CompuServe Thread

#3D CHALLENGE!

85 messages in this thread
#58716From: Suresh KavanJun 12, 1993 2:13 PM
OK .. many of us have 3D Studio, Lightwave, Imagine, Aladdin and other 3D programs..why don't we build a scene and render it on each platform and compare rendering speed and output quality..and really figure out which product is the winner! With Interchange Plus..we can covert the objects pretty easily .. Anyone got any good ideas for a test ?? Mebbe a room with furniture, lights, books etc.. ?? If we've really got nerve, let's post the challenge in the 3D Studio, MAC and other relevant areas….. Lemme know what you think…unless your chicken that is!
#58729From: Jim S. RoofJun 12, 1993 10:51 PM
I will volunteer for the test. I am running a 2000 with a GVP 33mhz '040. I have the 2.0 release of Lightwave as well as an early beta copy of the 3.0 release. It's basically the same as 2.0 but redirected for the '040. Let me know, I am up to the challenge.
#58760From: Suresh KavanJun 13, 1993 7:39 PM
OK .. cool, what shall we use as the the 3D model….anything in mind..?? Where should we post the files…??? How should we judge the winner..??? SK.
#58781From: Jim S. RoofJun 14, 1993 8:12 AM
All good questions. I think we should choose something of average complexity, say, no more than 20,000 or 30,000 polys. This would allow for more people to get involved and would also keep the download time reasonable. As far as subject matter, hmmmm. I'll think about it and get back to you. I guess the files could be put in Lib 10 of the Amiga Arts forum or possibly here in the Newtek area. Let's get it on! (typed with a New York accent, a la boxing announcer) Jim Roof
#58752From: Erik FlomJun 13, 1993 3:54 PM
It seems like a good test would be one of the old standard images that has already been made on a number of platforms. (2 that come to mind are the Hurricane Lamp used for the Mac render-off a few years ago, and the Blurred Bowling Pins from the Renderman cover.) Sounds good to me! I'm really interested in finding out how the packages compare. I don't care about what platform it's on, I want SPEED! :^) Keep On Toastin' Erik Flom P.S. I read your initial statement, and would agree with most of it. (The point about the which versions of the renderers were being used is a valid one, IMHO. LW3.0 is MUCHO faster, even if it doesn't have think like the IPAS modelers.)
#58761From: Suresh KavanJun 13, 1993 7:44 PM
Thx Eric. Now, you've been busy lately with 'museum' pieces (very impressive..) so, I'm wondering, what have you got that we can convert to 3DS format etc.. Unless these models are highly proprietary, it'd be a good place to start – multiple relections/lights/object mapping types/surfaces etc… Whaddya think. I have Interchange Plus and the 3DS add-on so if nesceccary could be the official converter! Mebbe Foust will jump in and donate a scene from his upcoming 3D object library??
#58766From: Erik FlomJun 13, 1993 10:13 PM
I could probably make parts of the museum public – though folks that have paid me already might not be too crazy. :^o I'll be getting Interchange+ in the mail on Monday, so I'll be in a better position to work on this. We may want to consider setting up one or two scenes which will showcase both Raytracing options, and std Phong/Goraud shading renderers? (Since some packages are optimized for tracing/non-tracing, it'd only be fair.) (Also, alot of my models rely heavily on lots of 24bit bitmaps – we'd have to convert those too.) I like the idea of using something from the 3D CD-ROM. Since those would likely use mostly procedural surfaces, the resulting 'test' scene(s) would be smaller (for distribution). Keep in touch. I'd like to do anything I can to help. (This sort of comparison has been done on single platforms, but it'd be nice to have a 'suite' of test scene which we could use across platforms as a benchmarking utility.) Keep On Toastin' Erik Flom
#58850From: Stan Kalisher (Impulse)Jun 15, 1993 2:26 PM
Erik, unfortunately, LW 3.0 is significantly slower than 3D Studio and slightly slower than Imagine. In the past few months, I have been heavily involved in the creation of a new product for all three platforms called the VRS Media Master Designers' Objects Series. Each scene we rendered is quite large (at least 8 Megs) and requires reflection, shadows, etc. A scene which might require several HOURS in Lightwave (with raytracing) could be produced in UNDER 5 MINUTES on 3D Studio. Imagine tracing came in at about 45 minutes. If you use shadow mapping and reflection maps (which is all 3D Studio does, anyhow), Lightwave is still significantly slower, coming in at somewhere around 15 minutes. If you'd like, I can prepare a more detailed posting of our findings. The machines we used were an A2000 with 28 Mhz 040 and 18 megs of RAM and a 486/33 with 16 megs.
