38699

神经网络昆斯林的新闻组分类2006

Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!noc.near.net!howland.reston.ans.net!zaphod.mps.ohio-state.edu!swrinde!dptspd!tamsun.tamu.edu!tamsun.tamu.edu!news
From: [e-mail] (The Man behind The Curtain)
Newsgroups: comp.graphics
Subject: Views on isomorphic perspectives?
Date: 23 Apr 1993 11:39:06 -0500
Organization: Texas A&M University
Lines: 87
Sender: [e-mail]
Message-ID: <[e-mail]>
NNTP-Posting-Host: tamsun.tamu.edu
Keywords: isomorphic perspectives

                             
I'm working upon a game using an isometric perspective, similar to
that used in Populous.  Basically, you look into a room that looks
similar to the following:

                             xxxx
                         xxxxx   xxxx
                     xxxx    x       xxxx
                 xxxx        x           xxxx
             xxxx     2      xxxx     1      xxxx
             x           xxxx    xxxx           x
             x       xxxx            xxxx       x
             x   xxxx             o      xxxx   x
             xxxx              3 /|\         xxxx
                 xxxx            /~\     xxxx
                     xxxx            xxxx
                         xxxx    xxxx
                             xxxx

The good thing about this perspective is that you can look and move
around in three dimensions and still maintain your peripheral vision. [*]

Since your viewpoint is always the same, the routines can be hard-coded
for a particular vantage.  In my case, wall two's rising edge has a slope
of 1/4. (I'm also using Mode X, 320x240).

I've run into two problems; I'm sure that other readers have tried this
before, and have perhaps formulated their own opinions:

1) The routines for drawing walls 1 & 2 were trivial, but when I ran a
packed->planar image through them, I was dismayed by the "jaggies."  I'm
now considered some anti-aliasing routines (speed is not really necessary).
Is it worth the effort to have the artist draw the wall already skewed,
thus being assured of nice image, or is this too much of a burden?

2) Wall 3 presents a problem; the algorithm I used tends to overly distort
the original.  I tried to decide on paper what pixels go where, and failed.
Has anyone come up with method for mapping a planar to crosswise sheared
shape?

Currently I take:

  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16
 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

and produce:
            
             1  2  3  4
33 34 35 36 17 18 19 20  5  6  7  8
49 50 51 52 37 38 39 40 21 22 23 24  9 10 11 12
            53 54 55 56 41 42 43 44 25 26 27 28 13 14 15 16
                        57 58 59 60 45 46 47 48 29 30 31 32
                                    61 62 63 64

Line 1 follows the slope.  Line 2 is directly under line 1.
Line 3 moves up a line and left 4 pixels.  Line 4 is under line 3.
This fills the shape exactly without any unfilled pixels.  But
it causes distortions.  Has anyone come up with a better way?
Perhaps it is necessary to simply draw the original bitmap
already skewed?

Are there any other particularly sticky problems with this perspective?
I was planning on having hidden plane removal by using z-buffering.
Locations are stored in (x,y,z) form.

[*] For those of you who noticed, the top lines of wall 2 (and wall 1)
*are* parallel with its bottom lines.  This is why there appears to
be an optical illusion (ie. it appears to be either the inside or outside
of a cube, depending on your mood).  There are no vanishing points.
This simplifies the drawing code for objects (which don't have to
change size as they move about in the room).  I've decided that this
approximation is alright, since small displacements at a large enough
distance cause very little change in the apparent size of an object in
a real perspective drawing.

Hopefully the "context" of the picture (ie. chairs on the floor, torches
hanging on the walls) will dispell any visual ambiguity.

Thanks in advance for any help.

-- 
Till next time,                \o/   \o/
                                V \o/ V     email:[e-mail]
<>  Sam  Inala  <>                 V