Suzy B Software 2/Suzy B Software CD-ROM 2 (1994).iso/picmanip/pic_r2z/raynews5/raynews5.txt

Text File | 1995-05-05 | 23KB | 440 lines

*************************************************************
*    RAYTRACE USER NEWS      ISSUE #5         JULY, 1992    *
*************************************************************

SOME DESPERATELY NEEDED PUBLICITY

The July 92 issue of ST FORMAT UK has followed up last month's
publishing of Raytrace 2.1 with a trio of handy utilities;
Rayview, a pic viewer I hacked from the demo code on the V2.1
distribution disks, Lonny Pursell's SUL2SPU Spectrum converter
demo, as well as a converter I had never seen that does 16 color/
grayscale Degas conversion.  Two tracings I've done got little
thumbnail pics to accompany the coverdisk article - Andy Warhol was
right, and I guess I got my 15 minues of fame!  That I sent Rayview
and SUL2SPU and some pics on disk off to the editors may have had
something to do with it...   But it proved to be very timely, keeping
Raytrace in the limelight for another issue, and provided those who
took the program for a spin over the last month a way to get the pics
they created viewable with something other than Raytrace itself.

In the Rayview .DOC file is my ongoing plea to see other raytraced
pics - hopefully some of Format's 70,000+ readers will send a disk
my way - how about you???  Hmmmm? (Nag, whine, etc.)


COMPATIBILITY PROBLEMS, PROBLEMS, PROBLEMS...

Also in the July Format issue, are two references claiming that
Raytrace 2.1 runs on TOS 2.06 - at least on upgraded ST/E's -
That was not my impression when I went to my Atari dealer and tried
to run Raytrace when the first Mega STE's came in  - all I got was
bombs.  It seems I'd been under the mistaken impression that TOS
2.X is the culprit, it seems not - so I've done what I've wanted to
do for months and upgraded to a new 1040 STE with 2.06 installed.

For the most part, Raytrace 2.1 runs okay - with one exception.
The program will bomb out when hitting the ESCape key to abort
a tracing in progress.  Whether this is a major problem or minor
inconvenience is your call.  You just have to wait till a small
1/8, 1/4 scale test finishes before returning to Edit.  The HELP
key still works, so you can get a sneak peek when you need to.
So that kills the theory that switching between video modes is
the cause.  Very puzzling.  All other operations work normally.

The more serious problem is that Spectrum 512 has a major conflict
the new TOS.  Oddly, the program loads just fine, and you can
select all functions with a mouse.  But ANY KEYPRESS will bomb out
the program - not acceptable in my book - too many of Spectrum's
functions depend on keyboard commands...  So unless I can find
someone to perhaps piggyback the ROM's and let me switch between
TOS 2.x and 1.6X I will just go back to the older TOS and trade
Newdesk's nifty features in for the compatibility I need.

I'm still very puzzled about the Mega STE, I know that although
Raytrace gets along with my AdSpeed STE at 8mhz, just having
a 68881 math chip plugged in killed Raytrace - Two Bombs.
Maybe my dealer's Mega STE has a 68881 installed and thats the
culprit.  So went to my dealer and tried it again, 8Mhz, no cache.
Two Bombs.  We didn't have a program to test for the presence of
an FPU, (they didn't know if it had one or not, duh!) so its still
unresolved as to what's causing the problem.

Please, if anyone out there can positively verify they can run
Raytrace on a Mega STE - whether it has TOS 2.05 or 2.06, and
whether or not it has a FPU installed - let me know and help me
settle this issue once and for all.

A related issue:  Screen accelerators such as Turbo ST and Quick ST
definitely had problems and prevented Raytrace from even loading.
The Codehead's upgrade to Quick ST - WARP 9 resolved some of the
problems.  At least Raytrace will load and run now - and Edit mode
is noticably faster when creating wireframe models.  But in the
testing process of switching from Edit to Trace mode and back I've
had the screen fail to update properly.  The Edit windows may not
refresh between editing changes and end up a scrambled mess of lines
- close, but no cigar as they say.  Can't blame the Codeheads for
this, since Raytrace isn't using real GEM windows in the editor,
and can't call the WIND_UPDATE function proper to do a refresh.

