learn Asm and read up on Binary Space Partitions, and mesh collision detection

or just fake the 3d models.

actualy, ben reves has a project that could do a number of those things and was true 3d (gemeni is a raycaster that is actualy 2d with texture angling, as is zdoom).  The problem with real 3d is it requires a LOT of memory to sotre x, y, and z for each vertices meaning models for enemies is likely going to be very limmited if it ever happens.  While Im not sure it can't be doen as I only recently began programming even a little assembly, raycasters can achive most of the effects of 3d with much less memory cost and much higher speed and ultimately are the most accesible and proven method of 3d on calc.  On the 89s, there it might be more likely.

edit: sorry ben, I mixxed you and kerm...

All of your ideas can be achieved. However, the game would be too slow to be any fun.

Here is a great tutorial for raycasting engines -- it includes variable height walls, climbing, ceilings and floors, etc:


http://www.permadi.com/tutorial/raycast/

ya, basically what everyone else said.  the reason why most calc fps are raycasted is because raycasted is pretty watered down compared to what a comp does real time for PC games.

my only complaint about gemini was that for some reason, I couldn't see enemies unless they were facing me, which kinda frustrated me.  and zDoom, people always seemed to be moving through walls and the guys were huge.  both were still really fun though

benryves on MC was working on a 3D engine for the z80 calcs.  duno where that's going though.

before i forget though, I should respond to each point individually to give a little explanation why raycasting limits your ability to achieve those ends

1.) In raycasting, all walls HAVE TO BE FLAT.  so textures are more or less generated by "texture maps" probably not the right word, but i hope you get the idea.  In gemini's engine, i wouldn't be surprised if coelerus simply generated his 3d world from a 2d tilemap that simply indicated what kinda "brick" to use.  the characteristic of the brick was probably what kind of sprite would be used to paint the wall.

in zDoom, I'm not sure how many angles there really are, but it also is probably generated by tilemap.  I'm not sure if doing both would be overtaxing on the processor, but one reason is probably that angled walls require the "texture" to be transformed by perspective, in addition to simply scaling the wall sprite to give the illusion of the wall textures becoming clearer.  When you look at a wall at an angle, the farther end of the wall is squished, so all the textures have to be squished.

2.)  Not sure what you mean, but if you mean by height of the wall, I'm not sure,  I imagine that it could be possible w/ raycasting, but it wouldn't be as watered down.  iirc, raycasting at its fullest also assumes that all objects have the same height.  That way you can render an entire column of pixels at once because you don't have to worry if the top half of the screen will be empty and you need to cast the ray further than where the pixels on the bottom half "hit the wall"

3.) probably linked to 1 + 2, but keep in mind it's easiest to generate from a tilemap.

4.)  Like I said before, raycasting assumes that everything's flat and 90 degrees from the floor so that all rays in one column will travel the same distance and can be processed at the same time.  I assume this also makes it difficult to tilt the view up and down, although a translational up and down motion (like in zDoom's death sequence) is just a matter of readjusting the camera height.

5.) eh, I agree w/ elfprince there, but my complaint about gemini still stands, though my ASM skills don't give me the right to say to much about it.  if only the enemies were also blocks, so you would have textures on 4 sides of it.  I kept missing them for some reason.

OVERALL
Don't think I'll discourage you because my ASM skills are still at a recently-not-n00b state, and my skill probably won't progress for awhile until I start churning out my first game.

however, based on what I've seen and forum discussion i've read, it all depends on how you build your engines and what limitations and assumptions you set.

3D modeling like on the comp might actually be feasible on a z80 calc, but I wouldn't know.

Voxel (is that the right one) on 68k has shown it's possible (right?).  it's probably a matter of time before something good appears.

maybe gs fps will come first though (unless there'a already one I'm not aware of?)

just ask for help.  I at least can point you in the right direction  And there are lots of other really good ASM programmers here too.

i think that's how Zeromus was going to make his basic fps (+xlib?)

same goes for the lurking shades project too i think