Re: Perspective not quite correct?

This thread has become fragmented on my newsreader, so I don't know what the
original post was.
I had a quick read and I think I know what you're getting at.

I approach this from an artist's point of view who paints trompe l'oeil
(trick the eye) murals.
I design images for walls like false doors, and other openings etc.
I see it simply like this ...
Any point in space gets mapped to the wall using a light ray from that point
to your eye at the origin. This has nothing to do with field of view etc.
If you had your large chequered square behind the wall and parallel to it,
it would get mapped to the wall exactly in the same proportions as the
original but scaled down. (Similar triangles.)
Meaning, the square in the top left corner would be the same size as the one
in the centre (and they would have straight edges).
What happens with a mural is that because that mapped corner square on the
wall is further away, it automatically gets foreshortened and looks smaller
because of the viewer's position in relation to it. It doesn't have to be
manipulated because real life is doing it for you.
So, mapping to a PC screen is the same except of course, much smaller.
Think of it this way ...
Imagine you have a *** of glass in front of you and you traced the image
behind it onto the glass.
You can still look around the glass turning your head, but the glass stays
in the same place. So as Geoffrey implied, the z-axis for the glass is not
changing.
Of course the mapping does change if you turn the glass with you, but I
don't think that this is what you're talking about.

Regards,
Ron Francis
www.RonaldFrancis.com


"Villiam Jurnheim" <wjh@xxxxxxx> wrote in message
news:465f2d9b$0$21925$157c6196@xxxxxxxxxxxxxxxxxxxxxxxx
But hang on to the 'should be' by all means. It isn't true,
but what the heck, people believed the earth was flat and
the sun revolved around it. As I said before, when you say
it's smaller, it's because you are looking directly at it,
not down the z-axis, that changes the ENTIRE resulting image.

Imagine a large square which is has a checked pattern on it. Each little
subsquare is 1 by 1 meter.
If you look at an object at twize the distance, then it is half height and
half width and the area is one quater of what it was before.
If you look at one of the little checkers at a corner, it has a distance
from you of sqrt((½w)^2+(½h)^2). If you look at one check at the left
vertical side at y=0, its distance to you is sqrt((½w)^2+0). Which
distance is the biggest? w and h is width and height of the square which
has its zenter at x=y=0 and some z distance in front of you. The square is
lying in the xy plane.

Can we at least agree on this much?
1:Objects shrink the farther they are away from the observer.
2:The corners are farther away from the observer than the center of the
edges are.

Straight
lines in world space will still be straight lines in screen
space.

Yes. Because one calculates the perspective divide based on z alone. Again
you argue that things are as they are, because this happens in 3dgraphics.
Ignoring the fact that I am saying that things in 3dgraphics seem a little
simplified in this aspect. I am not saying that in 3dgraphics things are
not parallel. I am saying that they are, and they should not always be.
With small fov, the difference will be tiny






.