[Poll] Which technology will be more important to reducing game development costs over the next 3 years of gaming?

[soccerballtux]

Just thought we'd take a look at which technology we think will be more important to the industry. For any lowered production cost, risk of production in a depressed market (where there's no guaranteed sale) will be minimized.

It looks to me like Tesselation will be a huge cost reduction option-- developers won't need to spend so much time modelling the world but will be able to give it a general mold and just randomize/tesselate the rest of it-- similar to Id's latest texturizing (beautiful technology btw check out Quake Wars: Enemy Territory as an example).

 

[lopri]

I think the question is somewhat biased.

 

[jimhsu]

I think the best answer to this is better tools. Fractal geometry tools like Terragen, sculpting tools like ZBrush, procedural texture generation, etc. It's time we move away from hand-editing meshes -- that's so 2000. In that regard, yes tessellation wins.

Some quick demonstrations of how to generate characters in a few hours with ZBrush: http://www.veoh.com/group/mojettetimelapse

 

[sandorski]

Which is going to be Supported by ATI and Nvidia? Therein is your answer.

 

[zephyrprime]

Better tools for sure. I don't expect either Tesselation or PhysX to have any impact on development time. Tesselation will improve visual quality and PhysX will add physics realism but neither will have much impact on development time.

 

[Kakkoii]

Uhh, I don't see the point of this poll... Did you not know that Nvidia will be making a DX11 card? And that in turn will be supporting Tessellation. There's no need to choose lol.

Tessellation doesn't really reduce the cost's, it merely allows for more detail to be run with less hardware. PhysX I guess would a bit, since you don't need to create animations for a lot of things anymore. Just let the physics handle it. But PhysX will probably die out, unless Nvidia ports it to OpenCL/OpenGL.

 

[Idontcare]

Originally posted by: Kakkoii
Uhh, I don't see the point of this poll... Did you not know that Nvidia will be making a DX11 card? And that in turn will be supporting Tessellation. There's no need to choose lol.

I don't see where the OP is implying game developers had to make a choice regarding Tessellation vs. physx, he's asking which of the two is going to be more important to lowering game development cost in the next three years.

 

[mmnno]

PhysX is middleware, and tessellation is a new hardware capability. Generally costs go up as technology advances, so it's probably too much to hope that tessellation will reduce costs.

 

[akugami]

Originally posted by: mmnno
PhysX is middleware, and tessellation is a new hardware capability. Generally costs go up as technology advances, so it's probably too much to hope that tessellation will reduce costs.

You're right about new technologies increasing the costs of developing games. In the short term, tessellation will more than likely increase costs. However, costs should decrease due to savings that tessellation allows.

Tessellation should allow you to get effects done faster and that will allow savings because the less time needed to do something, the more you save. These savings can be used to add polish to a game and increase other areas such as the story and hopefully gameplay and game AI. Efficiency is something that's sorely needed in the games development field. Middleware will contribute to the savings down the road as well.

The whole general movement from DX9 to DX10 and DX11 has been more about efficient use of available resources than about true graphical improvement IMHO. Granted this does have the benefit of allowing you to display better graphics because of the efficiency. While DX10 has been somewhat slow on the uptake due to Vista being perceived as a flop, I think DX11 will be much more supported because Win7 has been generating a lot of positive buzz.

 

[RavenGuard]

Tesselation is not really a "new" technology, the Xbox 360 has a tesselator, and ATi cards since the 2000 series had tesselators.

I Agree that tesselation won't technically make development easier and cheaper, but it is possible because it allows developers to get a better result with less, though it's place is likely to efficiently improve graphics that are already pretty good to begin with.

 

[Idontcare]

Originally posted by: mmnno
PhysX is middleware, and tessellation is a new hardware capability. Generally costs go up as technology advances, so it's probably too much to hope that tessellation will reduce costs.

I think the premise of stating "reducing development costs" means reducing the rate of increase in the game development budget, not say reducing the development cost below that of the budget of prior developed games.

To achieve a game title delivering xyz features and eye-candy, will tessellation enable the same with less cost? What about physx? If both enable same sellable game then which enables a lower development model? (still expected to be higher cost than developing yesteryear's game, but you can't sell yesteryear's game at tomorrow's ASP's so that is kinda a moot comparison)

Something can cost 10% more than its predecessor but still be 5% lower in cost than otherwise projected/budgeted/etc.

