Ray casting
 

is there a good ray casting tutorial for amiga?

I don't know of one, but isn't it effectively the same thing as on the PC, but adding C2P conversion?

That's exactly what is is, there is no "Amiga raycasting" at all.

Well, on Amiga you usually do it into a chunky buffer in fastmem and then use C2P into chip mem, but that's the only real difference.

Bisquit on YouTube has some good tutorials on raycasting.

What does ray casting have to do with chunky graphics?

Nothing per se - except that (assuming your rays are intersecting with vertical walls) you're processing the scene one column at a time, which is horribly inefficient if you're rendering to a planar framebuffer.

Since Amiga is a word(16 bit) machine, was thinking about doing ray casting with this in mind, and maybe get rid of c2p

Yes, but why not just use integer based vectors? Seems faster than bitmap based ray casting, unless I'm wrong (never done it).

Sure, that's doable if you're rendering walls with flat shading - but as soon as you introduce texture mapping into the mix you have loads of shifting and masking to deal with before you can write a 16-bit word.

What's the actual use case for ray casting here? I'm thinking of ray casting as used in current 3D games...

I've written a raycaster but not done 3D graphics with vectors. The big draw of raycasting is simplicity - a simple step-by-step procedure for each column on the screen tells you which walls have to be drawn (just the first one you hit unless you're dealing with transparent textures), which texture column should be drawn, and at which scale. No need to depth-sort polygons, deal with backface culling, view cones, clipping and all the stuff that comes with polygon-based 3D graphics, even before you've added texture mapping into the mix.

(Just to be clear, I'm talking about raycasting as in Wolfenstein 3D here - Sandruzzo, is that what you meant?)

Right, now it makes sense. Didn't realize the term was used in that context.

Yes. I meant that:great

Maybe we some good lookup table we should be able to do texture mapping word based.

What if we use just one bitplane texture?

We still have the problem of assembling a 16-bit word from 16 independently selected texture pixels - so still lots of masking and shifting.