I realize it's kind of lame to ask but I can't come up with a good project idea for my digital hardware design class. Basically I've got about 3 weeks to do something; I've got a Spartan 3 FPGA, a Diligent A/D A/C, and some kind of VGA board I think as well. I was thinking about a hardware ray tracer, but A) it's been done and b) I don't think I care about hardware ray tracing. I'd like to do something that pertains to 3d though, or I don't know, just something cutting edge, even if it's already been done and in production, just something with a relative newness. The last thing we did was a FIR filter, so I've got some VHDL components from that, but I'm just ...bored. Anyone got any ideas for something fun?
What would be a fun project?
- Thread starter Otto Dafe
- Start date
Killer-Kris
Regular
Since memory constraints would likely be a problem, 
you could always look into doing a HW rasterizer (ie. primitive GPU).
you could always look into doing a HW rasterizer (ie. primitive GPU).Since memory constraints would likely be a problem,
That could be interesting, I think I have some vague notes about rasterizers in an old OGL textboook, in the history section. I've got 2 MiBs offboard I think (or maybe onboard I dunno). Is a rasterizer just a serial out kind of thing, I guess it must be, as I come up with almost 10 MiBs for a 640x480 frame buffer.
Blazkowicz
Legend
why 10 megs? looking at the Voodoo1, it has 2MB framebuffer memory (and 2 or 4MB separate texture memory) and does 640x480 16bit, double buffering, Z-buffer.
what I'm thinking of is vector graphics. you know, stuff like Happy Tree Friends, Another World and boring fonts, that would be something great to accelerate.
what I'm thinking of is vector graphics. you know, stuff like Happy Tree Friends, Another World and boring fonts, that would be something great to accelerate.
Killer-Kris
Regular
Yeah, one of the benefits to a rasterizer is the relatively low memory requirements, basically a front buffer, back buffer, and z-buffer, of which you could even actually do away with the back buffer if you can tolerate the tearing.
I might start by looking at the old OpenGL 1.1 state diagram
http://opengl.org/documentation/specs/version1.1/state.pdf
Being that it's a class project you obviously won't want to design something that complex but I imagine you could cut it down to something quite reasonable. At which point that leaves plenty of room for discussion in any reports on directions the project could be expanded as well as a discussion on current GPU development trends.
Vector graphics? As in OpenVG?
3 weeks? If it's supposed to work, that somewhat limits your options, doesn't it?
Forget about rasterizers, ray tracers, voxel stuff etc. Way to complex for such a limited time. That kind of stuff would be more suitable for a master's thesis.
If, indeed, you have a working VGA controller such that it display whatever has been stored in memory, how about a small computer game? Nothing fancy: a square moving from left to right on the top of the screen. A canon (also a square) at the bottom in the middle of the screen. A bullet going straight up. Collision detection.
Extra points if the bullet can do not-so-straight lines and if the score is updated on screen.
Sounds simple but usually college level assignments are about traffic light controllers and money counting state machines and even there inexperienced students often need a week to get it right.
If you have access to a analog oscilloscope with X-Y mode you could make a vector interface for drawing triangles. This is easy to do and should be possible in 3 weeks time. You will also need fast D/A converters, 12 bit will give you a display of 4096x4096 pixels.
If you want even more fun you could get the triangle data from your own rewritten version of 3Dfx Glide such that you can play old games in wireframe on the oscilloscope.
If you want even more fun you could get the triangle data from your own rewritten version of 3Dfx Glide such that you can play old games in wireframe on the oscilloscope.
