your program keeps on working on all classes of video card hardware and looks the best with the best hardware available. The Radeon 9000 is a sub $100 card, TRUFORM aware programs will work just fine on it but without the enhancement. I don't see a valid reason why a developer wouldn't use TRUFORM to show case their game off unless it will conflict with how they want to render something as in J.C. DoomIII engine. It isn't a time consuming ordeal at all as far as I know. It is supported in the API and should be used until the general hardware can support something better.
But why do you need truform in the first place? Most (everyone I know) artists will make a high poly model and then drop polys to make things run OK. There is no problem to ship a game with couple meshes for each model (ala mip mapping).
Truform is just a last minute solution for low poly problems. You don't need to use it if your game will ship with models that are highly detailed (Unreal 2). You can only make use of it if your models are still low poly (and I don't know why should they still be? We have hardware t&l for 2 years now). Even on GF 2 MX or Radeon 7500 t&l speed is not a problem...
Displacement mapping and other stuff coming down on us are much different - much more flexible.
Truform is just a last minute solution for low poly problems. You don't need to use it if your game will ship with models that are highly detailed (Unreal 2). You can only make use of it if your models are still low poly (and I don't know why should they still be? We have hardware t&l for 2 years now). Even on GF 2 MX or Radeon 7500 t&l speed is not a problem...
Displacement mapping and other stuff coming down on us are much different - much more flexible.
I don't think it ever really had much of a future. Even upon release I never had any chance of really being supported. If it was supported on both ATi and Nvidia cards, then it might have seen some use. But even if ATi were to have 50% market share (closer to 30% I think?), not all of that is R8500/R9000/R9700 and most developers just won't be bothered (even if it's not that hard).
nooneyouknow
Newcomer
Many points:
1 - It is still being supported in games. Expect a new title to announce support by the end of the month.
2 - Matrox also supports it (it is called n-patches in DX lingo and the Parhelia supports it)
3 - Displacement Mapping NEEDS it to do the tesselation.
'Nuff said...
1 - It is still being supported in games. Expect a new title to announce support by the end of the month.
2 - Matrox also supports it (it is called n-patches in DX lingo and the Parhelia supports it)
3 - Displacement Mapping NEEDS it to do the tesselation.
'Nuff said...
Well, first of all, because of displacement mapping, most new video cards are going to support N-patches, at least in hardware.
Personally, I really don't see any need for N-patches...they are just a shoddy subdivision surface implementation. Perhaps if there was some better interpolation done, they would be more viable. Right now the problems (and the need to fiddle with meshes) just seems like too much hassle for most game developers to bother.
In particular, N-patches will almost certainly be done away with in lieu of more flexibile HOS techniques. Since it appears that RT-patches will never see support (apparently game developers didn't like them as much as I thought they might), and Displacement Mapping is only good for rather specific instances, it may be a short while before we see a "good enough" HOS technique. Perhaps the rumored "Primitive Processor" that didn't quite make it to the NV30 will fill that gap. I'm willing to bet that something similar will.
In sum, I feel that N-patches will end up like EMBM. It will be sort of an, "Hrm, that looks kinda cool." feature for a little while, until something better comes along. After all, every mid-high end video card out today supports EMBM...but does anybody use it?
Update: Despite hardware support for N-patches, whether or not that support will be exposed is as yet unknown. As an aside, does anybody know if the N-patches support is exposed in the Parhelia's drivers at the moment?
Personally, I really don't see any need for N-patches...they are just a shoddy subdivision surface implementation. Perhaps if there was some better interpolation done, they would be more viable. Right now the problems (and the need to fiddle with meshes) just seems like too much hassle for most game developers to bother.
In particular, N-patches will almost certainly be done away with in lieu of more flexibile HOS techniques. Since it appears that RT-patches will never see support (apparently game developers didn't like them as much as I thought they might), and Displacement Mapping is only good for rather specific instances, it may be a short while before we see a "good enough" HOS technique. Perhaps the rumored "Primitive Processor" that didn't quite make it to the NV30 will fill that gap. I'm willing to bet that something similar will.
In sum, I feel that N-patches will end up like EMBM. It will be sort of an, "Hrm, that looks kinda cool." feature for a little while, until something better comes along. After all, every mid-high end video card out today supports EMBM...but does anybody use it?
Update: Despite hardware support for N-patches, whether or not that support will be exposed is as yet unknown. As an aside, does anybody know if the N-patches support is exposed in the Parhelia's drivers at the moment?
Definitely a very good idea with a lot of obvious benefits. I think it'll be replaced in the near future by GPU's with programmable primitive processors. The idea of subdivision on the GPU makes a lot of sense especially as GPU clocks get increasingly higher and bus bandwidth does not keep up. It's a pioneering technique for something that will see widespread use in the future -- albeit in a more flexible fashion.
Hellbinder
Banned
Its all really great that you have all decided that Trueform is a *wate of time* and all.......
But ATI says that 100 PC games comming out in 2003 will have *native* Trueform support. It *is* an intelligent and usefull technology and Displacement mapping is *not* a replacement for it.
You mean like the *rumored* Nvidia toy? From what i have read is that a tech like that will see its biggest beneffit dynamic Terrain.
But ATI says that 100 PC games comming out in 2003 will have *native* Trueform support. It *is* an intelligent and usefull technology and Displacement mapping is *not* a replacement for it.
You mean like the *rumored* Nvidia toy? From what i have read is that a tech like that will see its biggest beneffit dynamic Terrain.
The point of a primitive processor is to allow any geometry amplification mechanism to be used by making it programmable. Right now, n-patches, rt-patches, et al are similar to older hardware's fixed T&L and fixed texture blending modes.
Allowing the GPU to be programmed with the tesselation algorithm of choice allows flexible geometry compression/amplification and allows it to work better with algorithms that require generation of additional scene geometry like shadow volumes.
How can you call it a "toy" when you no nothing about it?
Allowing the GPU to be programmed with the tesselation algorithm of choice allows flexible geometry compression/amplification and allows it to work better with algorithms that require generation of additional scene geometry like shadow volumes.
How can you call it a "toy" when you no nothing about it?