home *** CD-ROM | disk | FTP | other *** search
- > I've tested the options of the c version, I've only seen the bugs with
- > wallopt, I mean when I press <SHIFT>+w . I haven't found any bug with
- > <SHIFT>+f . Did you ?
-
- The bugs are just not visible when you don't select wall optimization.
- They're always there, but without wall opt all walls are drawn after all
- floors and therefore obscure the errors.
-
- I guess I should get that snap-shot I promised earlier.
-
- > In the case that only wallopt bugs, we can use flooropt directly for the
- > floors, and the normal function written in drawseg for the walls.
-
- I principle we could do that, but the bugs will sometimes cause a lot of
- unnecessary drawing to be done. Since the clipping structures are screwed
- up, I think the sprite drawing will be impossible as well unless the bug
- is fixed.
-
- > Do you agree with me ?
-
- No. ;-)
- Apart from what I said above, I believe the optimized wall drawing will cut
- down on the amount of DSP->'030 communication since a single wall is drawn
- at a time. Without the optimizing, I think some extra information will have
- to be stored and transfered (possibly more calculations as well).
-
- > Or do you think there'a a possibility to correct the bug of Wallopt ?
-
- The bug isn't in the wall optimization as I said above. I can't believe it
- isn't possible to fix it. It should even be easy...
-
- > I'd like to optimise the code to its maximum speed. For that, I musn't
- > take in considerations different calculs options, but only one. So, if
- > Wallopt doesn't work, do you think we should forget it and keep flooropt ?
-
- I'd say forget the non-optimized wall drawing. At the very least that'll
- put some extra pressure on us to fix the bug. ;-)
-
- > Another question : did you make the drawmap function ? (I mean a function
- > that draws the level map with different colors for the walls, stairs, ...) ?
-
- The rotating map of the walls the engine considers could easily be adapted
- for other uses.
-
- > If yes, could you send it to me ? I'd like to convert it, and then, to
- > make some tests with it ...
-
- Sure, I'll do that ASAP. I was going to wait until I had the time to do the
- wall checks, but I never seem to get around to that.
- The routine could really benefit from some optimzed line routines. I use a
- simple fixed point one that would probably be pretty good if it wasn't all
- C code, but I do the clipping on a point by point basis...
-
- > The last point I'd like to talk about is the textures. I haven't optimised at
- > all the drawseg, coldraw and floordraw functions, as they'll change for the
- > mapping. Do you thing a version of the mapping will run soon ?
-
- Not really, unless someone other than me does it. A course I'm taking at the
- moment requires me to write a compiler for a functional language more or less
- in my spare time, so most of my programming efforts will have to go into that.
-
- > Do you know how to load the textures in a wad, convert the pictures in a
- > chunky mode ( to access them in an easy way ), and map them on the walls ?
- > Or do you need some help ?
-
- I have not looked at the texture specs, but I can't believe loading them
- could be difficult. They are already in some kind of chunky format as well.
- The mapping itself should not be a problem. Getting hold of the extra info
- needed from the engine might be tougher.
- As I said, I don't really have much time now, however, so any help is much
- appreciated.
-
- The extra info I'm talking about is needed because the mapper must know
- where to start in the texture (x and y coordinates) and how far away the
- line is.
-
- > I think in the first time, we should try to map the walls only, and the
- > floor after. (For 2 reasons : it's easier to map walls than floors, and it
- > takes less machine time).
-
- Yes, I agree.
-
- --
- Chalmers University | Why are these | e-mail: d8klojo@dtek.chalmers.se
- of Technology | .signatures | rand@cd.chalmers.se
- | so hard to do | www/ftp: rand.thn.htu.se
- Gothenburg, Sweden | well? | (MGIFv5 and QLem)
-
-