dirish@glab1.math.utah.edu (Dudley Irish) (05/16/91)
We are in the process of re-evaluating how we do our video recording. To this end I would like everybody who is doing video recording to drop me a note describing the hardware you use. We are primarily interested in animation recording, but I will happily collect any information you are willing to send to me. I will then tabulate the responses and post the results. Thank you, -- Dudley Irish / dirish@math.utah.edu / Manager Computer Operations Center for Scientific Computing, Dept of Mathematics, University of Utah The views expressed in this message do not reflect the views of the Dept of Mathematics, the University of Utah, or the State of Utah.
CDO@ibm-b.rutherford.ac.UK (Chris) (05/17/91)
Here are descriptions of (a) our initial system, (b) our current system and (c) what we may be about to add. (a) Initial (pilot) system VME crate with 68010 processor and 2 24-bit 768x576 video framestores, genlocked to Acron sync pulse gen, RGB output going via Abekas Cox PAL encoder to Sony VO-5850 (modified to record Hi-Band SP) via Lyon-Lamb MiniVas animation controller. Input to system: already rasterized images or (more usually) CGM (Computer Graphics Metafile, IS 8632) files. Time to render: anything between 10 seconds and 10 minutes; Time to record: 40 seconds because of PAL colour framing requirements. (b) Current production system VME crate with 68030 processor and 1 24-bit framestore with convertor to CCIR 601 (4:2:2) digital video standard; YUV output from this (equivalent to YIQ on the other side of the Atlantic) via parallel link (so real-time) in Abekas A60 magnetic video disk with 30 second recording capability Input to system: as before, plus UNIRAS; also, because Abekas also has Ethernet interface, direct download of images from mainframes (Cray and IBM 3090) over Ethernet onto the Abekas Time to render: subsecond to 15 seconds Time to record: close to nil onto Abekas (being lazy we send the record command over Ethernet which is slow but not slow enough to worry about); we download replay sequences into the Abekas (maximum 100 segments, each can be at a different playback speed from 0.001x to 32x, forwards or backwards) and then the Lyon-Lamb drops the whole sequence onto tape in one go; typically users add captions and loop or repeat sequences, so we usually drop a minute or two's video from disk to tape for a single 40 second insert edit. (c) Additions in the wings Seriously looking at LVR-6000 system from Sony (component WORM disk, 24 minutes per side of a $290 (450 dollars equivalent) disk. This would connect straight up to the existing system, as we are already running the computer in YUV although the recorder is composite. Hope that is the sort of summary you are after - get back if you need more information. Be interested to see the reuslts you get. Regards Chris Osland Head of Computer Graphics Rutherford Appleton Laboratory Chilton, DIDCOT, Oxfordshire, UK tel +44 235 44 6565
e._john_love@icecave.wimsey.bc.ca (E. John Love) (05/18/91)
I assume you're interested in low-end options as well as higher-end ones, so here goes... I'm using Amiga 1000 and 2000 computers to create 2D and 3D images and animations for educational television programs for our provincial educational TV channel. Most of our images work at the level of 'diagrams' and as such do not reallt need that many colours - 32 is usually enough. For any Amiga system there are a number of low-cost genlocks and encoders available to translate the computer's RGB signal to a composite NTSC video signal. At the Emily Carr College of Art and Design, here in Vancouver, we currently are using the 'SuperGen' genlock which is good quality, but might not be the best alternative if you want true broadcast quality. The best quality genlock I know of for the Amiga is called the 'Magni' by Magni Ltd. They are producers of broadcast quality equipment for the professional video industry, and as such have a good rep. We have used the Magni genlock in a professional video setting, recording the Magni's output directly to 1 inch tape, and the quality is most likely the best possible! I would highly reccomend this genlock - the model we used had an S-VHS output as well as a composite output. For higher quality images, I use the DCTV unit from Digital Creations. (They are also makers of the SuperGen genlock.) DCTV allows me to display 24 bit graphics on any NTSC monitor using the DCTV's composite output signal. It comes bundled with paint, digitizing, and conversion software. I use it to compile 3D animations created with the Imagine! software. (Once all the 24 bit frames are converted to DCTV's own compressed video format, the Amiga thinks they are actually hi-res picture files, so you can load them into Deluxe Paint III and compile them into a RAM animation quite easily!) I haven't had a chance to compare the quality of the output from the DCTV to that of something like NewTek's Video Toaster, but I've seen others call the Toaster's output quality 'excellent', with the DCTV's output being 'good'. Hope this info is useful to you! Cheers!! "Imitation is the Sincerest Form of Television..." ------------------------------------------------------- E. John Love The Ice-Cave BBS Vancouver BC Canada (604) 873-8452 e._john_love@icecave.wimsey.bc.ca
cn09+@andrew.cmu.edu (Christopher Kalevi Nuuja) (05/18/91)
Here's our (Pittsburgh Supercomputing Center) old and new system:
OLD: 3/4 inch umatic tape <-(controlled by)-- diaquest controller (pc based
board)
<-(fed by)-- peritek frame buffer.
: This system took about 40 sec to record a frame to tape, and between
1 second and a few minutes to render each frame. All animations
are cgm based, and are mostly from our remote users. This system is
about 4 years old
NEW: Sony LVR <-(fed by)-- sgi 4d20.
Custom software controls the LVR. The LVR can be played directly
to tape. If the animation is not all in order on the disk, or if
you just want to do some other editing tricks, the diaquest controller
can be used to control which frames and in what order the lvr outputs
to tape. Recording time is to short to measure per frame. Rendering
time is a lot quicker, but I'm not sure what the average is anymore.
We use cgm files, as well as P3D files. P3D is a lisp-based 3d modelling
format. P3D files are animated to tape either by using the sgi gl
interface, or by raytracing them and converting the resulting rasters into
a cgm file. Our cgm and p3d interpreters come with the code to control
the sony lvr, as well as the diaquest controller. They are available
via anonymous ftp from calpe.psc.edu. Please send mail to kar@a.psc.edu
requesting to ba added to either the gplot (cgm interpreter) or
p3d mailing lists, so we can keep you informed on updates, and just so
we know how many people are using our software.
Chris Nuuja
Sys Prog/Graphics App
Pittsburgh Supercomputing Center
nuuja@a.psc.edu