By creating a WARP9.DAT file, you can tell Warp 9 to automatically
turn acceleration off when RAYTRACE.PRG is launced if need be.
I'm still a little leery of this though.  Having a handful of AUTO
programs and ACCs in memory when Raytrace is running can cause
Raytrace to behave unpredictably - it's definitely one of those
programs that is best loaded from a cold boot...


THE 'NEW' MATH: 512 - 510 = 2

I've finally gotten a chance to play around with Raytrace running
in black and white by using Mick West's PD monochrome emulator -
MONO_EMU.PRG.  Other emulators I've tried have just caused Raytrace
to crash, but this one works pretty well.  The CPU overhead of the
emulator seems to be offset by not having to calculate all the
colors of color mode, so tracing times aren't much different.  I'm
gonna search some bulletin boards for a converter to get my color
texture maps into hi-rez Degas .PI3 format and experiment with this
mode further.  The 9G nine 'grayscale' dither pattern mode in
particular yields pleasing results both onscreen and printed out in
black and white.

Raytrace is rather unique in that it's almost two versions of the
program rolled into one.  The source code is interesting - rather
than writing separate programs for color and mono users, the code
frequently checks what rez it's in - if its not mono 640 x 400 then
halves coordinates, cricumference, length, etc., variables to low-rez
320 x 200 and acts accordingly.  Its an interesting example of how to
write resolution independent code.  But there is alot of overhead
in testing for this condition all the time and jumping to the
appropriate procedure.  Resolution specific versions of the program
would be much smaller and the Edit and Animate modes would probably
be more responsive and quick to redraw, though it wouldn't affect
rendering times.


THAT SICK FEELING...

Where is the future of raytracing going on the Atari platform?
The TT, and the upcoming Falcon have hardware that just begs
for a successor to GFA Raytrace, but I'm afraid that's far down
the road.  The Lexicor series of programs just doesn't seem the
answer - it still lacks its own 3-D modeler, Prism Render is an
add-on package that outputs non standard Targa files, and though
a powerful animation suite in its own right, just isn't an
integrated/dedicated raytracer, and isn't optimized for the bulk of
low end Atari ST/STE machines out there.  Its unfortunate I can't
afford to really see what this suite of programs can do - but if
you've got a TT and want to see what a successor to Raytrace is,
Lexicor is leading the way.

Maybe the Falcon series will change things.  With a fair amount
of horsepower, some excellent graphic specs, and some degree of
affordability, maybe someone will take up the challenge of
bringing a capable raytracing package to market for it.  There
certainly is a niche for one.  GFA seems to be moving out of the
applications market, so I'm not expecting the next generation to
come from them.  But I do appreciate what they accomplished with
GFA Raytrace.  As often as I bump into the programs limitations,
I still feel a sense of awe at what it can do, and know as the only
commercially successful tracer the ST/E has, it's set the minimum
standard for what the next generation of tracers must do.  I
honestly hope Atari and some software developer can come up with a
hardware/software rendering solution at a low price and keep us in
the Atari family for many years to come.


TOO BRIGHT FOR ITS OWN GOOD

A few issues back I mentioned that Bright objects will 'wash' out
the  ground  pattern,  turning it to white at least in the area
surrounding the object.  Quite accidentally I discovered that this
only occurs when the Bright object is close to the ground.  If you
raise the object up high enough, the ground pattern will return.
If you must have a bright object near the ground in a scene, try
using a large texture mapped rectangle as your 'ground' instead,
and this problem will not appear.


