nVIDIA GeForce256 and T&L
More effects: Cube Environment Mapping
There is going to have left us an impressive, sure cars game, because we are going to make it as real as possible. Because we have a fantastic black Jaguar that has just gone out of the newly washed garage, with what we will be able to turn out to be reflected in him, as the trees of the highway and everything what approaches...
Until now everything had been almost impossible to apply to the games, because the used method was consisting of surrounding the objects with a sphere, which interior surface they were going to reflect (that is to say, that was using a "map of spherical environment of reflexes"). This was giving a few quite satisfactory results... if we had A LOT OF time to be lost to do the necessary calculations, which were even more complicated for objects in movement (and a game is not precisely slightly static). And in addition, it was turning out to be slightly difficult to programme...
Now our card GeForce yes that will do it. Have you ever looked in a mirror? Since this is what does the GeForce; it takes the object that can reflect things, wraps it with a bucket and projects the images of flat form in this bucket, what needs great less calculations. And also, now when we move the object the reflexes will not distort, we will be able to calculate them again quickly from the bucket. We are going to see it:
We project what will be seen in each of the sides of the object and keep it in the lower (unfolded) bucket; now we have the kept reflexes information and, unless the point of view changes radically, in the following frame at least we will have to calculate six reflexes again, very probably we will be able to make use of several of the faces of the bucket.
Also, we can use the bucket to look for the guided lights, saving time and improving the quality of the image.
Interpolation 3D, shaped by means of NURBS
This sounds fantastically but neither get excited too much, it is one of the characteristics already waited in DirectX 7, the acquaintance like Vertex Blending. Now on already having joined two objects a "break" will not be evident, but with the new graph, the union elbow will be smoothed between the objects:
To the left without DirectX 7, to the right with DirectX 7 and one GeForce. I suppose that you all will appreciate the difference...
Very well, after having realized all the calculations, we will need to draw things on the screen (we are not going to extract only the numbers and leave that the player imagines the images); of this the accelerating one takes charge. This is the phase of the render, and in her we are going to "fill" all the polygons of the scene with the colors and corresponding texture, after having applied the corresponding lights.
The polygons refill pixel to pixel by means of the units of render of the graphic card; major all that is this Fill Rate, more píxels we will be able to paint and more frames (images, pictures) will show per second.
In case of the GeForce we have 4 render units. Also, the card at present is the one that more píxels can paint on the screen of all that they are on the market for an "accessible" price. The magic number remains in 480 million (theoretical) píxels per second. Also, it is necessary to bear in mind all the capacity of calculation that we have seen earlier.
In these moments there is certain discussion on which it is more important: to liberate the calculations CPU (with T&L) or to have a better Fill Rate. Without realizing we have come to two different positions that they support nVIDIA and 3Dfx. It seems that the programmers see more advantages in T&L that in increasing the Fill Rate and that it is a question of the technology that will dominate the future.