#58854From: Vance SchowalterJun 15, 1993 3:42 PM
Someone said that 3D Studio's current version wasn't using real raytracing. Scanline with simulated shadows and other things would be in most cases significantly faster than true raytracing.
#58913From: Stan Kalisher (Impulse)Jun 16, 1993 9:02 AM
That's right, 3D STudio doesn't raytrace at all. But even when using shadow mapping and reflection mapping in LW 3.0, 3DS still is at least three or four times faster. –Rick Rodriguez, VRS Media
#58922From: Vance SchowalterJun 16, 1993 11:19 AM
Ok. Just wanted to clarify that. *8) -= Vance =- Image Master using Black Belt Systems' Imagemaster!
#59010From: Jim S. RoofJun 17, 1993 1:12 PM
You don't mention anything about how much oversampling and at what threshold you had LW 3.0 set for. You also do not mention the resolution of the image or the degree of antialiasing. These attributes need to be included in your comments or there is no true comparison. I once passed a Ferarri Testerossa in my VW Rabbit. Jim Roof
#59014From: Stan Kalisher (Impulse)Jun 17, 1993 2:34 PM
Jim, the images were rendered at 768×482. Antialiasing was set to medium and the threshold and sampling were left at their defaults, which I have been told is too much. Imagine edge level was set to zero. 3DS' antialiasing was set to max. –Rick
#58867From: Peter GreciJun 15, 1993 4:38 PM
Hiya Rick, any news on Imagine 3.0? I know this is getting redundant but us Imagineer's grow impatient after seeing Lightwave 3D 3.0 wether it's slower than Imagine or not.
#58914From: Stan Kalisher (Impulse)Jun 16, 1993 9:12 AM
Peter, from what I've seen so far, I would say that LW 3.0 is certainly faster than previous incarnations. However, I think that much of the talk surrounding its incredible speed may be attributed to hype. Likewise, Imagine has been given a bum rap for being exceedingly slow. Which it certainly isn't. Scanline rendering for both programs is about the same speed and Imagine doesn't add the kind of potential penalties for maximum antialiasing as LW does ( although I grant you that the maximum antialiasing in the new Lightwave is significantly superior to Imagine's maximum antialiasing). When raytracing is thrown into the conversation, my experience shows that Imagine can be up to four times faster than LW 3.0 on an '040 28 Mhz. For example, the bathroom scene contained in Volume 1 of the VRS Media Master Designers' Objects Series took 40 minutes to raytrace in Imagine. It took nearly four hours in Lightwave 3.0. On a 486/33, the Imagine raytracing took 15 minutes!!! Generally, 3D STudio is far faster than either Imagine PC/Amiga or Lightwave. However, for this scene, I turned on automatic reflection mapping, which meant that the program had to generate new images for every reflective surface in the scene. This bogged down the program and I ended up with a 30 minute rendering.
#58940From: Peter GreciJun 16, 1993 5:11 PM
Thanx for the info Rick.
#58962From: Charles HillJun 16, 1993 7:52 PM
This is interesting. The beta version of LW 3.0 (to the best of my knowledge, it hasn't ben publically released yet) works only on an A4000….that is, it requires AGA. Hmmm…wait a minute…that may be the TOASTER software 3.0 that requires AGA…I've access to both. I have tested a number of "real world" scenes on LW 2.x beta (real world meaning they have been commissioned and paid for) and NONE of the images has taken over an hour to render. Most (without raytracing, but with shadow mapping) render in between 2 and 4 mintues. All my images are done in medium res with either low or medium anti-aliasing. With raytracing, the time depends on the number of reflections/refractions but most of mine have averaged under 10 minutes (for scenes >6 Mb). Granted, if I do a scene of chrome balls in a mirror room light by a couple of candles, etc. then things slow down. But…that isn't real world. I wanna participate in this test. I've an A4000 (25 MHz 68040) with 18 Mb of RAM and the Toaster 4000 (among other things…but they aren't relevant). I've a beta version of LW 3.0. Oh, yes. I also have a normal Toaster (2.0) running on a 25 MHz 68040 based A3000 and LW 2.0. The beta version of LW3.0 does not run on this system. What revision of LW are you running? The version I run will (as long as it isn't raytracing) match or beat 3D Studio on my 33 MHz 80486DX in both speed an quality. I haven't seen Imagine since 2.0. Somebody pick a scene an upload it either here or in AmigaArts 3D library.
#59044From: Trent K. JohnsonJun 18, 1993 2:44 AM
Rick, did you happen to mention when your Object Disk series will be available? If it is already, how can I get catalog/order info?