A WELL LIT SHADOW

Say you've got the classic Mirrored Spheres Floating Above The
Checkered Ground scene constructed, and it renders out well.
Except... the shadows from the spheres leave completely black
shadows on the ground pattern, where in real life you'd still
see the checkered pattern lit dimly.  Its the lack of ambient
lighting in Raytrace's world is the culprit.  To light that
shadowed area, you need to dedicate an additional lamp just for
the ground itself as shown below:

Spheres - o,O                             L1  +
Lamps - L                              O o         L2
Ground - ___                       ____________
Observer - +                         Side View

Here, because of its proximity, the main lamp (L1) in front and
above the spheres dominates the scenes lighting - as we want it to.
the second lamp (L2) is placed much farther away, and closer to
the ground to lighten the shadows, but keep its effect secondary.
This little trick is what can make a good scene great, and helps
restore the realism that a totally black shadow was destroying.

An auxillary lamp enhances nearly every scene, the trick is
experimenting with its placement/distance so the effect is subtle.
You may also want to try coloring this secondary lamp to add a
tinted glow to the side of objects.  It really helps to 'round
out' objects like the sphere, cone, & cylinder, since the sides
of these objects often get lost in blackness and you can't tell
where the object ends and the shadow begins.


WEIRD AND NOT SO WEIRD THINGS TO TRY

Experimentation and trial and error are pretty much a normal
operating mode for anyone using Raytrace.  If you persist long
enough to understand all of RT's features, you'll begin to start
exploring the edge of the program's limits, but more importantly
some of the limits you perceive are in your own head.  Here's some
examples of some unusual things I tried:

Underlighting:  Illogically, a Ground in Raytrace is NOT an opaque
object!  You can place a lamp _beneath_ ground level and it will
highlight the underside of objects placed on it and illuminate the
ground pattern itself.

The Glow Within:  What would happen if you placed a lamp INSIDE
a transparent sphere - would it glow like a light bulb?  No...
The sphere ends up black - but the rays cast from it would illuminate
nearby objects.  Raytrace is very quirky about multiple lamps and is
very sensitive to distance when multiple lamps are used.  I this case,
the lamp in the sphere would be dominant because of its closeness,
preventing other lamps from illuminating its outer surface.

Tapered Cylinders:  Something like a coffee cup seemed impossible
at first.  Only mugs are really cylindrical, but that isn't what I
wanted.  So I used a cone to give a tapered shape - by placing the
apex of the cone well below ground level.  (Then used an Ellipse
Segment to create the 'C' shaped handle.)

            C\-----/
          ____\   /_____G
               \ /
                v

You could take beverages a step further; to create a glass of
orange juice, rather than just use an orange cylinder object - think
of it as it is, two objects  - a glass, with orange juice in it - a
90% transparent cylinder object for the glass itself with another
slightly smaller orange dull cylinder of 'juice' inside it.  If you
wanted to view it from a somewhat overhead observer, use an ellipse
(perhaps a bumpy orange texture mapped one) to be the surface of
the juice itself.

Transparent objects: are quite challenging to get good effects with.
Its often important to have some patterned object behind or perhaps
a textured surface underneath them to really show off their surface
quality.  Transparent objects refract (bend) the light rays and
appearance of adjectant objects - as they should.  Keep your main
lamp at a reasonable distance.  Too close and the 'surface glare'
of the lamp will override the transparency of the object.

Tunnel vision: You can put your observer inside a cylinder and
get a view of looking down a pipe by adjusting your view angles.
Don't know what kind of scene its good for, but its possible...

Something to reflect on: Raytrace's mirrors are very farsighted!
Reflected objects often appear much farther away in the mirror's
surface than they should be - much like the view in a cars side
view mirror, or a Xmas ornament.  But one interesting thing to try
is a scene with objects in front of a mirror, but the observer placed
in-between, only viewing what's reflected, not seeing the objects
directly themselves.

