Elite Demo
CHARVEL DEMO DISK#1 October 1991
Welcome to the first (official) Charvel Demo Disk (Don't buy
inferior versions), featuring:
*A three dimensional development collection from the basics to
active three dimensional environments.
*A hands off (the keyboard onto the joystick!) three dimensional
Elite Demo.
*Pictures drawn with peintre,eye q and painter (in that order).
*QSPEC a spectrum compatable tape system with conversion routines.
3Dv1-5 value meanings.
RI R oll (0-254 step2) X co-ordinate (wraps)
HI H eading (0-254 step2) Y co-ordinate (wraps)
PI P itch (0-254 step2) D istance (wraps)
I ncrement (for X,Y&D) P erspective (or off)
(MODE) (DIRECTION) (STYLE) F rames per second
RI,HI & PI:
These are increments for the roll, heading and pitch used during
rotation (by ROT or R+D mode or space and left/right). I.E:
If RI=1, HI=2 and PI=0 after each rotation frame roll would
increase by 2 (1*2) heading by 4(2*2) and pich by none.
Modes:
ROT ate continuous, DIR ect rotation, R+D rotate and DIRect active
Direction:
Rotation direction L eft, R ight (with using space and left/right)
Style:
O verdraw (don't CLS after frames), L ucid faces (Draws one face
each frame), W ireframe (no hidden lines).
A HISTORY OF OUR THREE DIMENSION EXPLOITS.
3Dv1
This was a result of initial experiments into three dimensional
graphics back in early 89. A myth hung over Rob and me (Dave) that
it was a very complex subject but in it wasn't, just a serial
rotation around three circles, easy! Having already written a line
routine it wasn't long before our first object was moving on the
screen (Actually both of us wrote versions during a late night and
early morn at Rob's nan's but Rob's didn't work).
Rob's the mathamatician and realy without his assistance I
probably would have stuck to 2D programming (for better or worse).
Next was a line clipping routine. Again the battle was set, Rob and
Me were at ends to write one in Basic first; Rob did, followed by me
,though mine was shorter and Rob's was an endless list of exception
clauses. After studying the problem from a clearer view at work (in
a Pub) I found a tidier logic method. It was soon written in basic
(much better and shorter) and then converted into machine code (Rob
did the math half I did the logic), after a heavy night. Rob fixed
some signed branch decisisions a bug that I was too tired to notice.
I wrote some control code for the 3D routine which checked
keys and displayed values. We both designed and coded graphics. I
did the perspective (in a couple of minutes) after Rob with all his
mathamatical alogarythms failed. The perspective can glitch though
when the object is too close because values are magnified so much
more due to the logarithmical nature of perspective. Perspective
also looks wierd on wireframe.
Version 1 was tidied up after version 2 to conform to some new
keypress's and visual layout.
3Dv2
The step to the faced version 2 was realy a big one. We
laboured on a the problem of whether a face was visiable or no for
over a week trying numerous methods in Basic. Rob came up with the
most, half of which I couln't understand and insisted wouldn't work,
and they didn't! Finally Rob came up with a method that worked, and
we employed it. It worked by each face (side) having a point that
was especially coded, which was on a vector at right angles from the
object's centre and the plane of the face, and if after rotation the
z value was positive the face was not visable. The only problem with
this method was calculating these points! It was 1% calculation and
99% guess work, and coding all version 2's graphics was a real bane.
Perspective doesn't work with this method either but was kept
anyway. The object data was completely diferent to that of version 1
and took up much more memory.
Mark appears on the scene (a weirdo rocker come sex maniac who
liked computers (the usual specification for an Amiga owner)). We
drew all the Elite ships from the Elite manual and the Ami Elite
startup (where all the ships appear, stay a bit then dissapear) in
Mark's bedroom one evening (We discover Rob who finished his third
in no time can't draw (his Thargoid looked like an upside down
yoghurt pot)). Mark discovers he isn't able to code in the graphics
(despite much primary style tuition) after he realises how boring it
is (smart kid).