#59058From: Stan Kalisher (Impulse)Jun 18, 1993 10:30 AM
Trent, volumes 1 and 2 are available for Lightwave, Imagine and 3D Studio. You can see two images from volume 1 in Lib 4. Just browse VRS. Email me your name, address, phone and fax number and I'll send you something ASAP. –Rick Rodriguez, VRS Media
#58898From: Erik FlomJun 16, 1993 1:25 AM
This sort of information sounds very useful! Please upload a text file to the the appropriate area. (I'm not sure where, but maybe in the AmigaArts RayTracing group?) I have to admit that I AM sorry to hear your results. Is there a chance you could suggest (or create) some good 'test' scenes that would work on all platforms? Thanks for the input, Keep On Toastin' Erik Flom
#58915From: Stan Kalisher (Impulse)Jun 16, 1993 9:18 AM
Erik, prior to commencing the Master Designers' Objects Series, I designed a western scene for a new videotape we're developing called "3D Shootout." This video takes on this precise question head on by putting each program through its paces both for modeling and rendering. The scene is included on Syndesis' 3D ROM. Once again, our findings were that 3D STudio rendered the scene with mapped shadows and reflections in about 2.5 minutes. Lightwave took about 15 minutes to map shadows and reflections and Imagine took about 20 minutes to raytrace. –Rick Rodriguez, VRS Media
#58965From: Erik FlomJun 16, 1993 9:05 PM
Sounds like I need to talk to John Foust, and see if he can supply us with the 'shoot-out' scene as a sampler from his disk. Would you have any objections to that?
#58997From: Stan Kalisher (Impulse)Jun 17, 1993 9:44 AM
None whatsoever. Since I've laid down the gauntlet, I suppose I should sit back and see what results come back from independent sources. –Rick
#59007From: Erik FlomJun 17, 1993 1:04 PM
Sounds good! I'll be getting my new system in the next few weeks, so I'll be able to test it on both version of LW. Thanks, Erik Flom
#58923From: Vance SchowalterJun 16, 1993 11:19 AM
Well, I think it's all a matter of programming technique. I think that 3D Studio's approach must have been unique to garner such speed increases. It certainly isn't the 486. I'd like to see these speed comparisons sent to NewTek so they'll work on a solution to get Lightwave up to the speed it is capable of. I'm definitely getting Imagine 3.0 when it comes out. I don't have the energy to adjust to another renderer outside of Lightwave and Imagine. So, I'll leave it up to the rest of you to push the envelopes of the other Amiga renderers.
#58966From: Erik FlomJun 16, 1993 9:05 PM
I WILL be curious to perform some of these tests myself. Since I just found out 3D Studio was used for the 7th Guest CD-ROM game, I'm beginning to wonder if I did the right thing by just buying another $4K of C= crap. :^(
#58988From: John GrossJun 17, 1993 1:54 AM
I'd be interested to see what the time for LW is if you used a 33 mhz 040 which should be more comparable to the 486/33. Also, are you using Med res images in LW or High Res…and what level of antialiasing. My experience has shown that even with ray tracing on in LW, it's difficult to get frames to render in over an hour unless there are over 100 lights.
#58998From: Stan Kalisher (Impulse)Jun 17, 1993 9:48 AM
John, the image I traced can be found in the Imagine lib as bathroom.jpg. The image posted there is the Imagine version, which required an hour to trace. The Lightwave version took several hours. This result was confirmed by former VTU editor David Duberman. The same scene takes 45 minutes in 3D STudio. –Rick Rodriguez
#59116From: John GrossJun 19, 1993 8:10 PM
was that done with 2.0 or 3.0 LW? also, what resolutions were used?
#58793From: James RoathJun 14, 1993 1:45 PM
Ok, I'm in, but I'll raise the anty… One of the SGI dead heads around here has offered to render the scene on his workstation, during lunch. I have seen how it renders and it would be interesting to see how it works out. Some of the scenes they have reproduced from 3d studio files have render in 1/60th the time. In addition, I can setup a 486 50 mhz with 3d studio, and a 33mhz 040 Amiga. Of course there will be some advantages to each of the platforms that should be taken into account. The 486 50 has only got 32 megs of ram, the 040 Amiga has 72 megs of ram, and the SGI has some ungodly amount of ram in it… Also, how will we handicap the 3d studio images in regards to ray tracing ? Oooppssss…. I can't use lightwave in this test but can use Imagine 2.0 ….