A Not-A-Sky sky pic: I've never been a big fan of the Sky function.
Partly because sky pics get their brightness level from the colors
in the loaded pic itself, and has nothing to do with the lighting
level from the lamps used in a scene.  So often your ground will
recede into darkness at the back of your scene, then Boom! the
horizon of the sky pic will abruptly rise up.  To prevent this,
make sure the rear of your scene is adequately lit (Lamp BL):

BL=Back lighting lamp               BL
ML=Main lamp                       |       ML
O=Sphere                       Sky>|            +
+=Observer                     Pic |    O
|=Sky background                   --------------
                                     (Side view)

The alternate method is to 'zoom out' in the editor by editing at
a 1/1 scale and creating a large vertical rectangle at the back
of your scene (in effect, a Wall) the full width of your raytrace
world, and apply the sky pic as a texture map instead.  Now your
'sky' will be under the same lighting influence as all the other
objects in your scene.  This also helps preserve the aspect ratio
of the pic used: Sky pics tend to be scrunched - say your horizon
is at mid-screen - Raytrace literally throws away every other line
in your picture to squeeze it into the upper half of the screen,
severely distorting the pic.  Texture mapped as a Wall, the aspect
ratio will be preserved.


3-D TO 2-D

Raytracing is all about creating illusions - and particularly with
the ST's limited resolution and palette, your visual system and
brain have to make some 'leaps of faith' and plug alot of holes and
inconsistencies to perceive the realism you're striving for.
Someone once commented to me that they felt Raytrace created rather
'flat' looking scenes - but thats really just the result of bad
input.  In some ways exaggeration is needed when building a scene
to retain the depth you're striving for.

I've previously mentioned separating your observer and main lamp
in opposite directions to enhance shadows.  Keep chanting the
mantra of a good photographer or illustrator: 'light and shadow'.
But a second trick is to really play on the depth perspective to its
utmost by really moving your observer close in to your objects, then
using various Rayscale setting to 'wide angle' your view.  This is
especially noticable in scenes with more than one object, say a scene
with a pack of cigs, a matchbook, and an ashtray - here you have
multiple reference points for depth cues.

Think of 9th grade English class when you studied -prepositions-
(<note positions) 'the ball is _under_ the table', 'the rake is
leaning _against_ the garage'; atop, beneath, behind, over, beside...
If you can build a scene that visually contains many of these
words you're moving in the right direction.  There is a direct
correlation between the number of objects you have and the number
of aspects to each other that result in a convincing 3-D scene.


A B C's/1 2 3's

Raytrace's opening title screen shows how text can be created
with multiple objects.  If you look at the wireframe file included
on your distribution disk, you can appreciate just how much work
it took to spell out GFA RAYTRACE - 30 rectangles and ellipse
segments all together - and it takes EONS in tracing time.
(Especially since many lamps were used for the rainbow effect.)
There is a cheat to this, and a shortcut can be done by
using text created in a paint program and used as a Sky Pic
placed as a backdrop, or texture mapped onto a rectangle.  Both
methods tend to be reduced in size onscreen, so you need to
compensate for the distortion scaling down causes by creating
large text with adequate spacing between the letters.

Here are a few tips you can use in creating text the hard way:
First, let the program handle part of the placement for you.
The depth (Z axis) always snaps to the same Z coordinate by
default when placing objects, and its easiest to let the program
do that for most of the letter segments you place - its one less
coordinate you have to deal with, then move your observer if
necessary to keep all your text in view.  The exception might be
the letter 'A' where you might want to place the horizontal
segment one pixel behind the other two.  If you wanted to tweak the
positioning, its often alot easier to adjust and move objects when
they aren't all on the same plane.  And in the 'A' example, the cross
bar behind won't cast shadows on the other two letter segments.

Secondly, here's where the CONNECT function can be important.  To
give the appearance of a letter being a single rather than complex
object, the Connect function 'joins' them when it comes to how it
calculates the lighting.  You would do this with the EDIT/OBJECT/
CONNECT function by selecting each object in the letter and
assigning it the same connection number, so that the nearest lamp
treats it as a single object.

Another interesting possibility with text would be to texture
map the segments so that you can have wood or stone letters
instead of just differently colored ones.  This can be tricky
since _not all object types can be texture mapped_.  Instead of
using the ellipse segment to create the rounded parts of letters,
you need to use the mappable ellipse object.  You would then need
to mask over unwanted areas with a black object, or edit them out
in Spectrum 512 after the scene was calculated.

