Comment by MBCook
The second revision with the better “head tracking” screen (I don’t remember how it worked) was much better than the launch model.
But mostly, you’d just turn 3D off. It was a gimmick, cut the resolution in half, and doubled processing requirements.
It just worked way better as a 2D screen. As you said, 2DS was good.
They should have just made the DS2, not gone for 3D. But such was the time.
> The second revision with the better “head tracking” screen (I don’t remember how it worked) was much better than the launch model.
It worked with a sheet of lenticular lenses on top of the screen, which separated the odd and even pixel columns for the left and the right eye. The sheet could also be moved a tiny amount by a motor, to change the "center" of the separated light, depending on whether the head was tracked more to the left or more to the right.
> But mostly, you’d just turn 3D off. It was a gimmick, cut the resolution in half, and doubled processing requirements.
Well, it did cut the resolution in half, but this meant the GPU cost stayed the same. They could instead double the horizontal resolution of the screen to keep the same effective resolution, in which case the GPU performance cost would indeed double. But as I said, in future systems this performance cost could likely be reduced with a form of frame generation by a machine learning model.