differences between the DX9 SDK reference rasterizer and the r300
1) where wood base hit floor the shadow is more to the right ... rather minimal but reference and both nv drivers have it in exact same place it seems
otherwise its basically the same
(yes that was nitpicking lol)
nv31 driver revs...
could it be this "dull lighting" issue is a light not being rendered or being substituted for a quicker to process one ? i pose this question with much ignorance as prob shows
it just seems on that wooden base that the way the light catches the rim(meaning the highest part of the base) that its like a light is missing from affecting that one object
the problem is the pesky animals dont seem affected so doing it for some objects only ?
actually i take that back
the texturing of animals is much more the same in 43 vs reference and in 43.45 they change
number of lights always kills performance i see in benches but maybe not enough for such gains?
sonix666 said:
The animals are exactly the same for the eye except:
However, the animals are almost indistinguishable for the eye when flipping between the images fast. So if nVidia is cheating, then they are doing it very well.
hmm yes apart from the lighjting on the wooden base its certainly quite decent looking... whatever trick they use.... the simple way to deal with it is surely as a driver *optional* optimisation... not exactly the first one.... it seems common enough to have setting along the lines of
blah blah feature.... set it to the left for faster less quality/ to the right for more quality but slower.....
as new games came along and u werent getting the speed u need u might decide its a worthwwhile compromise.... nothing too wrong with a bit of choice....
of course its a nightmare from a "standards" point of view
but what do nvidia care about that.... nothing if they have to soldier on with that architecture for 1.5 years <GULP!>
So no, I think don't there's any magic involved, just pure math and making the right decisions. 