#58818From: Suresh KavanJun 14, 1993 10:09 PM
Well guys, this wild!!..we've got 3DS, Imageine, Lightwave and SGI… so the question remains whats the scene that we'll render. Taking suggestions given, I reckon that two scenes is definetly the way to go : (1) Raytracing ON (s/w that doesn't do it, will just have to do it's best!).. (2) Phong only. Who knows Foust well enuf to persuade him to donate a scene or two from his forthcoming CD-ROM title?? Lastly, I'll start to build a library of the Hardware we're gonna put on this project….and I'll post the images/scenes to the 3DS studio when we're ready. This'll be fun. SK.
#58848From: Stan Kalisher (Impulse)Jun 15, 1993 2:18 PM
Suresh, I undertook this exercise in the last few months as I began developing my new object series, the VRS Media Master Designers' Objects. If you'd like, I'd be glad to share my findings with you regarding 3D Studio, Imagine (Amiga and PC) and Lightwave. –Rick Rodriguez, VRS Media
#58852From: Suresh KavanJun 15, 1993 2:50 PM
I'd absolutely like! Just one thing…aren't you a little biased <gr>. SK.
#58916From: Stan Kalisher (Impulse)Jun 16, 1993 9:22 AM
My association with Impulse is based primarily on friendship. I am not now, nor have I ever been employed by the company. I've used the Toaster since day one and have known Alan Hastings since way back in '86 or '87 with Videoscape 1.0. I began using 3D Studio about a year ago. If I'm biased, I'm biased toward speed. I've always used these products inn a production environment where deadlines are everything. It's unfortunate that there is no room for unbiased journalism in the 3D arena. There are too many vested interests clouding writers' perceptions of the products they are covering.
#58873From: James RoathJun 15, 1993 5:28 PM
I'd be interested in seeing those numbers as well! And why on earth does the pc version of imagine render faster than the amiga version ??? 486 50 and 040 33 were the test machines… Is imagine 2.0 not optimized for 040 operation ? I assume that 3.0 will be ???
#58917From: Stan Kalisher (Impulse)Jun 16, 1993 9:32 AM
Imagine 2.0 has been optimized for the '040 for over a year now! My test machines are a '040/28 Mhz with 18 megs of RAM and a 486/33 with 16 megs of RAM. From what I understand (and I'm not a techie, so forgive me if I'm mistaken) the PC's architecture allows a single program to run more efficiently. Even though you may not be multitasking on your Amiga, the computer is always polling every device to make sure you haven't inserted a disk in a drive or performed another task which requires attention. Thus the Amiga's strength becomes a weakness when it comes to pure performance. Reading the above, I know I'm in for a lambasting from a lot of you! Please, if I'm mistaken set me straight. However, before sending off a flame, remem- ber that I am working with real world objects in real world products. Imagine PC is up to three times faster than Imagine for the Amiga when raytracing the same scene. 3D STudio is much faster than anything on the Amiga. Go figure. –Rick Rodriguez, VRS Media
#58924From: Vance SchowalterJun 16, 1993 11:19 AM
Well, couldn't Impulse devise some sort of patch for Imagine to gain similar speeds (ie. deactivate polling devices)? Or would that cause more trouble than it's worth? Maybe we should just wait for the next generation of accellerators? -= Vance =- Image Master using Black Belt Systems' Imagemaster!
#58934From: Stan Kalisher (Impulse)Jun 16, 1993 4:06 PM
Like I said before, I'm not a techie except when it comes to 3D and video. I tried to play around with task priorities on the Amiga to see whether or not I could simulate the "single-mindedness" of the PC. I ran a pd program that listed all the tasks going on at once and I was shocked to see so many of them! Anyway, task priorities are very delicate if you raise or lower a priority too much. I tweaked the system a bit and got the refresh to work a bit faster, but rendering ended up being the same.
#58978From: Michael C. NielsenJun 17, 1993 12:24 AM
Rick, I can corroborate your findings on 3D Studio vs Imagine. I run 3DS on a 486-33 vlb and I have Imagine on an Amiga 3000 25. Both machines have 16 megs of ram (and of course the Amiga has an additional 2 megs on Agnus while the 486-33 has one meg on the vlb video card. My 3DS not only renders faster, but also displays 24-bit .TGA images very quickly. I would never give up 3DS, esp since the manual and tutorial were so nicely done that even a dummy like me could learn 3D imaging <g>. BTW, you are not the guy who wrote that manual for Imagine 2.0 are you?
#58979From: Vance SchowalterJun 17, 1993 12:32 AM
A better comparison would have been if you had a 68040 instead of a 68030. *8) -= Vance =- Image Master using Black Belt Systems' Imagemaster!
