#Sparks
43 messages in this thread
Can anyone give me an idea how good the Sparks front end for light wave is? I'm
considering something like that, but haven't seen too much about it. I am
curious about how easy is it to use? Does it get cumbersome when you have a
large number of particles? 10, 100, 1000 ? Does it create motion paths, or
what trick does it use to animate the particles? Any info would be helpful
Thanks – Joel
Currently, LW seems to be limited to about 400 objects, each particle is a
seperate object.
The interface is somewhat limited, but I would recommend trying to contact
the developer directly. Macrosystems is no longer distributing his
product.
Can't find the number right now, but Adib Nasle is supposed to be one of
his 'buds'.
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Eric,
As a neutral voice in terms of speed claims, and someone familiar with the
raptor (I built the Screamer prototype unvailed at SigGraph) I would like to
make the following points on speed:
The speed increase seen will depend HEAVILY on the scene content.
The raptor can be slower than the amiga, due to fixed overhead, but only for
very easy scenes. For example, a scene that renders in 20 seconds on an Amiga
might render in 1 or 2 seconds on a raptor, but include 60 seconds or more of
file transfer.
The processers doing the rendering are fundamentaly different (040 vs r4400).
This is IMPORTANT to speed claims. The big advantage of the 4400 (other than
generally being faster) is its pipelined FPU. The 040 as you may recall has a
limited FPU and requires software libraries to do most of the higher functions
(sin, tan, etc..) while the R4400 does all this in hardware. This means that
for certian types of images that are FPU intensive (it also helps if most of
the rendering code fits in the cache of the 4400).
I did most of my work with these things using two processors of the older (and
slower) deskstation machines. Results varied from perceptually quite a bit
faster to way,way,way faster, as long as the scene was sufficiently complex
(i.e. more than a few min rendering time). Based on my experience, the 10x
stock amiga 4000-040 is the 'typical' number. Very few things will go any
slower than this at all, and certian classes of scenes will go MUCH faster.
All of this indicates it is very difficult to determine reletive speed between
architecurally different machines, even if they are running the same
application.
To wrap up, from everyone I know doing 3D graphics for $, the Raptor is a very
effective, revenue enhancing tool. You certianly don't have to buy one (it is
a free country). Deskstation, however, does seem to make the fastest R4400
based NT machines around, and as I understand, does a good job of installing
the ScreamerNet software and making things work out of the box (not a trivial
task with an NT machine, as I know).
Charles
Charles:
Thanks much for the even-handed and lucid explanation of the variation in
rendering times. I had expected things like this to crop up, given the
'minor' differences between the 040 & R4400, and I'm _VERY_ curious to
know "just how fast it'll go." Like Tony D. Tracer says "Any rendering
time >0 needs acceleration." :^)
I remember running into performance problems because of the limited FPU in
the 040, and am not surprised that the R4400 can do transcendentals mucho
faster. You also mention the cache size of the 4400, again another
important performance booster.
Now, if this was the sort of argument I had received initially – as
opposed to a highly emotional, personal attack reminiscent of being on a
playground – things would have ended up much more amicably.
As you and others have pointed out, the Raptor is the 'only thing out
there' right now, and if a time should come when I truly need the speed,
I'll put on my black trench coat and go out and buy one under a pseudonym.
<g> Until then, I'll keep on asking questions.
Thanks again. This was the exact sort of realworld/technical explanation I
was looking for.
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Have noticed the "any render time >0 needs acceleration" before. This is a
rather impossible goal. If you said however, "any render time > effective
frame rate needs acceleration" you might eventually get your wish.
Real time rendering/ray tracing video resolution images in real time (1/60 sec
to calculate 4.2 MHz video screen) is doable today with alternative archetecure
processers (i.e. wierd supercomputer that is hard to program, but kicks butt on
vectorized ray trace calculations). The hardware for such a machine could even
be rather inexpensive (like < 50K) in low production quantities. The problem
is amortizing the design cost of such a beast if you wern't going to sell as
many as something like a Pentium.
