home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Amiga Format CD 28
/
amigaformatcd28.iso
/
-seriously_amiga-
/
programming
/
amos
/
a3dmodelsim
/
readme
< prev
next >
Wrap
Text File
|
1998-05-09
|
12KB
|
256 lines
A 3D simulation of radio control model jet.
(using amos 3d)
Author: David Lawrence. e-mail.. muddle@voyager.co.nz
Best with fast A1200. ( There is a crude time limit to prevent very fast
machines from producing uncontrollable flight. Remove at will)
...........................
Notes on A500 use.
A500 with 1mb, will ONLY WORK IF the user has Amos Voodoo 3d xtn. to run the
source code. Ideally control and speed variables should be changed to allow
for slower frame rate. Then the sound frequency variable would have to be
changed from it's direct relation to the SPEED variable. ( you have to poke
hardware to alter frequency while a sample loop is 'playing' in amos.
See code)
An amos autoexec floppy disc version of the program will also run on
an A500 with 1 mb. Which would be nice for those without amos or the
3d extension, but I don't think this can be archived to aminet in a
way that will work..?? And would that be legal?
Any advice .. anyone?
also.. I'm not too keen on making and sending floppy disks all round
the world from way down here (new zealand)... Sorry.
........................................
Will run on PAL or NSTC, the last by accident since the model sits on the
ground in sight in nstc with some screen out of view. Ideally the user could
redo the prog. to smaller screen.
The archive includes:-
- a compiled workbench version for non-amos users
- amos source code
- the necessary sample bank, screens, 3d objects
- ( note- the 3d library is part of the compiled version ..it's not
useable as an amos extension.
Also note, the 3d extension referred to herein is
the voodoo 3d extn, not 'stars' or other)
The program expects all the above archive to be extracted into
the same directory.(any name)
Please advise author if things don't appear to be working, as I can't
be quite sure if the prog. is relying on something on my hard drive
which may not be on others. I think it's just getting the normal math lib.
-------------
This archive is put here so that other amos programmers can freely
modify, improve on, (or just have a good laugh at ) one persons attempt
to overcome rotation limits in amos 3d. It should be possible to make
some sort of more interesting game out of the basic program. e.g. Missiles,
3d targets, or 'round the pylons' racing against a computer opponent.
Or for those who can happily use the object modeller, one could improve the
plane or convert it ( and the sound bank) to various curious aircraft.
Or to a flapping dragon or whatever.
The whole scene could be changed from a vaguely N.Z. rural flavour to
an airport, or fururistic .....
The background could just be a 'rainbow' horizon to enable greater speed
for more 3d interaction. The rainbow has to move up and down with the
view elevation. It works, I tried it somewhere on the old A500 for speed.
It might be best to just poke a custom copper list on the fly to create a
split or change in a colour register(s) at a movable raster line.
( A command to do this is an almost incomprehensible
ommission from standard Amos, of a basic, simple and very useful command..
The ' rainbow' commands are usable, but complex and limited. Good grief.
you can only change one colour register for a start.. a sad effort I think )
Amos programmers, note that the full rotation in 3d requires a background
much larger than a standard screen can be. Hence two linked screens
are used.. anyone who knows how to simplify this, or create a larger screen
is hereby urged to tell me....
If A3d didn't use a masking method to draw one could use the machine code
copy command, which would be faster and could handle the size... but...
. ......................
If anyone wants to obtain amos 3d I have no suggestions that are legal.
You could try annoying Europress?. They are unlikely to respond, and it
sounds like maybe someone else has the rights to the 3d code and object
modeller. (But you could still annoy them any way..... )
It was on a mag. disk some years ago, you may find a friend who has a copy.
They must be around somewhere.
As an alternative :-
It should be possible to use AMCAF vector rotations for Amos pro to make a
similar program. I think that's what I would do now if starting again.
And you could make use of full colours etc. And make a better object modeller.
.....at least as long as you kept it all simple.
But it would be a LOT OF WORK,and probably need some assembler routines
if you wanted to handle large numbers of co-ords. Or to check for surfaces
and lines both in front of and behind the viewpoint.
This would not matter with a simple 'object way out there' model sim.
........ ....................
ABOUT AMOS 3D
This writer was depressed to discover that amos3d is not ideal.
It is quite fast for a blitter system, and I find the commands fit my
approach .
BUT...
There are very restricted colour choices, it can write over screen
edges and crash from (seemingly) memory problems, and since it uses
masking to draw objects ( probably to gain speed), it is incompatible
with other drawing operations. Also the screen conversion doesn't handle
off screen positions.
And the object modeller though it has some good ideas, drives me completely
up the wall !!!!..
Hence the minimal objects in programs I make).
But above all the rotation angles are limited, and do not match the
manual which implies the ability to rotate objects and viewpoint about
their own axes with simple commands..
In fact the z rotation axis of objects and viewpoint remain
permanently aligned with the z axis of the co-ord world.