#58999From: Stan Kalisher (Impulse)Jun 17, 1993 9:49 AM
No, I wrote the documentation for versions .9 and 1.0. –Rick
#59095From: Michael C. NielsenJun 19, 1993 8:10 AM
>No, I wrote the documentation for versions .9 and 1.0. –Rick Then we can speak freely <g>. That 2.0 manual is quite an interesting book. But it is not a manual for 2.0. I did get Steve Worley's book and I am now actually beginning to understand the program. Since you use both 3D Studio and Imagine, I have a question for you. Since we agree that 3D Studio is faster, when would you have a need to use Imagine? I have heard people complain that 3D Studio is not a ray tracer, but when I look at the results of mapping, I say, "Who cares?" What do you think?
#59098From: Vance SchowalterJun 19, 1993 10:25 AM
Well, unless newer techniques are found that will render raytrace-accurate shadows and refractions, etc., raytracing is almost always preferable for complex/high quality images.
#59192From: Stan Kalisher (Impulse)Jun 21, 1993 9:37 AM
I think Imagine is a superior object creation tool. Perhaps it's the years of familiarity with Imagine, but I also think it's easier to animate within Imagine than 3DS. For the most part, I would agree that I prefer to render in 3DS; however, there are some things that annoy me. First off, 3DS apparently can't reflect map onto a flat surface. Am I doing something wrong? Secondly, in a highly reflective scene like the bathroom (which has a mirror, reflective tile walls and floors and reflective porcelain), rendering reflection maps can actually take LONGER than simply raytracing the scene in Imagine. To sum it up, I use Imagine, 3D Studio and Lightwave. Each has its strengths. Each has its flaws. I'm also trying to figure out how to swing purchasing a Softimage system. So it goes… –Rick Rodriguez
#59038From: Joe StratmannJun 17, 1993 11:24 PM
I'm looking at 3D studeio for the IBM and I saw your message on the Newtek forum. I have a 486/66 Gateway system with an ATI Ultra Pro video card – it's capable of 24bit display. Are you using a Targa board or another video card similar to mine? I'm curious because I keep hearing conflicting information from different people.
#59061From: Stan Kalisher (Impulse)Jun 18, 1993 10:36 AM
Joe, I'm using both a 24bit graphics card and the Targa. If you need to go to videotape, you need the latter. –Rick Rodriguez
#59096From: Michael C. NielsenJun 19, 1993 8:10 AM
Joe, I am not sure what your question is, but here goes… I am porting my .TGA files generated in 3D Studio over to the Amiga and converting them to 24 bit DCTV anims. My video card for the 486 is a Cirrus Logic Vesa Local Bus card that cost me $120. It displays 640 by 480 at 24 bits. I had thought at the time that I got it (and the upgraded VLB motherboard) that I would use a program called F16 to compile 16 bit (64k colors) animations that I would play in real time through some sort of VGA to NTSC box. But after a guy sold me his Amiga 3000 for $800 ($400 in cash and the rest in used video equipment that I had no use for), I changed direction. With your setup, if all you want to do is dump 256 color animations to tape, you just need an external VGA to NTSC converter. If on the other hand you want to do higher color anims such as 15 or 16 bit, I would look into the F16 program. You can look at a demo by downloading it from their BBS at (602) 956-5959. You will find that even though your ATI Ultra Pro is fast, the ISA bus throughput of your Gateway system will be too slow for more complex F16 files with large delta (change between frames) values. The Targa card is nice if yoiu have a Single Frame Capable VCR such as the Panasonic AG-7750. But I am not sure how it would help if you were trying to do what I do on the Amiga, which is recording to a Panasonic AG1960 in real time. So, for my money (and I got an incredible deal on 3D Studio), the best is to render with 3D Studio and transfer to tape on the Amiga via DCTV.
#59148From: Peter GreciJun 20, 1993 10:09 AM
Hiya Mike, I'm glad the DCTV is working out for you. Haven't seen ya much on FIDONET lately tho.
#59190From: Michael C. NielsenJun 21, 1993 6:33 AM
We had a very immature person at the hub of our fido feed and when he quit, he simply "unplugged" the rest of us from the system. No fido feeds for over a month now. :-< Someone commented on his abrasive style and he just dumped all of us into the trash. I have been capturing the video and amiga-video echos from Ron's bbs in Michigan. I got his Toaster, btw. I have not got it up and running yet because of problems with installation and bad disks. I am rebuilding my a3000 ("We can make him bigger, stronger…" -The 6 million dollar man) with a 330 meg Fuji drive, a 200 watt power supply, the new Personal Component Adapter (from DPS for S-video input and output to the Toaster) and the Toaster. I already have one TBC and once I have the other one, I have a complete S-Video A/B roll editing system. Now, if I only had some clients…
#59099From: Peter GreciJun 19, 1993 11:11 AM
Mike uses both a IBM/Clone and an Amiga, I believe he uses a DCTV on the Amiga to output to Video tape.
