stalking elusive GUI's

stalking elusive GUI's

Post by Rion D L » Fri, 08 Mar 1996 04:00:00



Hi Everybody:
        I'm a programmer, using a linux/MS platform.  First, i'd like to express my
appreciation for all the shareware/freeware developers that make computing fun.  I
have no regrets about paying for an application which i find useful.  As an aside, when
i first started writing html, i went out and grabbed h-text editors, with all the little
pointy-clicky thingys.  It wasn't long though before i realized that it was easier
to have a few xterms open and simply cut n paste in a plain ol' editor.
        Being new to raytracing, i grabbed the available engines, pov (2.2, 3.0),
polyray, radiance, and all the GUI's i could find; pored through the docs and man pages, and took em for a test spin.  2 weeks later and what i've got is 40-50M worth of somebody's hard
labor and very little images.  On the linux side, mnm is nice but the colormaping is a problem.
tkPovShell and tkpov are both limited  (in beta) with  essential functions disabled or
nonexistant.  I haven't figured out radiance yet; theres tkrad as a gui, but i'v got so
many versions of tcl/tk .... bla bla bla.
        On the windows side, ive got GUM, but i guess i haven't got it configured right,
because the library editor always gives me error messages. PovLab is the best looking
interface, with all the bells and whistles, but running under dos4gw, it bombs before it
can render anything.  Besides, it's a good thing to be able to have extra windows open
with help/reference info.  I'd almost want a dedicated pc to use povlab.
        I realize that complicated images require a modeller.  I'm beginning to think
that one modeller isn't sufficient.  I'm also wondering if a good imagination and fast
typing at the command line isn't the most effective way to create a simple picture.
i'm d/l'ing breeze now, but for all the time i've spent hunting the elusive modeller,
i'm beginning to think i could have saved some disk space, used pov, and made some nice
graphics.
        Could some of you tell me your experiences? What you started with, what you're
using now and why. Also how you've got dos4gw set up if the app requires it?  email
responses will be greatly appreciated and followed up.
                        Thanks for your time

 
 
 

stalking elusive GUI's

Post by Chip Richar » Sat, 09 Mar 1996 04:00:00


   I have no regrets about paying for an application which i find useful.

Same here.  I'd like to see more of that attitude among the Linux community.
There's a widespread "Gimme everything for free" mindset right now.

   I realize that complicated images require a modeller.

Not necessarily.  Some of the most stunning images I've ever seen were created
by hand, or with only minimal use of a modeller for certain shapes.

   I'm beginning to think that one modeller isn't sufficient.

Depends on your style.  On average, though, I'd say you're right.

   I'm also wondering if a good imagination and fast typing at the command
   line isn't the most effective way to create a simple picture.

Again, it depends.  These days, I usually start with the modeller for simple
scenes, sometimes switching to hand-editing if things get tricky.

   Could some of you tell me your experiences? What you started with, what
   you're using now and why.

I started raytracing in 1979, I believe; may have been earlier, but I didn't
do much in those days.  I used custom software (mostly written by others,
mostly in FORTRAN or assembler) and extremely expensive hardware (usually
called "graphics terminals" back then).

When I found POVray around 1989 (may have been '88 or '90) it was a real
boost.  I started ray-tracing quite a lot after that, using DOS and the
standard graph-paper-and-ruler modelling system.  When I graduated to Unix in
1990, I found it a *very* friendly environment for POV.  In the past year or
so, I've discovered the joys of using the sced modeller under Linux.

Nowdays, I use sced mostly, with help from some external modelling programs
like gforge, levcon, etc., and occasionally even mnm.  I agree with you (and
with others) about the horrid (and unchangeable) colors used by mnm under
Linux.  And now that sced has wireframe input, and native triangles and bezier
patches, I find I hardly use mnm at all any more.  Contrary to what I've heard
from others, I find sced's 3-d view very natural, and find the fixed X-Y-Z
views of mnm to be limiting.  Sced could still use some features, but it is
*extremely* nice even now.  And it can write Radiance, Renderman, and VRML, as
well as POV (and others too, that I don't care about).

