Hardware collision detection in Amiga games?

[Photon]

Originally Posted by nandius_c

Which game did you code?

Sad fact is none were released. "Super Munch-Man" (guess which game it emulated ) was finished & scheduled to go on a CU Amiga cover disk in mid 1991, but I needed help from a friend to make enough levels in time. Instead I went to England to apply for work. Remember, I was young and expected Amiga 500 to go on forever... While applying at Graftgold I coded a shoot'em up demo (no enemies) using gfx Uncle Tom had drawn. Also applied at Argonaut Software with my vectors and that's where I worked briefly. Didn't get to code any vectors though so home I went.

Before that I did Fetris which actually surfaced at a xmas visit to a friend, a Tetris Grand Master aficionado was present so we talked some more about block bag algorithm (it's a thing...) and the host started a logo replacement for my old attempt

Demoed my cross between Gravity-Force and Exile for a demogroup member at a recent retro meeting, flying around in caves... <3

Also talked to Uncle Tom again to get some of his game gfx out some damn time a few weeks ago...

Also have a sprite/bob multiplexer thingy with isometric 3d levels similar to Snake, Rattle 'n' Roll / Fire & Ice but with penguins and ice floes, a Breakout thingy with creatures clamboring to drown your bat and water level rising.

Most of these use my game system with SFX system, level design in DPaint and auto pic bob scanners/mask gen.

Still intend to get some games on the board but at this day & age it's sadly pending time being stolen by work and other projects that I stupidly sign up for like a lollybobs.

[Lonewolf10]

Photon, any chance of releasing the finished games?

[nandius_c]

Originally Posted by Lonewolf10

Photon, any chance of releasing the finished games?

+1! By the way, nice story, Photon .

[Photon]

Originally Posted by Lonewolf10

Photon, any chance of releasing the finished games?

Yeah I've started thinking about releasing Super Munch-Man and Fetris. That's why last xmas there I showed it to Illmidus & he got interested, so I AGA-fixed it and started looking at the panel graphics layout to make it multiplayer.

Super Munch-Man is trackdisk while Fetris is a small executable to just run it quickly and get your fix then continue.

I should be able to get them out during summer holidays, I'll keep it mind & let you know in another thread maybe? Topic was collision detection (sorry about that).

[nandius_c]

It would be great to see that thread started, Photon!

As I said before, I found tunneling (fast moving objects going through others with no collision being detected) one of the hardest problems to solve. I guess it can be kind of "solved" when designing the game. But anybody knows which way was this usually done back in the day? Here's a thread where Mrs Beanbag talked about a possible approach: http://eab.abime.net/showthread.php?t=71288. What I did when I first met this problem was simply reducing the time slice of the simulation, so that tunneling would not happen so often but I guess this can be heavy if there's a big a mount of objects to check...

Anybody can bring some light about this point?

[nocash]

Theoretically someway like this:

  4x4 pixel sprite (motion step = 0 pixels)
    ##
   ####
   ####
    ##

  4x4 pixel sprite (motion step = 10 pixels)
    ############
   ##############
   ##############
    ############

Ie. for the 4x4 sprite moving at 10-pixel step, treat the 4x4 original sprite as being 10-pixels wider in collision logic (ie. 14x4 pixels). Then you could do a pixel-to-pixel mask check, or bounding rectangle check.

The rectangle won't work well in case of diagonal movements; you could eventually brew up some paralellogram / diagonal line maths instead of rectangle checks.

You are still working on the Arkanoid-style game mentioned in the http://eab.abime.net/showthread.php?t=71288 thread? Then you would need some kind of line/vector maths solution that can compute if there is a collision - and that could also compute if the collision occurred at the left or bottom side of a brickstone (so your ball could bounce-back in the opposite direction (using the blitter/sprite hardware's pixel-collision tests couldn't be used for such bounce-back stuff)).

Originally Posted by Boo Boo

or when math check correct then do blit check for 100% shape check.

Also a good idea - first do some fast but inaccurate "nearby" check, then do some more accurate check if needed. Of course, you won't need that method if you can make test that is both, fast AND accurate.