home *** CD-ROM | disk | FTP | other *** search
- > Who are the new participants ? Coders, do they code in assembler or in C ?
-
- I recognized more than half of the email addresses, but if they want to
- introduce themselves, it's up to them.
-
- > What are they doing ?
-
- Very possibly nothing at all. The list in publically open, which means
- we may well have participants who only want to hear what's going on.
- Let's hope some of them are interested enough to help out a bit or at
- least take part in the discussions.
-
- > For the conversion, I know lot's of parts of the code that I can improve,
- > we'll see at the end how far I can improve this ...
-
- :-)
-
- > I'm starting to look at the code that'll go into the DSP ( probably the
- > BSP tree, I'm studying this...). If we could put it into the DSP, it would
- > improve the code a lot, because of the big number of muls and divs.
-
- Yes, just getting rid of all the clipping and rotation stuff would
- probably be helpful. Then there are those tables...
-
- > For the algorithm question, I just thought of something, perharps stupid,
- > perharps not :-)
-
- [some details about a bitmap of the level map deleted]
-
- As Bertrand mentioned it's a rather good idea, but in contrast to him I
- see some problems with it. I'd very much like to be proven wrong, though.
-
- I don't know if there are any limits to the coordinates that can be used
- on a DOOM map, but figures between 1000 and 2000 are common. That means
- that we need to keep a map of at least 2048x2048 in memory since I'm
- pretty sure we'll to represent at least the vertices exactly.
- With a single bit map that would be 512kbyte.
-
- Now, how would that map be drawn?
- We really need every possible sector in there, which means that we'll
- have to go through the data structures and fill in all the sectors.
- But wait, we can't do that since some sectors are not always accessible,
- such as doors. The doors can of course be handled by repainting part of
- the map when they're opened/closed.
-
- OK, assume we have a map. How do we use it?
- The map was drawn using some normal line routine, which means that the
- walls are (almost) never exactly drawn. Take a look at the following:
- (The player is the '+')
-
- O****** O******
- OO***** OO*****
- OOOO*** -> OOO+***
- OO+OO** OOOOO**
-
- Both before and after the movement the player is on an 'O' which is
- supposed to mean open area in this case. In the latter case, however,
- the open space is only there because the line routine made a step.
- In reality that position is inside the wall.
- To make matters worse, the player's (and the monsters') position will most
- likely have to be stored as some kind of fixed point number. I don't think
- it's done like that right now since the current DVIEW is so slow anyway,
- but for fluid motion it will be needed.
-
- If a position has a '*' (wall) close by, how do we find out if we're
- inside a wall or not? As far as I can see the only way to find out for
- sure (we _really_ don't want to draw from inside the wall) is to check
- against the line. But which line? Our bitmap does not say anything about
- that. And we're back to the original problem (IMO).
-
- The bitmap could be used for a quick check that we're safe, but in most
- cases I think that could be done rather easily with the line coordinates
- as well.
-
- By the way, can all trigger mechanisms and such be handled only by knowing
- which sector the player is in? Or is there ever a need to know which line
- was crossed?
-
- > If I'm totally wrong, just tell me. :-)
-
- You're certainly not totally wrong, but I doubt if it's possible to use
- your idea. I hope I'm mistaken, though.
-
- > What about the square technic described by Bertrand ? What can we use it for ?
-
- As Bertrand said, that's more or less the same thing but on a larger scale.
- For collision detection against objects I think that will work fine since
- the those are just sprites. There's no great problem if you don't get
- their positions exactly right.
-
-
- --
- 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)
-
-