#59309From: Laurie CarrollJun 23, 1993 3:38 AM
As the Amiga is multitasking while doing rendering,then to "equalise" these tests, shouldnt the PC be doing some other real world task. Like running another programme. EG getting info from CIS while rendering, which would seem to me to be a real world like activity for us. 🙂
#59320From: Stan Kalisher (Impulse)Jun 23, 1993 8:13 AM
Unfortunately, DOS doesn't multitask and neither Imagine PC nor 3D Studio may be launched from Windows. Also, I don't multitask with my Amiga when rendering, either. My point was that whether you like it or not, the Amiga's always off doing something while the user thinks it is performing only one task. –Rick Rodriguez
#59367From: Jim S. RoofJun 24, 1993 7:06 AM
So what it sounds like is this – You're saying "Unfortuneately, neither the PC nor the Mac multitask. Unfortuneately, the Amiga does." That's weird.
#59373From: Stan Kalisher (Impulse)Jun 24, 1993 8:48 AM
Kinda sucks, doesn't it? Fortunately, any SGI box multitasks and doesn't penalize you for it. Unfortunately, the minimum real investment is $20-$30,000 minus software. –Rick
#59444From: John DowdellJun 25, 1993 3:54 PM
Hi Rick, When you say "multitask", what exactly d'ya mean? I've heard the term "preemptive multitasking" used before, specifically in contrast to what our desktop machines do, but am not exactly sure what it signifies…is this what you're talking about? From what I understand, DOS can only run one app at a time. But Windows allows scheduling of multiple apps at once, even DOS apps, and these can be prioritized by the user. How does this differ from the Amiga's setup? And can't you turn off if you want to? Thanks, John D.
#59446From: Stan Kalisher (Impulse)Jun 25, 1993 5:40 PM
To backtrack a little on this thread…We were lamenting that Amigas don't seem to have the throughput that 486-based machines have. Impulse has always told me that the PC Imagine would be faster than the Amiga version, but I didn't believe it until I verified it myself. Rendering can be as much as three times faster on the PC version! One of the reasons for the improved throughput is that the PC does not have to contend with the constant overhead required by true multitasking. Windows' flavor of multitasking is more akin to task switching. For example, the Amiga is always polling its disk drives to see whether you have popped in a new disk. When a machine is on a network, this can be a serious drain on resources. The PC, on the other hand, does only one thing at a time. So you have to tell it when you've changed a disk. Screen refreshes are very fast on the PC, but the mouse is momentarily disabled as the screen updates. I tried to play with task priorities on the Amiga to achieve faster screen refreshes, but there simply is too much going on at one time to make much difference. In fact, minor changes can really foul up the system, disabling the mouse altogether or other similar havoc. I'd like to hear other more technical folks explain some of the other possible causes for differences in throughput on the two platforms. –Rick Rodriguez
#59622From: John DowdellJun 30, 1993 1:53 AM
Rick, Ah! <lightbulb flashes> It's that background polling for system events like floppy activity that makes the difference between the two — gotcha! Still, it seems there are advantages either way…guess that's what you meant by your "unfortunately/unfortunately" line. The NeXT folks used to say that their multitasking was unique, because if (say) Adobe Illustrator crashed, it would not affect other programs running at the same time. Is this in the same vein as what you're talking about? And fwiw, other multi-platform renderers also report similar speedups when moving to '486. More speed is great…until we get used to it! Thanks, John D.
#59569From: Laurie CarrollJun 29, 1993 5:00 AM
<Unfortunately, DOS doesn't multitask and neither Imagine PC nor 3D Studio may be launched from Windows. Also, I don't multitask with my Amiga when rendering, either. My point was that whether you like it or not, the Amiga's always off doing something while the user thinks it is performing only one task> Well Rick, personally I love my Amiga's multitasking. :-> Laurie Carroll AKL NZ Longdistance Autopilot
#59603From: Stan Kalisher (Impulse)Jun 29, 1993 5:32 PM
So do I, Laurie. I just wish I had the fast screen refreshes of the '486. And just wait til Imagine is running on a Pentium! Ouch! –Rick
#59653From: Peter GreciJun 30, 1993 4:21 PM
I can't wait till Imaine is running on the SGI, Forget the Bloody Pentium.
