By using this site, you agree to our Privacy Policy and our Terms of Use. Close
fatslob-:O said:
JoeTheBro said:

Your link shows FPS in games with 3D on and 3D off. Was there something else at that link I was supposed to notice? Otherwise my comment is just as on topic as your link.

As far as being an expert, I'm obsessed with 3D even more than Kaizar is. So obsessed that I incorporated it into my profession. I'm not going to spill the beans on my credentials since I'd like to stay anonymous at least a little bit longer so if you don't believe me, I can find other sources that say the same thing.

The link was supposed to show how computationally expensive 3D is. It seems you missed the point of that link.

Then I'm sure you know why I explained how computationally expenisve it is. Do you have any explanation as to why it's not very expensive if i'm missing something ?

Sure.

First off it's practically impossible for 3D to take more than double the resources. In a hypothetical situation you could just have two 360s (or any other system really) and feed each the same controller input. I'm of course assuming the two games stay perfectly synced.

If each 360 was programed to have the camera slightly shifted to the side, one left and the other right, the outputs would be perfect for 3D. If you used active shutter glasses then the TV would alternate between each 360 twice a game frame. So if the two 360 outputs were 1080p at 30fps, the TV would be displaying a 1080p at 60fps video. If you used a passive technique like the 3DS each view would lose half of its pixels and the video would only be 1080p at 30fps. This would be perfect stereoscopic 3D with zero optimizations, and it uses 2X the power.

 

However lots of calculations don't depend on the camera's location, at least relative to the other camera. Practically everything except the GPUs are calculating identical information in the two 360 example. So this in itself gets 3D using less than 2X the computational power.

Some of the GPU calculations as well can be shared between viewpoints, but this really depends on how well the dev optimizes the code. Thus it's between 1 and 2 times the computation power, depending on the 3D techniques used.