It did not have to be that way as the amcaf vector rotations for example
seem to rotate relative to their current alignment.. using matrix maths.
It may be possible to use complex sin* cos formula to make amos3d
work that way, but all my attempts failed at the angle extremes..
(My '3dtests' dir is full of things like;- 3dcalc, novertcalc, nearlythere,
finally?, .... very depressing.)
-------------------
The partial solution used here and in my other programs, is to change the
normal point of view and hence the apparent way the objects rotate.
It's as simple as setting the initial view angle at rt. angles to normal
on the A angle axis .
And then being sure you put your objects in the new world ground plane
of X, Y co-ords (instead of x,z as normal) with their A angles also moved
one rt angle...
It may be clearest to visualize as follows;-
Imagine you have stuck your head into the co-ord world through your monitor
and are looking straight down. Now twist your head to look to your right.
The effect from your point of view, is that the world z axis now seems
vertical ( i.e. off the top of your head.. Ha! joke) while the x goes away
into the distance ahead and the y goes from side to side.
Your ground objects are put on the new ground plane,( which from the
old point of view was the plane parallel with the screen )
For convenience I chose start angles which gave positive z upwards,
positive x away into distance, y positive to the right, but the x,y
(set by initial angles ) could be any chosen direction.
The main thing is that while the z axis rotations still remain aligned
with the 'world' z axis, this is no longer the longitudinal Z axis of the
objects and viewpoint... from your new point of view.
--------------
It is possible now to point an object in any world direction on the
z axis (angle c), elevate with angle B, and roll any angle with A.
I'll call this;- mode3
The limits of this are:-
1. The object can not use td forward, since the rolling angle A
is not the axis used by td forward.
Instead you have to use a sin/cos formula to move.( see source code.)
2. since the viewpoint cannot align along the A angle/ axis
the VIEWPOINT cannot be flown about as in a proper flight sim.
3. A seperate 'yaw' response is too difficult to arrange.
None the less it is possible to fly an object around properly if
it is saved from the object modeller at right angles (pointed left) to the
normal angle. (if you used td forward it would go sideways)
This is the mode used by the jet model simulator. It was very tricky
to understand how to adjust the rotation angles as the object pointed
upwards in the world.. because the roll angle (A) gradually becomes a change
in the direction in the world. (angle C )
It's in the source code, but if anyone needs more I could supply extra
notes... once I've re- worked it out myself! (Don't think this was worked
out by a blinding quick brain. It mostly happened by hours of experiment
over many months, and step by step advance. NOW it seems simple)
..............................
It is possible though, to use td foward for objects and viewpoint
with the world in this orientation, if you use angle B for rolling and A
for elevation. Angle C still sets world direction.
Call this mode 2
The limits of this mode are:-
1. the B rolling axis is always parallel to the 'ground' plane.
(Though you can point it in any direction on that plane.)
Therefore when you elevate with angle A and try to roll with B, the object
or view 'skews' or 'barrel rolls' it's nose around in an arc.
The greater the elevation the greater the skew.
It is possible to use a simple sin (or cos?, I forget) to reverse the
skew in the inverted part of the roll.
This is just enough to make it useable as a limited flight
sim with a proper travelling viewpoint but held within limits up and down.
See example. ( to be uploaded to aminet if this gets there!).. A3Dsimplesim.
Or to use to fly independent targets or enemies which bank and
fly with some realism.
See example .to be uploaded .. A3Ddefence
There are other possibilities with both modes, and there are probably
proper games that could be written along these lines- especially for fast
machines.
.. but the writer is done for now. Too much trial and error.
I should mention that the td 'bearing' command does not work in the
new world order. It works only for direction in the x,z plane and not
the new x,y plane. I have a method which should enable it's use, but since I
never needed the command it is untried. ( You have to substitute the co-ord
positions as if they were in the old mode, use td bearing command, then
put results back into the new world.. I think. e-mail if desperate)
Td face command is also affected- it puts the object horizontal on the OLD
ground plane i.e. stands it on end in the new world, and then spins it on
that plane to face the viewpoint or other ob. It would be rather
complicated to try and use that.
------------------------
Feel free to use anything here for your own programs. Or to upgrade the
whole thing, I have spent too much time worrying to optimise !
There may well be some other way to improve A3d .. go to it.
I really believe that a complete new, and faster 3d system- using fast ram
(bit planes or chunky) and cpu drawing techniques is what is needed for amos.
It may be possible to do quite a lot with the amcaf vector rotation
commands. But then, one needs a polygon command which doesn't crash
the machine when off screen right somewhere!
I hope someone might come up with something before the Amiga expiry date
passes.
Or my expiry date!!
Also... sorry to have not got this stuff out much earlier when amos and 3d
were still around, and this might have been a bit more use.
But I felt isolated in a bit of a personal black hole, and the internet
was not on my event horizon at all.
Good luck all. David L.