#58995From: James RoathJun 17, 1993 8:55 AM
If they did something like that I hope they make it an option and not a default! You would lose your multitasking.
#59005From: Vance SchowalterJun 17, 1993 12:38 PM
That's kinda what I meant… toggle multitasking to allow faster rendering… Unlikely, I suppose. I'll live. *8) -= Vance =- Image Master using Black Belt Systems' Imagemaster!
#58994From: James RoathJun 17, 1993 8:55 AM
OK, so that would make some sense… and while we are on the subject of multitasking and such, what plans does imulse have to support the new generation of operating systems ? I know that Autodesk is supposed to be working on getting 3d studio to run under os/2 and windows, and it's obvious that Imagine/pc needs to have this ability as well. I think that once those things happen, we can get a better impression of how all these programs work on equal terms. We did have a windows ray tracer for about 3 days, Renderize was the name, and it was painfully slow! It's very dissapointing to witness the speed of the pc version vs the amiga version, even screen updates are faster…
#58912From: SyndesisJun 16, 1993 8:36 AM
Maybe it's time to carry out the multi-platform test? I think you should be meticulous about it… such as, document the machines used in great detail, and let each platform's official representatives and interested users comment on the rules of the test beforehand. For example, it would be good to test LW and Imagine on a variety of Amiga systems. Similarly, test a half-dozen PC systems. As for the scene, I'll lend whatever conversion help is necessary! We might want to add POV, etc. to the mix, too!
#58918From: Stan Kalisher (Impulse)Jun 16, 1993 9:33 AM
I'm all for it. Actually the scenes are already designed and ready to go! Which would you recommend, the western scene on the 3D ROM, the bed or bath or one of the houses? –Rick
#58930From: SyndesisJun 16, 1993 2:34 PM
I think the Western scene is good. Care to mention this project in the 3D Studio forum?
#58935From: Stan Kalisher (Impulse)Jun 16, 1993 4:08 PM
Do you think anyone over on the 3DS forum cares? I've found those guys to be pretty sold on their software and hardware!
#58952From: SyndesisJun 16, 1993 6:06 PM
I think they're less fanatical than many Amiga-based users!
#59000From: Stan Kalisher (Impulse)Jun 17, 1993 9:50 AM
Ok, PC users may not be so fanatical, because they think they're above it all! 🙂 –Rick
#58927From: Suresh KavanJun 16, 1993 2:04 PM
Great to see you volunteering to get involved! Now we've talked about the things we want to do, let's spec it out. Everyone who wants to particpate in the rendering stage of this project, E-MAIL me with your hardware and software specifications. Include details of machine type, Ram (and RAM type), Processor (and type – motherboard vs. accelerator), Chip Ram, Hard Disk type and speed, Hard Disk Interface Type, Software and Revision Number, OS revision and anything I've missed! I will compile a database of this information. John, can you be responsible for putting a scene together (or getting one donated) and making it available in multiple formats. If you need help I just bought your program and the 3DS add-on and will happily fire it up. Eric, maybe you can act as the Amiga referee. Making sure the test isn't inadvertantly *biased* against the Amiga and or Lightwave, Imagine etc.. Can someone volunteer to take this *challenge* to the MAC forums ?? Maybe John and I can do the 3D Studio forum presentation. Comments? Other suggestions?
#58929From: SyndesisJun 16, 1993 2:34 PM
I think you'll need to coordinate this effort by a team of "experts" from each market… no ringers. Multi-platform people allowed, but they need to be considered "experts" in the programs in question.
#58957From: Suresh KavanJun 16, 1993 6:33 PM
John, Why do you say "experts only" ? If we create a scene that encompasses every major facet of 3d rendering (i.e reflections, shadows, opacity, bump maps, lighting, colour spectrum etc..) and convert this scene into the relevant formats, surely anyone with the software is capable of giving us a time result?? Am I being dense? SK.