I think my years of using POV input language directly make it easier and more
fun to use modellers.  I'm not afraid to look at what the modeller has
written, in case I want to understand exactly why I'm not seeing what I
expected to see, or in case I want to add a few more tweaks that the modeller
can't do easily.  Often, I'll code a POV include file or three with some weird
objects or textures in them, use a support program and/or mnm to create one or
more wireframe objects, then bring up sced, create the bulk of my scene using
sced's native objects and my created ones, then export the POV source and
trace it.  Sometimes (though I try like hell to avoid it) I will hand-mangle
the exported POV source before tracing it.

I suspect I'll probably use direct input with POV 3 (oh when oh when is the
Linux build coming out?????!?!?!?) until I'm familiar with the features.  I'm
not sure what to do with animation--so far, I've used a motley collection of
perl scripts to animate scenes, eschewing the POV clock variable and sceda
techniques.  With the better animation support in POV 3, I may change that.

   Also how you've got dos4gw set up if the app requires it?

Sorry, I've switched totally to Linux; I don't have any DOS/Win on my system
any more.  Good luck getting it set up!

--
Chip

 
 
 

stalking elusive GUI's

Post by Michael J. Hamme » Sun, 10 Mar 1996 04:00:00




>    I'm also wondering if a good imagination and fast typing at the command
>    line isn't the most effective way to create a simple picture.

> Again, it depends.  These days, I usually start with the modeller for simple
> scenes, sometimes switching to hand-editing if things get tricky.

I agree.  Modellers, if nothing else, make the simple things easy.  The
measure of their usefulness is probably on how easy they make the more complex
scene generation.

Quote:>    Could some of you tell me your experiences? What you started with, what
>    you're using now and why.

I've only been at this for a couple of months.  I don't even have any really
interesting renderings to offer up (but will by the end of the month, methinks).
I tend to gather as much info as I can on a new subject and then weed through
it to find a good starting point.  For POVRay (and raytracing in general) this
helped me sift through the terminology and tools and so forth.  Then I grabbed
a fairly simple scene of a space fighter and started to manipulate it by hand.
I ended up with what looks like a dimented washtub in space (which I proudly
display on my web pages ;-).  Not pretty, but it gave me a base on which to work.
I learned how the scene files are organized, which parts are generally mandatory
(gotta have lights and cameras!), and so forth.  Next I jumped into modellers.
I now use sced on Linux (don't use MS environments much) and am quite pleased
with it.  I just finished my first "interesting" model - an android which will
make its appearance as the mascot of my raytracing web pages!

Quote:> I started raytracing in 1979, I believe; may have been earlier, but I didn't

Gads, I didn't even know this stuff existed till Toy Story came along.  I really
gotta get out more often....

Quote:> Nowdays, I use sced mostly, with help from some external modelling programs
> like gforge, levcon, etc., and occasionally even mnm.  I agree with you (and
> with others) about the horrid (and unchangeable) colors used by mnm under
> Linux.  And now that sced has wireframe input, and native triangles and bezier
> patches, I find I hardly use mnm at all any more.  Contrary to what I've heard
> from others, I find sced's 3-d view very natural, and find the fixed X-Y-Z
> views of mnm to be limiting.  Sced could still use some features, but it is
> *extremely* nice even now.  And it can write Radiance, Renderman, and VRML, as
> well as POV (and others too, that I don't care about).

I originally said I preferred MNM's multiply pane display to sced's interface.
Now that I've played with sced a bit I have to say its quite easy to use.  I
just had to get used to rotating the wireframe display often.  In conjunction
with the Layers and various View functions I can easily position any object.
My apologies to Stephen Chenney for doubting his design choice.  :-)

For now I can't say much about the various output formats.  I expect I'll
appreciate the Renderman format before too long (I'd like to get hold of BRML (?)
at some point, but I'm just not there yet).  I think a drawback to this might
be the fact that it may limit SCED's ability to handle renderer-specific
functions.  I'm not sure of that, but I really miss the ability to do clippings
in SCED.  I've assumed its missing due to the difficulty in adding it and
maintaining compatibility with all the other output formats.