Anyway, now that you're all excited, sit back down, because even 'tho it's
possible today, you probably won't see anything neat for 5 to 10 years, as the
market gradually gets big enough to become 'interesting' to those with enough
$$ (and know how) to make such things.
In the short term remember you've already got the power of a couple of early
Cray's lying around. Now if only you could get your hands on some of the
processing power they have online at the NSA……
The comment about 'rendering time >0' needing 'acceleration' is a
signature tag of one of the guys who frequents this forum. While your
reinterpretation of the addage to say '> effective frame rate' is _indeed_
more accurate, I believe he is speaking for those of us that _still_
wouldn't be satisfied……Currently, the frame rate of VR rides is upto
60-80 fps using film. We wouldn't need 'motion blur' as an option if the
frame rate was upto 100fps….:^)
And once the question of frame rate is licked, I for one would like to see
them tackle the problem of producing a truly 'holographic' display not
requiring glasses. Then there's the question of tactile feedback,
holographic audio, temperature, and smell….
Anyway, suffice to say that I won't get 'all excited' if you tell me
'gee-whiz' stuff like I've "got the power of a couple of early Cray's
lying around."
I'm aware of that, and now (thanks to modern culture), I WANT MORE!!
<G>
Anyway, I'm looking forward to hearing what Deskstation has to say at
SIGGRAPH. I wish them the best of luck.
Thanks again for supplying us with some information about the Raptor..
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Eric,
All the talk about high speed VR systems reminds me about the magazine
article they had a few years back of a prototype AR headset. The specs
included: stereoscopic 24 bit color with realtime response, automatic
motion blur, automatic collision detection, object interaction, gravity,
wind, atmospheric effects (fog, rain snow), radiosity type lighting,
better than CD quality sound, and much, much more.
One of the big drawbacks was that the editor was hard to use.
Of course, the AR headset didn't have much built in processing power since
it's only job was to look cool. By the way, AR stands for Actual Reality.
;^) (over 5 billion users daily)
Jeff H…
:^)
Aww man, REAL life sucks…Give me a VIRTUAL life where I can be a GOD!!!
(Now, that should be good for a few trips to the psychiatrist. )>:
AR _is_ cheaper, but it's the user interface that takes some getting used
to…..(Where's the FIRE button? :^)
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Erik,
But you ARE God – you just don't KNOW it yet. :^}
wmc – via Autopilot!
Make you a deal.
You provide the direct link to the cerebral cortex, I'll whip up the machine to
pipe in the data.
In the real world 🙁 I know deskstation has been speeding up their machines,
using the newer/faster/lower voltage MIPS chips. I don't know what their
latest/greatest is, but I would expect to see improvements out of the
MIPS/deskstation camp closely pacing events in mainstream x86 land (i.e. apx 2x
the power each 18 months).
My previous message was intended to forshadow events unfolding in the next five
years or so, as the current systems hit the limit of improvement in terms of
clock rate/bit width/piplining and radically different approaches are used to
improve speed. From what I hear the P7 project is a drastic departure from
'conventional' processers, while being code compatible. The huge leaps being
made in hard drives this year and next (soon there will be no drive < 1GB or
slower than 4 MB/s) will hit processors in 3 to 5 years. In the meantime a
Raptor is I believe the most cost effective rendering machine ever available to
the general public (or anyone for that matter).
In terms of Gee whiz stuff, Some of the best stuff I have seen recently
is the tracking of human models, wired to Onyx, that allows for real time
caricature movement of a Shaded 3-D Object. Soon, imagine rendering
complex 3-D creatures and producing naturalistic motions by allowing a
dancer/actor to input movement in real time. Prehaps the son of the
Raptor, (the Brontosauraus), will allow realtime throughput for a variety
of uses…simulations, games, Special Effects. What better input for
motion can you have than the real thing? Great article in New Media this
month…"Hollywood Production Digital Style" Put the biz in great
perspective. Amiga/Toaster touted as the Can Do machine. GREAT Pic of
Conti's new series Hercules with an Aqua-Nymph. The concepts about real
time motion capture are explained well,too.