#58996From: SyndesisJun 17, 1993 9:44 AM
3D programs advance by differentiation, so each program has one or two whiz-bang features that the others don't have. Other features are more basic and have a more dramatic effect on rendering times. For example, have you given any thought to whether the test scene is made of randomly oriented or consistently normal'd polygons? Is it polygons or triangles? LightWave does better with multi-point polygons. Any LW "expert" will probably (rightfully) argue that a triangle-heavy scene is an unfair comparison. 3D Studio uses only triangles, but its triangles can be oriented, which speeds rendering times in the same way that LightWave's polygons can be one-sided or not. Imagine's triangles are randomly oriented. So, for example, converting an Imagine object to LW and 3DS would be non-optimal in several ways: triangle orientation and no polygons. Converting LW to Imagine or 3DS would reduce all those nice polygons to triangles. (If you use Pixel to do this, the triangles will be randomly oriented even though the original polygons were oriented; but InterChange will preserve the original normal on all the resulting triangles.) I'd recommend at least two test scenes: one that originated whoops, make that at least one test scene created solely in each program you intend to compare. Let the experts in that program optimize the scene however they like. Provide a rendered image to the other platform's teams and let them translate the data as they like and optimize as they like to obtain the "same" results. You may need to agree on what constitutes a fair "tweak": removing polygons would be illegal, but re-orienting them would be OK. Along another axis, you might want to vary the scenes according to all those other whiz-bang surface attributes: render once each in wireframe, "quick" renders with simple shading, then on to the more complex additions. It would help compare the programs. You'd certainly have a lot of data to present, and it would lead to a lot of useful hot air in all the 3D-related forums. You might invite the Macromedia 3D people from the Macromedia Forum, other CAD-related companies from the CADENCE, LEAP and ACAD forums, etc. beyond the 3D Studio people.
#59153From: Suresh KavanJun 20, 1993 12:30 PM
mmmmhhhh..I guess you just made 231 great points… I'll have to go away and mull over them. Back to you soon.
#59174From: SyndesisJun 20, 1993 10:34 PM
In short, I think it's going to turn into a religious war unless you're very careful. 🙂
#59213From: Peter GreciJun 21, 1993 3:25 PM
John have you recieved the HUMAN.ZIP DXF model I uploaded to the Syndesis area yet?
#59252From: SyndesisJun 22, 1993 8:30 AM
Yes, Peter – I'll try to get to it as soon as possible.
#58967From: Erik FlomJun 16, 1993 9:05 PM
I'll send you my system stats later. I just got Interchange+ today, so I should be in a better position to work on this. I'd be willing to act as Amiga 'guru' & referee for this test. It could be interesting. (And, hopefully, not too depressing for Amigans.) If you can supply me with a Stratovision, Swivel 3D, or Infini-D compatible scene, I have some friends with Quadras and the like who would probably be interested in taking part in this 'shoot-off.' Keep On Toastin' Erik Flom
#59154From: Suresh KavanJun 20, 1993 12:33 PM
Unless there are any objections…I hereby pronounce you.. man and wife..no no I mean… the official "Amiga Guru"… I had another thought..let's get some of the magazines interested in this… perhaps the results can be published eh? After I think about this thing some more (esp. in light of Fousts observations)..I';ll post another message suggesting some ground rules.
#59169From: Erik FlomJun 20, 1993 9:11 PM
Sounds great! If you need any kibbitzing on the ground-rules, pass them my way. I'll try to add what I can. (Like that's going to be much . . .:^) If we get something that's a valid test of a wide range of platforms, I could think of several mags that would be interested. Especially if we had 'finished' images from each package. Keep On Toastin' Erik Flom P.S. One ground rule: NO touch-up of the rendered image. The image to be judged should be the final rendered output of the package. (Adding lens flare in PhotoShop would be a no-no. :^)
#58989From: John GrossJun 17, 1993 1:54 AM
I'm in too. I have access to MANY different Amiga & the newest LW setups. 1: I've been trying to send Quickmail to you… which numbers a 2: re active, and at what times? I'm hoping to make the a 3: dditions to the DXF transltorator so it can handle those 4: hinifty polymeshes you were getting frmo the om the Symbolics?? 5: 6: What /exit Post action !mail Message mailed to 71740,2357 via CompuServe Mail
#58849From: Matthew J. W. RatcliffJun 15, 1993 2:20 PM
Traditionally, a Teapot is the object of choice for rendering tests…
#58853From: Suresh KavanJun 15, 1993 2:52 PM
But it's very boring! Also, some products do much better with single object scenes than others. So while it's a good idea, I don't think it's every programs cup of tea (ha ha – groan)! SK.
#58857From: Peter GreciJun 15, 1993 4:18 PM
Tis Easier to do, get the Amiga version of Imagine and the PC version and render the EXACT same Scene and compare times.
#59147From: LOUISJun 20, 1993 10:05 AM
I'm game! I have an A3000/25 and a A2000 w/ GVP '030 Combo. I also will donate a cyborg object I've created and named "Anti-Motion Man" for this project. The attribute for him is gold. For his attributes, the reflection is set all the way up with a yellow-like color.