CompuServe Thread

#FLC RT Playback Problems

30 messages in this thread
#26442From: Ken Loss-CutlerNov 7, 1992 5:04 PM
We have created a series of FLCs which are supposed to represent approximately 2 minutes of realtime action. Problem is that when we play them back, they take around 13 seconds longer. We have 32 meg of RAM on a 486/50. There does not appear to be any disk accessing going on during playback. Suggestions will be venerated. Ken Loss-Cutler
#26448From: Nov 7, 1992 6:08 PM
Ken: Your problem lies in the fact that real-time playback of flics is almost impossible to control down to the second. Depending on how much is changing between frames, backgrounds, etc. The only way to get EXACT speed on playback is to plan on 30 frames per second and do a frame-by-frame recording to tape or laser disc. I assume you are using the play within 3D Studio and setting the check time function??? This is the closest we could come to exact time. You could try setting to 15 frames per second and halving the animation… PS – I've seen some of your stuff – keep up the good work! Greg Pyros
#26472From: Ken Loss-CutlerNov 7, 1992 11:29 PM
Greg: Thanks for the kind words about our work. We are certainly in good company, judging from the demo reel at last year's SIGGRAPH party! We normally single-frame to a BVW-75, but we need 4 (!) 2 minutes animations by Tuesday. Thats 14,400 frames (at an average render of 1.25 minutes, the case will be decided long before our graphics are complete <grin>) so we can't single frame this time. The sequences are different views of an ill-fated flight, and the atty. wants to over-dub the tower/cockpit tape so we need as close to realtime as possible. I hope there's another answer. BTW: We have tried playing back in Animator Pro. Same results. Ken Loss-Cutler
#26504From: Nov 8, 1992 10:35 AM
Ken: The way I would do it, with your time constraint <g>, would be to render flics at 512×486 if possible, or 320×200 for speed, and use flitape with a laser disk. With no pre-roll required, and at the proper resolution so that flitape won't have to re-scale the images, each one gets recorded FAST! The rendering problem is simply a matter of using more machines each doing a segment of the flic with a custom pallette so you don't get flicker between sections. You don't even have to combine the flics in Ani Pro, just batch the whole thing out in the proper order. You can also tell it to lay out 2 frames at a time, giving the effect of 15 frames per second (should be OK for this), which halves both your rendering time and lay-off time. I wish we were a little closer, we could help you with both sides, but even Fed-Ex couldn't get it to us and back to you in time. Let me know how it works out (if at all)! Greg Pyros – Pyros Partnership, Newport Beach, CA
#26517From: Ken Loss-CutlerNov 8, 1992 2:00 PM
Gregory: Thanks so much for your ideas. We are rendering at 640×480 (the atty wouldn't have it any other way). Ken Loss-Cutler
#26521From: Nov 8, 1992 3:01 PM
Ken: Seems like 512×486 is a more common NTSC standard… Greg
#26692From: Ken Loss-CutlerNov 10, 1992 12:25 AM
Yes, Gregory, 512×486 IS the common lo-res format. We have AT-Vista boards in all our systems, though, so we can go to the maximum displayable resolutions of 756×486 on single-framing. We chose the maximum resolution of 640×486 for our FLCs. Ken Loss-Cutler
#26727From: Nov 10, 1992 11:10 AM
Ken: So what did you finally do?? Greg
#26814From: Ken Loss-CutlerNov 10, 1992 10:12 PM
I'm going to make Brian write you guys a letter and tell you how we solved the FLC dillemma. He'll be out from under the gun in a day or two. It's been quite an experience, I can tell you. Most of our solutions are based on variable frame rates, but I'll let him fill you in on the details. Thanks to all for your suggestions. I only hope you learned as much from this as we did <grin>. Ken Loss-Cutler
#26825From: Nov 10, 1992 11:56 PM
Ken: I hope I learn from it also! Looking forward to Brian's letter… Greg
#26694From: Ken Loss-CutlerNov 10, 1992 1:30 AM
…I meant 640×480 for our FLCs, of course…. Ken Loss-Cutler
#26564From: bruce gorenNov 9, 1992 2:09 AM
Ken ! Since you are using BVW-75's just play out the flics and record at whatever wild speed you can get, then go down a generation using dynamic tracking to fractionally slow down or speed up the playback from the 75 and re-record. Beta-SP should hold up fine over just one generation down. Bruce
#26611From: CSA/CANov 9, 1992 1:59 PM
Bruce, Brilliant idea! I guess there is a use in computer animation for the slo-mo of a BVW-75 or a BVW-70. Kevin Krell – Computer Support Associates
#26615From: CSA/CANov 9, 1992 2:48 PM
Ooops, make that *OVER* a BVW-70. Kevin Krell – Computer Support Associates
#26617From: Ken Loss-CutlerNov 9, 1992 3:25 PM
Ah, Bruce. You forget that we're on our own now. No more edit suite down the hall <sniff>. But thanks for the suggestion, anyway. Ken
#26695From: bruce gorenNov 10, 1992 2:30 AM
Ken, What ? You guys can't afford one lousy hour of Beta-SP on-line time in someones no-frills bay to pull this rabbit out of the hat ? Whats an hour of A/B roll go for in Dallas ?, I'd guess $150-$200 tops. You're getting how many THOUSANDS to do this gig ? You could RENT two machines for a day, cost you maybe, what?, $450-$600 each, pretty stiff — but if it's the difference between delivering and failing . . . Bruce