Quote:> I think my years of using POV input language directly make it easier and more
> fun to use modellers.  I'm not afraid to look at what the modeller has
> written, in case I want to understand exactly why I'm not seeing what I

I can warn newbies:  don't try to understand the sced output until you learn
about the "matrix" command.  Its a bit confusing (thew me off completely since
the Waite texts don't have any reference to it).  However, once I figured out
its nothing more than a positioning tool, I was able to create objects using sced
that were nothing more than place holders for objects I create by hand or with
some other tool, like helix or geodome.  Just make sure you name the objects
something thats easy to jump to in a text editor.  

Quote:> I suspect I'll probably use direct input with POV 3 (oh when oh when is the
> Linux build coming out?????!?!?!?) until I'm familiar with the features.  I'm
> not sure what to do with animation--so far, I've used a motley collection of
> perl scripts to animate scenes, eschewing the POV clock variable and sceda
> techniques.  With the better animation support in POV 3, I may change that.

I'm almost scared to move to 3.0 simply because I haven't come close to
exhausting what I can do with 2.2.  And I have nice handy printed texts for
2.2. :-)  But I'm intrigued by all the things I've been reading about 3.0,
especially the atmospheric effects.  John Beale's image of the lparser
tree backlit through fog was great!

--
Michael J. Hammel          |

http:/www.csn.net/~mjhammel|
---------------------------+

 
 
 

stalking elusive GUI's

Post by Chip Richar » Sun, 10 Mar 1996 04:00:00


   I've only been at this for a couple of months.

Well, I gotta say that your "Unix Utilities" web page is certainly well-done,
even for someone who's been at it for a long time.

   I just finished my first "interesting" model - an android which will make
   its appearance as the mascot of my raytracing web pages!

I look forward to seeing it.  I never saw any spacecraft on your web pages--is
it cloaked? <grin>

   I originally said I preferred MNM's multiply pane display to sced's
   interface.  Now that I've played with sced a bit I have to say its quite
   easy to use.  I just had to get used to rotating the wireframe display
   often.

Yes, frequent rotations are my best friend.  Get busy with that shift key!
What I do is decide on an initial camera angle, then set the camera to the
viewport.  Then I can feel free to spin the scene as much as necessary, cuz I
can always pop back to the "real" viewpoint using "Window -> Recall ->
Camera".  Sced lets you create as many viewpoints as you want, either in its
config file or on the fly with Window -> Save.  I haven't really needed that
much power so far, but it's nice to know it's there.  And with the three axial
views he gives in his example config file, you can simulate the front-side-top
views of the old CAD modellers well enough.

   In conjunction with the Layers and various View functions I can easily
   position any object.  My apologies to Stephen Chenney for doubting his
   design choice.  :-)

He's a pretty smart dude, all right! <grin>  It ain't perfect, but it's quite
nice enough for daily use.

   For now I can't say much about the various output formats.  I expect I'll
   appreciate the Renderman format before too long (I'd like to get hold of
   BRML (?)  at some point, but I'm just not there yet).

BMRT--the Blue Moon Rendering Tools, by Pixar's Larry Gritz.  A fine package
indeed, with excellent performance and output quality.  But before you get
into it, be sure to get a copy of THE RENDERMAN COMPANION at least, and
preferably any other books you can lay hands on that talk about shaders.  The
Renderman language is almost primitive compared to POV, but the programmatic
shaders make its power essentially unlimited.  Sorta like comparing MS-Word
with TeX.  I surely do hope that POV 3.1 (or at least 4.0) incorporates
something like the shader patches that exist for 2.x.  I have so far been
afraid to spend the time incorporating and using them, but it sure sounds like
a killer combo--POV with programmable surface characteristics.  Wowee!!

   I think a drawback to this might be the fact that it may limit SCED's
   ability to handle renderer-specific functions.

