» Sun Jul 17, 2011 7:20 pm
As Direct X any version is core code level, the odds of successfully hacking features onto the game kernel are slim and none.
I'm hoping that they are at least putting the code hooks in to allow later development of a replacement executable. Otherwise, this is model they are working from:
Let's build a game for 5 year old out of date hardware, knowing that new hardware is still at least 2 years away, concentrating on the platforms that are utterly non upgradeable and obsolesent heading towards antique, while depriving the section of the community that can actually showcase our technical skill (and is responsible for the true longevity of the particular title) of any advanced features. Knowing that one or more of those 'new platforms' may introduce code or design concepts that totally prevent the game from playing in any acceptable way on Platform whatever. Which would utterly leave the game's longevity at the mercy of the section of the community that we blew off with an attitude of 'you are not important enough'.
That may not be the attitude prevalent at BGS, but the simple truth is that in statements and tone of communications, that is what it -reads and sounds- like at times.
There are many good technical reasons why implementing DX-11 level coding would be a very good thing. First and foremost, I personally get tired of everyone waiting for the bloody shooters to try new tech before -they- consider it. There are many descriptive adjectives I can use to describe that behavior, but let's not get banned. Daggerfall was the odd child because it, a lowly CRPG, took that tech envelope and stretched it beyond a lot of what the shooters were capable of doing. In 1996. And in 2011, people =STILL= play that old DOS game and love it. That kind of loyalty is priceless in any endeavour, and worth far more than any monetary cost to achieve it, as word of mouth beats and trumps the slickest ad campaign ever devised.
Shader Model 5 is superior to the previous versions. Better water, terrain, color control, blending, mapping with normal maps, bump maps, light maps, shadow maps, etc. All the little tricks that are used to speed up game renderers. Then you get into things like the Compute shader, that let's you use your GPU as secondary processing power for your system. That would let you have a scaleable physics setup for example; a simplified one for weaker systems, and added features for the capable (ragdoll vs. collision accurate).
Tesselation. Here's a good example; dragon wings. Model the wings with accurate boning, then make the membrane low poly (like say 3-6 per section). Have the tesselator push that to a few thousand each section. At runtime, and solely in the video card environment. All your computer sees is the low poly wings it has in its memory,,,,,,leaving all that room for other things there. Drive the tesselated wings with an animated displacement map, and you get softbody like behavior without the penalty dynamic softbody calcs inflict. The amount of memory it will free from polygon heavy modeling is....shall we say....significant?
Far better lighting models supported.....up to HDRI.
DX-11 cards universally have much wider processing pipelines than DX-9 hardware, meaning more room to offload CPU based effects and modes onto the GPU without inflicting slowdowns.
Unless someone deliberately goes low end, any new computer is at least a dual core, most quad. That kind of power is still mostly untapped, as it has to be designed in from the get go.
I hope we get surprised.......but I fear we shall not be........