Joe Clasen
Joe's DB&G
Yeah, some of the real-time rendering stuff on those high end machines is
pretty awesome. The ability to use digitized motion paths from live actors
makes a world of difference when trying to do character animation.
One company that offers digitizing services would even offer output in LW
format – for only $7000/day, with a 2 day minimum!
Did you ever see the 'rapping' velociraptor animation someone showed at a
japanese graphics show? It was pretty cool! (I'll have to read that issue
of NewMedia…so many things, so little time. :^)
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
At Universal Studio in L.A., next to the Studio where Seaquest is taped,
ran a Jurrasic Park display. One display was a Dinosaur dancing in real
time from input from a volunteer. Probably the same system. I particlarly
like the way the jumps were tracked. Sythetic actors are on the verge of
becoming a reality. ( This doesn't include the Captain of Babylon 5.)
Joe Clasen
Joe's DB&G
What you are mentioning sounds alot like the WALDO machine for giving 3D
"Puppets" life
Simgraphics VACtor system produces a Waldo. Sony Pictures Imageworks has
one too. SuperFluo is another. Or you can fake it with the Toaster for a
tenth the cost. Depends on how many 0's are in the budget.
Joe Clasen
Joe's DB&G
😉 I like that. "Depends How many 0's in the Budget"
Tho I'm unsure how ya can simulate a WALDO with Lightwave.
What're ya using for feedback to the System? The old Mattel Power Glove?
There are elves tinkering away to do just that, but with bigger sights
than just a glove. FYI, Newtek and company are going to the L.A. Toaster
user group this Sat. in L.A. Lee and others should give a full 2 hour demo
of the Flyer and might have a few answers to the burning answers of our
time. Like when, in what housing, whose hard drive, and who will provide
the SMPTE interface? If they spill the beans, I will tell ya. They said
they are going to tape the meeting for a video to be distributed to the
dealer Network across the nation.
Joe Clasen
Joe's DB&G
But right now, NOTHING looks like it beats the ACCLAM motion capture system.
This setup is all optical….no cords, wires, etc. Just get out there, and
dance around!
GM
Hey Glen!
How much is the ACCLAM motion capture system? I could use something like that.
I'm thinking of building a motion tracking system.
The leg's almost healed!
Best!
–>Doyle Calvert Synergistic MultiMedia
Out of curiosity (and trivia)..
Do you know where the term WALDO comes from?
JG
John,
I'm not sure where the original name comes from, but waldos are what they
call those mechanical arms that were originally designed for handling
nuclear materials and other hazardous goods. The term's been around since
the late 50's.
Jeff H…
They come from a Robert Heinlein story/book called Waldo and Magic, Inc
about this large guy named Waldo that used such devices.
JG
I believe it was coined by Robert Heinlein in one of his stories way back
in the 40s/50s.
I vaguely remember they were either developed to aid a handicapped
protagonist, or to make very small things……
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
I'm uncertain where the Term WALDO originated John. Sorry.
John,
Doesn't Waldo mean a useful device that took forever to develop and wears a red
striped shirt? :}
I don't know! Call them there folks at SIMGRAPHICS and ask 'em!
GM
Heinlein — altho I forget the name of the book (wasn't it a short story?)
-sja
"Waldo and Magic, Inc", was the short story if memory serves. I don't
remember the plot exactly. Something to do with prosthetic limbs, I
think.
–
Jim
>Thanks much for the even-handed and lucid explanation of the variation in
>rendering times. … I'm _VERY_ curious to know "just how fast it'll go."
>Like Tony D. Tracer says "Any rendering time >0 needs acceleration." :^)
Erik, machines will get so fast that they will finish BEFORE you start the
project. :^)
I will say that my friend was testing the Raptor the local dealer had and
there was an acceleration of about 7-11 times for the normal production
type rendering. But when he was doing a Ray-Trace anim of a Diamond and
rendering times were in the 6-8 hours a frame on a GVP 33/040 but took
about 20 mins on the Raptor. So the claims for speed are very dependent on
scene content and hopefully this will be at least hinted to in the
broucher.
