SetPixel() allows you per-pixel access to any image loaded in Verge.
Say, you load your image for a health bar, but then realize there needs to be a red dot at position 0, 10 on it. SetPixel allows you to alter the image from within the code, instead of from outside, in a paint program.
Right now that's the only per-pixel function Verge has. Which means, if you want to create your own pictures, such as a Mode7 background (like those 3D backgrounds in Mario Kart), you've got to do it entirely with SetPixel.
But there's a problem. That means you have to call SetPixel over every pixel on the screen. If you have a 640x480 resolution, that's...a darn lot of pixels. Unfortunately, Verge's SetPixel is pretty slow because of function overhead. Well, technically, it's also slow because VergeC is just interpreted code, and isn't pure C code. Ika's SetPixel() and Pygame (a version of SDL for Python) has a SetAt() function that are all pretty much slow.
What Rage is talking about, is that instead of having to mess with every single call to SetPixel, you can instead assign values to an image array.
The idea is pretty nifty. Assigning a value is faster than a function call, so you could do something like:
int myimagearray[640][480]; //contains the data for a 640x480 image
//...Do some image drawing in the array
while(something)
{
myimagearray[x][y] = somecolor; //Theoretically, this is much faster than a ton of SetPixel() calls
}
//And then display the image, kind of like this.
int myimage = GetImageFromArray('myimagearray'); //Or something similar
Blit(x, y, myimage);
The idea is, a 'GetImageFromArray' would be a function that creates an image from an array much faster than you could do it with SetPixel. (Ever tried making a RotScale or your own Blit or Flipping function with SetPixel? The builtin functions are, in general, much faster).
So, RageCage says a method similar to this would be ideal for rapid image generation.
Does this make sense? ...I probably ended up repeating a lot of unneccesary stuff.
On a vaguely related side-note: As I recall, you could actually use the DMA functions on the 'screen' pointer in Verge2 (Verge3 doesn't allow it), but for some reason it was just about as slow as SetPixel anyway. I didn't, and still don't, quite understand why DMA assignment in VergeC was just as slow as SetPixel. I figured DMA would avoid a lot of overhead.
