home *** CD-ROM | disk | FTP | other *** search
- Path: nntp.teleport.com!sschaem
- From: sschaem@teleport.com (Stephan Schaem)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: 90 degree rotation
- Date: 9 Feb 1996 08:02:49 GMT
- Organization: Teleport - Portland's Public Access (503) 220-1016
- Message-ID: <4fev39$jid@maureen.teleport.com>
- References: <DMEpMn.Cvs.0.-s@cs.vu.nl>
- NNTP-Posting-Host: kelly.teleport.com
- X-Newsreader: TIN [version 1.2 PL2]
-
- Irmen de Jong (irmen@cs.vu.nl) wrote:
- : Regarding texturemapping (doom like):
- : Wouldn't it be wise to write the walls in a 90 degree rotated fashion?
- : I mean, not vertically, but horizontally. It might then be possible to write
- : whole longs at once, which is much faster than writing single bytes.
-
- Much faster, the following is basic wall tmaping?
-
- REPEAT 16
- move.b (a0,d0.w),(a1)
- addx.l d1,d0
- adda.l a2,a1
- ENDR
-
- Even if you trandform this to write long (Would be slower) how do you
- handle your chunky buffer? Do you tmap 1 line and c2p it?
-
- : Is 90 degree rotation in the C2P phase easy to do? Or do I miss something
- : here?
-
- In your c2p loop you want to read 32bit from your chunky buffer and
- write 32bit to chipram... You do this when the chunky buffer is
- rendered, not a few pixels at a time.
- But lets say you just wrote a Vline of 32pixel on the stack, now
- if you c2p this you will only write 1 pixel Hline in the chunky
- buffer. So 8*32*1 read+And+Or+write. What you want really is
- 8 write... So this would be more then 64 time slower....
-
- : I recall that the "PicCon" program can save 90 degree rotated chunky
- : textures. What other use might this have?
-
- You would use rotated tmap for walls only.... because you step verticaly
- in your display buffer, and horizontaly in your texture buffer.(Just
- a speedup for the inner mapping loop)
-
- Stephan
-