Well, I'm guessing that Mr. Chenney is torn between wanting to offer common
features and wanting to exploit the power of individual renderers.  Some areas
lean one way, some the other.  Merges, for example--why leave them out?  A
note in the manual about which renderers support them would probably be
sufficient.  (Is it *only* POV?  Maybe.)  I've often thought about making a
patch to sced to support them, but so far I've just spent the time (in the few
cases where I've needed it) to change the exported source before my final
render.  Sigh.

   I'm not sure of that, but I really miss the ability to do clippings in
   SCED.  I've assumed its missing due to the difficulty in adding it and
   maintaining compatibility with all the other output formats.

Clippings?  As in "clipped_by"?  I guess sced presumes that you'll use CSG
intersections instead.

   I'm almost scared to move to 3.0 simply because I haven't come close to
   exhausting what I can do with 2.2.  And I have nice handy printed texts for
   2.2. :-) But I'm intrigued by all the things I've been reading about 3.0,
   especially the atmospheric effects.  John Beale's image of the lparser tree
   backlit through fog was great!

Hehehe, darn that Beale!  Thanks to your web page, I found out that his HF-lab
was a replacement for gforge, and thanks to HF-lab, I've got more modelling
power than ever.  Now if someone would just make a Linux version of lparser!
(That's one I couldn't run under DOSEMU, and I've since toasted DOSEMU anyway.
I want native!!)

None of us have exhausted the potential of 2.2 yet.  So?  You'll go to 3.0,
you know you will! <grin>  I will too.  It's just too good to pass up!  But,
whine, whine, where's the LINUX BETA?????  I know, patience, patience.  Guess
I could re-install that blasted DOSEMU and run the DOS version, yuck.

--
Chip

 
 
 

stalking elusive GUI's

Post by Maan M. Ham » Mon, 11 Mar 1996 04:00:00



Quote:>..........................................................
>I haven't used lparser yet, although I've seen a bit of its results.  The
>"Ray Tracing Worlds" text talks about it at some length.  I can dual boot into
>MS-DOS/Windows so I'll probably grab it from pov.org at some point.

By the way, you can always find the latest lparser at Laurens Lapre's
home page at:
http://www.xs4all.nl/~ljlapre/

Maan M. Hamze

 
 
 

stalking elusive GUI's

Post by Chip Richar » Mon, 11 Mar 1996 04:00:00



   > > I started raytracing in 1979, I believe; may have been earlier, but I didn't

   That's funny.  I didn't think ray tracing existed until Turner Whitted's
   article in 1980.

I assume you are talking about his 1980 CACM article "An Improved Illumination
Model for Shaded Display".  I think if you check your references, Whitted
didn't actually invent ray-tracing, though he contributed to its current form.
A cursory check of the bibliography in my Foley and van Dam shows references
to the ray tracing technique from Appel, Goldstein, and Nagel as early as
1968.

However, I certainly didn't intend to mis-represent my own history with ray
tracing.  I didn't invent it, nor was I by any means a pioneer--I've pretty
much always been playing catch-up.  Heck, now I'm wondering when I first
downloaded POV ... maybe I got that wrong.  Gotta check those floppies!

I *was* working on my own ray-tracer (written in Modula-2) when I came upon
POV.  I tidied up my working directories, put the project on the shelf, and
never (well, hardly ever!) looked back.

--
Chip

 
 
 

1. Pickover's stalks in mandelbrot

Hi, algorithms for Mandelbrot stalks can be found in
Chapter 27 "Turning a Universe Inside-Out" in
Pickover, C. (1992) Computers and the Imagination.
St. Martin's Press: NY.

2. READ ME OUNCE . . . Riiiiiight!

3. OT: Cyber Stalking

4. NURBS multiplicity 4

5. Cyber Stalking

6. Best Riva TNT Driver for SI 3.8 ? Problems !!!

7. 2-D Clipping and GUI's

8. FREE GUI's

9. Windows FONT3D/GUI Shell ul'd to ftp.povray.org

10. Learning to program GUI's

11. FREE GUI's