CompuServe Thread

#Raptor questions

26 messages in this thread
#83941From: Leo HurtadoJun 6, 1994 10:21 PM
Hello! I just got my Raptor today (will hook it all up tomorrow), but I have some questions regarding Beta issues stated in the letter… What does it mean when it says "Image sequences sent to the Screamer processors, or Raptor, will cause both the Toaster and processor systems to lock. 1 to 60 frames may render before the lock occurs."? The way I understand it (and I hope I'm very wrong) is that after rendering 60 frames it will lock up?? I have been waiting for delivery since the pre-order stage 'cause I have a 1500+ frame animation so…. I would appreciate a quick response from NewTek employees in the forum, and anyone else that has a Raptor installed as well. Thank you all for your responses, in advance. POlo
#83965From: GREG TEEGARDENJun 7, 1994 2:51 AM
Any animation you try and render that has an animated texture, -i.e., image sequence, will lock up after 60 or so frames. Same for sequencial bkg images. For some reason, the ScreamerNet software dosen't like image sequences as part of a scene file. That does not mean that it will lock up after rendering 60 or frame of a normal animation, just one that contains sequentially numbered image maps. GT
#84032From: Leo HurtadoJun 7, 1994 10:51 PM
Greg, Thanks for putting my mind at ease! POlo
#84048From: Joe MaulucciJun 7, 1994 11:58 PM
I guess then… PAR Sequence + ScreamerNet = lock up This is good to know. Joe Maulucci >>> Accadia Electronic Arts DTV & DTP BBS 1-716-882-1774
#84144From: John GrossJun 9, 1994 2:53 AM
It's important to know that the reason for this (Image sequence probs) is not the Raptor or LightWave. By all indications it appears to be a networking bog down problem with the Networking software used to hook the Toaster to the Raptor. As greg said, eventually there will be a solution and you will not thve to babysit. JG
#84066From: Stephen BaileyJun 8, 1994 10:54 AM
I do "normal" animations all the time using sequential image maps. If what you say is true, then how does all the sequential mapping we see in SeaQuest get accomplished? Locking up after 60 frames sounds like it would be a big hassle. Perhaps, I misunderstood the previous posts…I would love to someday own das Rapteur 🙂 and hadn't heard about the problem. Stephen Bailey Sawyer & Bailey Film and Video – Chicago
#84122From: Joe MaulucciJun 8, 1994 11:28 PM
Nothing like rendering 60 frames, crash, reset, render the next 60, crash — a minor bug, right. Well I guess it would depend on rendering minutes per frame. 10 minutes per would mean resetting every 10 hours. Joe Maulucci >>> Accadia Electronic Arts DTV & DTP BBS 1-716-882-1774
#84135From: GREG TEEGARDENJun 9, 1994 12:45 AM
Nope, you didn't misunderstand my post. It is somewhat of a pain having to babysit the Raptors when we have image sequences, but we usally use the Toaster rendering farm if we have composite shots or other sequencial images as part of a scene. Our farm consists of over 40 dedicated machines run through our own proprietary control software. Although it requiers the use of a large number of machines, it is non the less very efficient. This is a brut force approach to solving a problem though, and not a very economical one either, hence the Raptor. As soon as the problem is fixed, I would imagine that we'll dump the Amiga farm altogether and replace it with Raptors. The Raptors have definately proven their ability to chew through complex scenes, as they excell at FPU intensive operations such as raytracing. I would imagine that future incarnations of the Raptor will be even more powerfull, especially at FPU operations, which will level the playing field for cost effective houses like ours, not to mention the huge advantages provided to the home based professional. In spite of this one shortcomming, it will be fixed, and anyone who wishes to pursue a professional career at home or through a small business would be served well by getting one. GT
#84143From: John GrossJun 9, 1994 2:53 AM
Greg, It sounds like you work for Deskstation! Much better than some of those software companies, eh?
#84197From: RAD Moose @ NewTek, Inc.Jun 9, 1994 6:56 PM
How is your Screamer? DH
#84229From: John GrossJun 10, 1994 1:16 AM
Hey David, Still Screaming. Only little prob is the sequences… Thanks for asking. JG
#84207From: GREG TEEGARDENJun 9, 1994 8:46 PM
My mom always told me, "if you can't say something nice…." GT
#84223From: Leo HurtadoJun 9, 1994 11:14 PM
PMFJI,~ Amen to that!
#84314From: Grant BoucherJun 11, 1994 12:48 PM
Was that Greg Teegarden speaking or someone using his Compuserve account!? I agree with Greg's post and predict that all your image sequencing problems will be fixed shortly… Grant Boucher – Amblin Imaging "It's hard to believe grown-ups do this for a living."