 

[novasatori]

of the two i'd say physx, as it makes adding game physics features easier and less time consuming

tessellation may lower hardware costs for better looking graphics, but development costs? i dunno, don't see it really.

its like saying which lowers costs more, licensing havok for your physics engine so you don't have to program it from the ground up or utilizing tessellation in your 3d engine

 

[Modelworks]

Work for a developer is really no different with or without hardware tessellation. If I want to put a model in a game now I have to limit the polygon count to that which I think the very best hardware that my game is targeting can render. Then I use LOD to reduce the polygon count for lower end hardware. That is okay if you are on lower end hardware than what I made the model for, but what if you have better hardware ? Then you are still stuck with the same detail that the game comes with. You could have a GPU twice as powerful but the game would still look the same. Tunnels would still be 8 sided, soda cans on tables would still be 8 sided.

When you add a GPU that can do hardware tessellation , now you can take a game that had 8 sided tunnels and make them perfectly round. The user can decide how detailed he wants the model to be VS the developer deciding based on whatever the medium specd hardware is at the time.

 

[thilanliyan]

Originally posted by: Modelworks
When you add a GPU that can do hardware tessellation , now you can take a game that had 8 sided tunnels and make them perfectly round. The user can decide how detailed he wants the model to be VS the developer deciding based on whatever the medium specd hardware is at the time.

So the game doesn't have to be programmed with tessellation in mind?

 

[DaveSimmons]

PhysX increases cost since the game probably already has a software-based cross-platform physics library.

Eventually Havok will add cross-GPU hardware acceleration (does it already?) or some otherGPU-neutral hardware-software library will allow adding better physics (not PhysX) "for free" but still not reducing cost.

 

[gorobei]

OP you are mistaken about the nature of the technology and the effect it has on production(cost and efficiency).

tessellation allows for seemingly lighter vertex/poly count models to look better(smoother) than the current straight rendered triangles. the final rendered polycount is magnitudes greater than the working mesh that is kept to (theoretically) manageable numbers. the savings in vertex/polygon dont translate into any other savings/reduction in time, effort, money.

- software like zbrush, mudbox, amorphium, etc seem to create impressive meshes in short periods of time, but most of those models are barely ready for production pipeline and most aren't even close to the final base mesh that would be used by the game engine. sculpting packages allow you to add details to properly setup/modeled/uv'd meshes relatively quickly. those physical details can then be stored as texture data, to be re-generated as tesselation surfaces or as normal map lighting(pixel) illusion.
- this doesn't reduce the time needed to model. it just gives the artist additional mechanisms to store the data. if anything, it increases the amount of time by adding in processing time to generate normal maps or displacement maps. the main advantage is that texture artists dont have to paint bump maps, render it to test the results, and then go back to fix the map. also there's less need to re-paint the same crack/scratch/scuff in the color, spec, diffuse maps.
- anyways, these tools already exist and are being used in production pipelines. adding tessellation to the end render engine doesnt change anything other than the way the data is stored. detail and quality will always take time.

 

[Dribble]

Silly poll - both increase costs as both add extra complexity. For results you might as well just have asked are you an Ati or nvidia fan boy?

Originally posted by: DaveSimmons
PhysX increases cost since the game probably already has a software-based cross-platform physics library.

Eventually Havok will add cross-GPU hardware acceleration (does it already?) or some otherGPU-neutral hardware-software library will allow adding better physics (not PhysX) "for free" but still not reducing cost.

That software-base cross-platform physics library probably is physx - it is the most popular software physics library right now. Just cause it supports hw physics doesn't mean you have to use hw physics. Havok is not neutral - it's owned by Intel and you can bet they will use it to maximise benefit to them. Bullet physics is the only neutral one that has plans for opencl hw physics.

 

[Lonyo]

Stupid poll because PhysX will do nothing.
If you had said physics accelerated on the GPU, it would be slightly different, but all PhysX itself can do at the moment is increase costs as developers have to either go with a single level of physics, or implement two lots, one for those with hardware accel. and one for those without.

As has already been mentioned, tesselation won't do much for reducing model complexity, since the models we see, and would see with tesselation, are nothing like what gets made by the art guys. They model to a super high level of detail and scale down, they will just be scaling down to a different level, and be able to give better looking models with reduced performance hit.