I wish I could afford a Raptor the only chance I may have is to lease one,
I'll have to look into that.
<<< Tony D. Tracer >>>
(Any render time >0 needs acceleration.)
Remember Tony, if you make instant Coffee in the Microwave there is a 30%
chance you will go back in time due to the extreme quickness the coffee is
made.
'…they will finish BEFORE you start the project.'
Those computers would be using the new ST:TNG FTL (Faster Than Light)
circuits, right? :^)
I'm not surprised that certain types of scenes will render faster than
others, given the difference in CPU architectures between the Amiga &
Raptor.
I'll be very curious to see what this rumoured Raptor 2 is all about. If
it's the Workstation _I've_ been waiting for, DeskStation may well get
some of my money. (Until then, I'm going to be upgrading my 3000 with a
40MHz 040. Know anyone looking to buy a slightly used GVP 28MHz 040 w/2MB
RAM for the Amiga 3000? :^)
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
oh oh pick me pick me……Wish I COULD buy yer 040….Saving for a
Engagement Ring/Wedding…..My Spending will be naught for some time now
methinks
For _you_, I might consider installment payments. (Sort of my little bit
of charity. :^)
It would be _ONLY_ around $650.00………
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Erik,
What sort of accelerator are you selling? I've been thinking of upgrading
my 2000. The chances of a Raptor in the near future is starting to look a
little dim.
Jeff H…
Alas, I'll be selling a GVP A3000-040/28 w/2MB-40ns RAM. It only works in
the CPU slot on a 3000/3000T. Sorry. :^(
Have you seen the rumours of the 'Raptor 2'? I have no idea what it is,
but I'm hoping maybe it'll be the single/double R4600 workstation version?
Keep On Toastin'
Erik Flom – ELF Works 3D Construction Co.
Erik,
Just my luck about the accelerator. :^(
Haven't heard anything about the new version of the Raptor. Gee, wouldn't
it be cool if the market was flooded with original Raptors dirt cheap?
After all, they WOULD be obsolete right? ;^)
Jeff H…
heheheheheheh Still couldn't afford it unless those installments were
about $25 a week. Thanx for the offer tho Erik.
Joel,
The Sparks interface for version 1.0 is not very "Lightwavish" and may take a
little getting used to.
John Tindall, who is the author of the Sparks is just about to release the
latest and greatest version of this particle animation program. I have not
seen the new interface, but I may see him next Thursday and hopefully I will
have some answers for you with regard to the interface. You can call him
directly yourself, he is a great guy and will answer any questions you have.
contact John Tindall at:
MetroGrafx
625 Newton Drive
Lake Orion, Michigan 48362
Phone: (810) 693-5134
Also, Erik is correct, MacroSystemsUS is no longer supporting Sparks, so I
recommend that you get it directly through John. It has new tutorials,
manuals, and a whole host of special effects not available on any platform –
including SGI. In fact, John's "real" job is doing animation on high-end SGI
machines.
Adib
Adib, I couldn't get my order canceled from MacroSys and I got Sparks today.
What rotten luck.<G>
jmb
Well, its not that bad! You can still upgrade to the new version through John
Tindall.
Don't be put off by the manual you have received with the MacroSystemsUS
version. The upgrade has a very intuitive manual and examples which will
provide you with a clear understanding of how Sparks works. The manual you
have has three examples which really are not that clear to some people, the new
manual will have at least seven and an extra disk which has sample scenes for
you to load and try out.
In fact, one of the reasons the relationship with MacroSystemsUS was
terminated, was because of the lousy job they did on the manual, no upgrade
support (John must have sent them at least 2 new versions which they did not
provide to users), no technical support, poor packaging..etc, etc.
Adib
Haven't had a chance to get knee deep in it yet but I will agree on the
packaging. Looked like generic fruit loops box (B&W).
jmb
I'll keep you informed on any new developments.