How future DCS iterations should look like

[Bucic]

Post #19 says km's have no direct impact on performance. Poly/object count has. You haven't said anything that would state otherwise :)

[winz]

That's just silly wordplay.

By increasing draw distance, you have to increase poly count, if you want to preserve the level of detail. Thus increasing draw distance is directly increasing poly count, which in turn has direct impact on performance.

[Bucic]

That's the vital thing - people think that if FPS has X polycount at Y distance a simulator that needs Y*100 view distance would need X*100 polycount. You may know that it's not the case but it needs to be higlighted. That's why I insist on NOT using distances as any type of argument in similar discussions.

[winz]

No, the poly count inrease would be even larger, because you're not drawing a line, but a cone. At 180 degree FOV it would be around 10 000* more polygons without lods.

Edited July 19, 2011 by winz

[Bucic]

Ah, right! Wrong equation. But my point is still valid. Simulator polycount for a view distance of X km is much lower than FPS polycount for distance of X km. And it's lower by an order of magnitude. I hope I was clear this time.

[EtherealN]

I've seen that argument many times already and I can't shake off the feeling that it's a fallacy. I don't mean to be grumpy here - just getting the facts straight. I'm not a pro graphics guy but AFAIK the size of everything in 3D world is relative and the only thing that counts are: polycount, objectcount etc. You can render the whole DCS world inside a single room and then all it will be is an extremely high-polycount room-sized map.

 

It all depends on how it's done. I'm not saying it's impossible. There are potential ways to create an engine that has Crysis-style graphics up close and still offers a long view range. Problem lies in implementing it, which (so far) no-one has done. If these guys have done it, which it really doesn't look like they have but it's impossible to make a full judgement based on just two short gameplay demo videos.

 

It's all about scalability. I don't think anyone should doubt that it would be possible, it's just having the right people, time and money to make an engine capable of doing it.

 

Note how I never said it's not possible. I said that that's what I want to see before I were to agree that this engine does it "better". If this engine does not have that capability then saying "this is what DCS should look like" can just as well post a few pictures of Crysis 2... It just becomes N/A DCS unless it can both serve such pictures AND serve areas as large as what DCS does. Not impossible, but hard enough that no-one has done it - yet.

 

DCS has a very hard balance to tread. It can't do the same kind of load/unload magic that X-plane and FSX can do, for example, since it has to be prepared to serve more areas than just what's relevant to the player, at full terrain LOD, due to the possibilities of having AI units all over the place. Makes it way harder to get something that's both pretty close-up and will run on modern computers.