As an example, the letter 'P' would be built of a vertical rectangle
and an ellipse, the opening in the letterform could be created with
a smaller black ellipse placed one position just in front.  The
letter 'C' could be a mapped ellipse with a black triangle and
smaller black ellipse used to create the opening.  But...

You can save yourself alot of extra objects and reduce calc time
just by editing it out later in Spectrum - Okay, for the umpteenth
time, if you really want to get the most out of Raytrace - you NEED
to get Lonny Pursell's registered RAY2SPEC converter program and
Spectrum 512 - they are essential tools - whether you are editing
out color glitches, antialiasing to smooth your pics appearance,
or in this case - just saving yourself a hell of alot of time when
it comes to creating letters and getting the look you want.
The $5 shareware fee is well worth it, write to:

Lonny Pursell re:RAY2SPEC  Box 145  Pandora, Ohio  45877


THE BOTTOM OF THE BARREL...

In five issues, I'm pretty much tapped for any more truly profound
insight on wrangling with GFA Raytrace.  I've tried not to be too
repititious while emphasizing important concepts, and yet touch on
as many different aspects of the program as possible.  I hope those
of you who've taken the time to download and read them have found
them helpful and helped you move beyond the Stumble-Bumble phase of
the learning curve.  And more importantly, got a sense of the
program's broad possibilities.  I know its still my ability to put
all the pieces together that's the bottle-neck, and will continue to
push the envelope in the months ahead.

I feel the Animate functions of the program are just too limited to
require much explanation.  Only the most simple, loop-based sequences
are possible in 10 frames.  But it is worth exploring, I've done an
interesting swinging pendulum of a tick-tock clock and a 'follow
the bouncing ball' with a moving ground pattern that was quite
convincing.

Now that I've gained some mastery over Spectrum 512 - particularly
it's Block/Scaling functions - I can merge two or more tracings onto
a single screen.  Its a way to show side-by-side comparisons of
some of the tips I've given, and how minor changes in lighting,
perspective, and approach can make dramatic improvements to a
scene.  There's not much need for another issue of RAYNEWS proper,
but keep your eyes peeled on GEnie, I may create an archive of
such pics to serve as a visual reference to some of the techniques
I've described in these issues.  And of course, if I get my mitts
on some good texture maps, or come up with some rendered scenes
worthy of public display, I'll be sure to upload those too!

I still have hopes for a Spectrum converter that can handle .SCL
-compressed- pic format, tweaks to resolve the minor problems
with machines beyond TOS 1.6x, and more...  GFA has sent me the
Basic and Assembler source to the program - if you feel you have
the programming talent to help make Raytrace better, more compatible,
or come up with a direct to Spectrum format save, please contact me.
Since GFA doesn't intend to provide further updates, if somebody
can come up with enhancements to the program, maybe we can negotiate
with GFA to allow public distribution of an update.


I'LL SHOW YOU MINE IF YOU SHOW ME YOURS!

I seem to operating in a vacuum here.  After a full year + of
exploring Raytrace, I've tried hundreds of different scenes with
varying viewpoints and subject matter.  (Many of which looked
better in my head than they did onscreen...)  At this point, I
really need to see examples of other people's work to broaden my
perspective and show me some fresh approaches.  I'd really like to
see what you've been able to do with Raytrace so send a disk my way!
I'll send you back a disk with my best pics, a .SUL pic viewer, the
demo version of Lonny's Spectrum 512 converter, and the back issues
of Raytrace User News in exchange.  Not a bad deal for the price of
a few postage stamps.

If you're a GEnie subscriber - drop by GFA Category 22 Topic 4
in the ST Roundtables and help me keep a discussion of Raytrace
lively, or e-mail to S.SMITH65 and let me know you're out there!

Write to: RAYTRACE USER NEWS
          c/o Nick S. Smith
          4406 5th Avenue South
          Minneapolis, MN 55409 USA