Re: 3D Landscapes - a closer look
>I'll take a look at your thing. I already do all the perspective
>calculations on mine, and I'm drawing back to front, so I don't have to
>worry about the hidden line problems.
If you are including perspective as against just rotation around each
axis, then as soon as you drop below satellite view you must also draw
from both edges of the screen inwards as well as from back to front to
ensure you are *eventually* displaying the correct polygon. This is the
method my 'thing' uses.
This is compounded when you are looking at the map from an odd angle
as you would have to find the central line of vision and display points,
in order, from most distant to this line, to the closest.
For example: if you are looking north up a city street, you see the east
face of the buildings on the west side of the road and the west face of
the buildings on the east side of the road. If you are drawing from back
to front and left to right, a building on the west side of the road will
be drawn correctly but the east face of a building on the east side of
the road (out of view) will be drawn over the top of the building's roof.
If you are looking up a street that points south south-west, you have to
work out which buildings are on which side of the road.
>If you set wire=0 it will draw
>wireframe with hidden lines. (actually I draw a black filled poly then a
>colored unfilled poly, so whatever was behind the current polygon is
>erased. Looks nice and clean, especially in 1024x768.
Looks great, but that's not the point, the point is that drawing several
hundred polygons, then drawing over the top of them can hardly be called
speed optomization. Surely to have any hope of doing any of this in
real-time, you'd have to first work out which bits can be seen, before you
bother drawing them. I'm assuming that this exercise is eventually leading
to more than just pretty fractal distractions.
(and they are very pretty - don't get me wrong 
>In my first few positings I had the perspective D and screen position MY
>positive instead of negative, which made the perspectives look wrong. It
>was a typo in the book I got the routine from. Ish. Ya get some weird
>perspective views by changing these values. If I get around to it, I'll
>do a version where you can place the camera and target and it will give
>you a perspective view from the surface, which will look much cooler.
>I want to get a fast texture mapping thing working first though.
Sounds great, I look forward to it.
Graeme
|
Not Categorized, Please Help
|
|