#26813From: Ken Loss-CutlerNov 10, 1992 10:12 PM
Oh, Bruce, you should be writing screenplays! I like the way you make with the drama <grin>. First, dynamic tracking has NEVER been a good way to alter time. The next generation always looks shaky and degraded. I have tried it MANY times. Second, every penny of those THOUSANDS is earmarked for someone or something more worthy than an edit bay in which I have no equity. I NEVER go out of house unless I am dragged by a client who is paying for it. Period. The difference between delievering and failing (a word you know is not in my vocabulary) is always in-house. Sometimes it just takes a while to figure it out. Going to someone else's "easy solution" and paying them to "do it for me" does nothing but tether me to others' resources. I'm not that kind of guy. The only out-of-house resource I would ever consider using is James Hyde. (and when HE gets involved, it's like he's in-house) Ken Loss-Cutler
#26621From: Evan Yares [WeirdTM]Nov 9, 1992 3:46 PM
Ken: Go out and get yourself a local-bus 486DX2-66. It'll probably play fast enough to do the job. Evan
#26629From: Ken Loss-CutlerNov 9, 1992 5:17 PM
Evan: A 486DX2-66. Tres drole. Ken
#26655From: Yost GroupNov 9, 1992 8:18 PM
The Dell DGX2/50 will actually play 640×480 flics back in real time off of the hard disk. Cruise on down to Austin and check one out! – G
#26465From: Yost GroupNov 7, 1992 10:38 PM
How are you playing them back, Ken? – G
#26473From: Ken Loss-CutlerNov 7, 1992 11:36 PM
How are we playing them back? You mean "time corrected" mode? 3DS won't play them that way. any ideas? AnimatorPro won't play them realtime either. Ken Loss-Cutler
#26497From: Tobias FranoszekNov 8, 1992 8:11 AM
Sorry for butting in. I had the same problem with FLCs with a 640x480x256 resolution. The problem is the vga adaptor. I tested plenty of VGAs for a while and had to give up the ISA/EISA bus. For realtime (pretty close) playing I use a 486/50DX local bus computer. Even when lots of pixels move on the screen the result is still ok (not perfect but good). If the preview has to be on cideo I set the VGA (ET4000) to 50 Hz (You probably have to use 60 HZ for NTSC) and film it right of the monitor (Should be 5FG or better). If the FLC does not fit into RAM you will have to truncate it into smaller parts and then reassemble them on the VCR side. The quality for these previews for customers is ok. It is a quick and dirty solution, but the timing is pretty good. — tobias
#26516From: Ken Loss-CutlerNov 8, 1992 2:00 PM
Thanks for your suggestions, Tobias. With 32 Meg of RAM, we don't have a problem fitting it all in <grin>, and we are using an "honest" 486/50, so I suppose there's no room for improvement there either. We have a VGA-recordable board, so (luckily) recording to tape is pretty easy for us, too. The experiment continues. Ken Loss-Cutler
#26531From: MARTIN G FOSTERNov 8, 1992 3:44 PM
Ken@MediaOfExchange (strang name <g>), Here's 2 thoughts FWIW, also quick and dirty and potentially effective. Try setting up a ram disk large enough for the whole animation, but leaving enough for 3ds (at least 4mb). Then use 3ds's preview at a fixed speed, 30fps with the check speed ON, and play the animation from the ramdisk. Alternatively, have you tried James Hydes poco routine with Animator Pro which is able to play the animation down to the second accurately? Here it is: main() { int frame_count; frame_count=GetFrameCount(); SetSpeed(31); PlayFlic(frame_count); } Enter this in the Poco editor then play the animation checking for timing. You can adjust the SetSpeed value of 31 by +/- 1 or 2 to get the perfect speed. If this works for you just add my tmcode.poc timecode display to the flic for really professional looking results (obvious plug). Good luck. Martin
#26572From: Yost GroupNov 9, 1992 10:06 AM
One way you can get 3DS to play back flics with Check Speed ON is to set up a huge RAMdisk before you boot up 3DS. (if you've got 32Mb of RAM, allocate a 27Mb RAMdisk) Then, put the flic inside that RAMdisk, and use 3DS to play it with a locked playback speed. That ought to work fine. – G
#26618From: Ken Loss-CutlerNov 9, 1992 3:27 PM
How about when the FLIC is bigger than 27 MEG <g>. One of them is 29 Meg. Ken
#26650From: Yost GroupNov 9, 1992 8:08 PM
Get more memory! Split it up into two pieces? Turn off Dithering to reduce size? Shrink the live area a little bit? Put a "picture frame" around the live area to reduce total updates by 8%? Choose one from column A, and one from column A. <g> – G
#26688From: MARTIN G FOSTERNov 9, 1992 11:16 PM
Ken, did either of my suggestions work? The suspense is killing moi! RSVP.
#26693From: Ken Loss-CutlerNov 10, 1992 12:46 AM
Martin: Brian is struggling with the sequences as we speak. He says that he is going to call James later this evening or tomorrow if he can't get things running right. Sort of funny to have someone refer us to James since he lives 20 minutes up the highway and we talk pretty frequently <g>. Thanks for your input. We'll let everyone know how it works out. I think that Brian's current efforts are directed towards reducing the fps rate to get the play time correct. Ken Loss-Cutler