home *** CD-ROM | disk | FTP | other *** search
- Hi,
-
- First of all I'd like to say that I really appreciate that you released
- the sources for DVIEW. I've read about DOOM type rendering for quite a
- while, but to actually study working code was wonderful. Many thanks!
- You really should avertise that code more in rec.games.programmer. Every
- other day there's another person asking questions that are all answered
- quickly by a look at your work.
-
- The reason I started looking deeper into this field was that I and a couple
- of other people on the internet decided to see if it was possible to write
- a DOOM compatible game for the Atari Falcon030. As the name suggests the
- machine is running a Motorola 68030, in this case at 16MHz with a 16 bit
- data bus. The CPU itself will of course have no chance of handling DOOM,
- but there's also a M56001 DSP, which we hope to use extensively.
-
- The port of DVIEW to the Falcon took only a couple of hours, but in its
- current form it's not very useful. Compiled with maximum optimizations,
- using GNU C++, a staggering 2.5 frames/second average was reached...
- Now, putting all the rendering code on the DSP would of course improve
- the frame rate tremendously, but there is one major problem with that:
- The DSP only has 32kword (24 bit) of memory
- (there is a two-way DMA channel at somewhere around 1Mbyte/s as well)
- So, the wallrun, intersections and floorlist arrays won't fit. :-(
-
- I managed to eliminate the arrays mentioned above by drawing all walls
- and floors as soon as they were available, which most often works rather
- nicely. The main problem is that floors behind doors are drawn after the
- door itself, but I think it'll be possible to handle that somehow.
-
- While this all appears to be quite fine at the first look, I've begun to
- worry about the continuation of the project. It seems to me that the
- inserting of the sprites into the environment will be extremely difficult.
- Before the arrays were removed, there was some information that could
- perhaps be used to determine what the sprites should be clipped against,
- but now that's not there anymore. :-(
-
- So, a couple of questions:
- - Have you given any though as to how sprite clipping would be handled?
- - Have you done anything else on the program since you released it?
- - Any ideas on what could be improved/optimized?
- - Do you know if anyone else is using the sources?
-
- Finally a couple of bug reports:
- - For some of the load routines the calculation of the number of
- structure items to load is less than perfect (negative results can occur).
- The (short) typecast is in the wrong position.
- - You may have noticed that there are sometimes very strange distortions
- near the 'front'. I've traced this to the first of your clippings, the
- one where you clip to an x=z line.
- If you remove that clipping completely, everything works nicely.
- Was there a reason for the clip?
- If so, why only clip the 'from' coordinates and not the 'to' ones (or
- perhaps it was the other way around)?
-
- By the way, the maze looks much better if you let the name of the textures
- used set the colours (I XOR the characters in the name right now, but a
- more sophisticated algorithm would be better).
-
- Regards,
- Johan
-
- --
- 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)
-
-