Matt Burch: The Mystery

NebuchadnezzaR, on May 15 2005, 08:54 PM, said:

Well, techincally, you can build the rotation into hardware. Ever heard of a graphics card? Thats most of what they do, rotate triangles of images. And they can do it damned fast, since it doesnt even use code.

And having the whole sprite is a huge drain on ram, which is a legit concern. In every situation, there is a tradeoff between ram usage and speed. There is always a way to rewrite some bit of code so it either takes up more ram, or it goes faster, but not always both (unless its crappily written). prerendered, prerotated wastes ram, rotating in the program (or rendering for real, on the fly) wastes time.

You're forgetting the key element here: difficulty of coding. Also, some people (albeit few) have no graphics card, so they're SOL on that aspect. And I think you mean takes up more RAM, or goes slower, not faster.

NebuchadnezzaR, on May 16 2005, 01:54 PM, said:

Well, techincally, you can build the rotation into hardware. Ever heard of a graphics card? Thats most of what they do, rotate triangles of images. And they can do it damned fast, since it doesnt even use code.

And having the whole sprite is a huge drain on ram, which is a legit concern. In every situation, there is a tradeoff between ram usage and speed. There is always a way to rewrite some bit of code so it either takes up more ram, or it goes faster, but not always both (unless its crappily written). prerendered, prerotated wastes ram, rotating in the program (or rendering for real, on the fly) wastes time.
View Post

Have you tried programming in OpenGL? It's not 'no code'. With sprites it's simply a case of copying some memory around. Easy. There's a system call for it. With OpenGL you need model mesh, you read it in using many glVertex calls, one for each triangle (a loop, fairly easy). Then you add lighting from a specific direction, and then render it to screen. Far more complex, as you still need to do all the maths to compute the rotations...

I think we have seen the end of pre-rendered sprite cells in major games for good now that everyone and his aunt has a 3D card. I still use them because I like the classic retro look.

Constructing your 2D and 3D sprites from polygons has lots of advantages - shadowing, dynamic lightsourcing, scaling, deformation etc.

But if you're not careful, you can end up using more memory to store all the buffers and textures.

(Edit: I remembered the name of that game: "Super Ships II")

This post has been edited by tycho61uk : 16 May 2005 - 03:14 PM