#84293From: glen millerJun 10, 1994 11:27 PM
GREG, So what are the work arounds to this sequence/Raptor problem? GOd, half of a show open I am working on uses sequences….sometimes six or seven at a time… what does that mean, it'll lock up after 10 frames? :} Well, I guess I have a lot to look foreward to! GM
#84011From: Don PetersonJun 7, 1994 4:47 PM
Leo, The message you posted is dated yesterday. That means you hooked up the Raptor today. SO?????? How's it doin' Thanks to Greg T. for explaining the problem on image sequencing. I couldn't have explained it any better. Don
#84037From: Leo HurtadoJun 7, 1994 11:27 PM
Don, WEll, it looks goos so far! I just came back from the studio about an hour ago, after hooking it all up (I have two VT4000 working with ParNet and sharing a Bernoulli 150MB Box, on top of that I have a PAR and a TBC IV for animation control and videotaping to HD)! It has not been an easy set-up, I guess it would be better if the PAR helped out a bit here! The render times look good, we tried rendering simple stuff and found out that although the rendering time itself was around 15 seconds for a basic pan movement from an object to another, the saving of the RGB data to the HD took somewhere around 32 seconds in the average! That's a hell of a long time for saving to disk, no? I figure that more complex scenes will take better care of the rendering/saving time. We left it working overnight on a frame that Raptor-less it took 64 hours to complete on a A4000 w/18MB RAM. We'll see how it goes, but I'll keep you posted on any new stuff I find (other than a few lock-ups I experienced today, but hey, it's beta software)… POlo
#84106From: Adib NasleJun 8, 1994 8:27 PM
Leo, You can still do image sequences, and I have seen someone do them on a pre-beta-beta version of the software and hardware. You just have to babysit it until Newtek solves the problem – but it can be done with professional results. My guess is that your 64 hour animation took around 6.3 hours with the Raptor – nice huh? When I first got Wavemaker, I tried rendering the Universe scene just to see how it looked – well it took 14 days. For the hell of it, I tried it with the Raptor and it took 1.3 days (with reflection and shadows ON). Thats a lot of days saved! By the way, try the independent rendering option – its suppose to be even faster, and the images are stored in sequence.
#84115From: Leo HurtadoJun 8, 1994 10:32 PM
Adib, 6 hours? Try __41 MINUTES__!!!!! I cannot believe it! We are already thinking of getting a second Raptor to network all the A4000s and the four 2000s we have! Unbelievable… POlo
#84198From: Adib NasleJun 9, 1994 7:24 PM
Get that on film! I am sure Don would like to document that for his advertising. What kind of scene was it that provided such awsome acceleration from the Raptor?
#84222From: Leo HurtadoJun 9, 1994 11:11 PM
Adib, I was just one frame from a 1500+ animation we just finished rendering prior to arrival of Raptor. The scene was finished (and began to be rendered) in October of last year, and it finished rendering a week before the Raptor came in. We had one of the frames in our "Hall of Fame" for rendering records (not exactly _SPEED_ records). This one frame is the last part of a telescope suspended in space. To create it with a realistic look of lens, we used lots of refraction coupled with reflection and a nice map of glass, all this with antialias in Med-Res. Plus ray traced shadows and reflections! This _ONE_ frame took 64 Hrs. 44 Mins. 34 Secs to render back in — I think — January. We saved it in case we would have a problem with the target tape (and out of pity for ourselves, really). When I got the Raptor, I wanted to put it to the test and try that one frame with the same settings as before, and there you have it: 64 Hours + = __41 MINUTES__! Incredible, really… POlo
#84329From: john bennettJun 11, 1994 5:40 PM
I think 'Maybe' your other rendering machine (A4000 ?) is tweaked from the sounds of the results of your Raptor tests. Some how the configuration may not be right. 65 hours for a Video res frame??? You must be doing some awsome stuff. The longest its taken so far on a single frame for me is about 12 hours, and thats for Print res with aliasing. I can't wait to hear reports of Raptor rendering times compared to a quick 2000/040 or Warp board. Have Fun. ….John
#84393From: Leo HurtadoJun 12, 1994 11:59 PM
John, I can hardly believe it myself! No tweaking on the Amiga though (basic 18MB RAM and 200MB HD(?)… The frame was #114 of a 1500+ animation with lots of reflection and refraction (plus antialias and such, at med-res). We plan on re-rendering maybe the whole thing in hi-res (we had some tape drop-outs when we recorded the animation). POlo
#84116From: Leo HurtadoJun 8, 1994 10:38 PM
Don, I just rendered a frame that Raptor-less took 64 hours to finish. ___41 MINUTES___!!!!! with Raptor! More exploring left to do…. Way to go NewTek and Deskstation Tech!! POlo
#84320From: Don PetersonJun 11, 1994 3:00 PM
Leo, Thanks for the kudos! Sounds like you found a test that offers improvements on the way, way upper end of the spectrum. I wouldn't expect to see that kind of improvement on many other things but I'm happy you feel you got your money's worth. I want to also thank those of you who have gotten Raptors for your generous feedback on the product and for sharing the rendering improvements you've seen. We have many new things planned for this market. Phil Kurz and Martin Fanning are doing a tremendous job building support for our product in the desk top video market. Even Microsoft has offered their support and are very excited about having a toe in this incredible market. In the next several months we intend to deliver even more exciting products. Stay tuned…………. Don
#84392From: Leo HurtadoJun 12, 1994 11:56 PM
Kudos well deserved, Don! POlo