I finish Version 2 and make many modifications to the control
code and other subroutines including a rewrite of the line clipping
routine after seeing a 3x3 boolean matrix method in a book (one
glance was all I needed. "Of course!"), it made version 1 run 15%
quicker mainly because it was self selective of the maths it had to
do and the Scrolly, (painfully slow, and I never replaced it after
the version 3 re-write) Mark drew the character set from a typestyle
book that I chose after winning after a guess the random number on
the calculator game (I actually won it three times ater the first
two were ruled void by Rob and Mark (predictable tactics!)).
3Dv3
All change again. Rob and Mark become rather allergic to QL's,
I'm on my own (again). Version 3 needed a fast shape fill routine
and it was a long time comming. I started it many times leaving bits
of code going 'in the wrong direction', though not essentially a
very complex program, but it had many ways of doing it (programmer's
bane), and I'm not good at deciding best ways and get frustrated.
It was written, and perfected many times which resulted in my most
prided piece of code. One thing that also came out of the FillSHAPE
routine was that it also did the face check (no more face points,
"YIPEE!").
After a quick sneak at a book in the local computer shop a
diagram explained all, points that were in a rotational order, so if
all the points were coded for example in a clockwise order from an
external viewpoint and when rotated were then anti-clockwise the
face wasn't visable. Easy, and when the FSHAPE routine goes to draw
a horizontal line and the points are not left to right when drawing
a face shape it's not visable. Another way to do it was find the
top point and calculate gradients to the two points left and right
of it, if the gradient to the right point was less than that of the
left point then it's not visable. Perspective worked fine with these
methods.
The scrolly was re-written (try messing with the cursor keys,
enter and \). Code was written to enable it to run on 128k machines
though it would not return to basic (CALL address,128) and it was
ported to the earlier versions. The new object data structure was
much better, and all those stripples (press Y), I did all the
colours (boring). Did some more graphics, and could get away with
more being able to overdraw. I was very pleased and relieved on the
final speed of the solid version. Seeing the 3D routines running on
a Gold Card makes me wonder why I spent so long perfecting routines
to obtain optimum speeds. Awesome!.
3Dv4
"Cor, wouldn't it be great to see x-wing in solid, DAVE!" said
Mark, after seeing version 3, hence version 4. It was meant to be
able to display any complex object, and it could cope with many but
the method was too simple to be very, effective, it virtually worked
with x-wing (two very minor gliches) but struggled on the guitar and
bird of prey. Version 4 decided not suitable, and left.
3Dv5
All change again. Wireframe, Faced and Solid combined version
(Use 'A' to toggle). The main problem with v5 was the data format
and the colours (I actually couldn't be bothered with them). It was
inevitable that this was written and as mentioned e The line routines were
later significantly speeded up 6-20% making the wireframe and faced
outputs extra zippy, with a smaller improvement to solid and scrolly
speeds.
3Dv6
Another try at complex shapes. Close but not perfect, faster
than version 4 and more adept, though and also with view-(A)xis and
0,0 cross(W)ires. It still looks like I'll have to call on even more
complex methods (and a considerably slower). I'm now thinking about
light shading. Hmm, that would be nice. Now lets see, plane normals,
colour list... The rotaing head looks alright, but thats about all!
I've worked out how to store 150+ equally complex animation frames
in 512k! Maybe later...
Elite Demo
A pretty boring demo, but I got fed up with it. I may write
the proper thing soon any way.
Dave Barker 22nd August 1992 (p.s. I need a job!)
Y & T the worlds greatest ever rock band. 1976 - 1991.
Dave Barker Demo disk assembly
021-350-2252 Programming 3D1, 3D2, 3D3, 3D4, 3D5, 3D6, Elite Demo
Qspec Picshow, All 2D & Some 3D Graphics
Robert Nash Some 3D Graphics
021-356-6875 Some programming 3D1,3D2 (visual face theory#1)
Mark Handley Some 3D Graphics
021-588-5532
[ back to the prod ]