tiefel@sunshine.Kodak.COM (Lenny Tiefel) (02/19/91)
Within the next two months, Eastman Kodak Research Laboratories will be generating video movie sequences. We are making these images available to promote greater interest in video animation algorithms research. Each frame is 512 pixel elements (one byte per pixel) by 512 lines in three color bands. Tentatively, we'll use gif format, but we are open to other suggestions. (In addition, I've written a video movie program in basic xlib to display these images at 24 frames per second.) But for security reasons we cannot allow anonymous ftp, so we would like some friendly university site to volunteer to store and distribute these images. If anyone is interested in receiving these images, or in video animation algorithms research contact me at the address below. ----------------- Len Tiefel | ___ ________ | Eastman Kodak Company | | / / | | Research Laboratories | | / / | | Imaging Science Laboratory | |< < K O D A K| | Rochester, NY 14650 | | \ \ | | (716) 722-2250 FAX (716) 722-0160 | |__\ \________| | email: tiefel@kodak.com | | Standard disclaimers apply ----------------- "Tell me what company thou keepest, and I'll tell thee what thou art." -- Cervantes
tiefel@sunshine.Kodak.COM (Lenny Tiefel) (03/16/91)
The Eastman Kodak Research Laboratories will be generating video movie sequences and providing these freely through anonymous ftp within the next three months. (The details will be given later.) So, to provide the world's research community with the most useful images, consider the following questions: What image format would be most useful? Example: gif What image size (pixels, lines, bands) would be most useful? Example: 256x512x3 Would monochrome be more useful than color? How many bits per pixel would be most useful? Example: 12 Would computer generated motion drawing be more useful than actual video movies? What sort of scenes would be most useful? Example: a moving car How many frames per scene would be most useful? (Assume 24 frames per second.) Example: 100 ----------------- Len Tiefel | ___ ________ | Eastman Kodak Company | | / / | | Research Laboratories | | / / | | Imaging Science Laboratory | |< < K O D A K| | Rochester, NY 14650 | | \ \ | | (716) 722-2250 FAX (716) 722-0160 | |__\ \________| | email: tiefel@kodak.com | | Standard disclaimers apply ----------------- "Productive achievement is our noblest activity" -- Ayn Rand, American author (1905-1982).
root@fuzzy (Dr. Michael M. Cohen) (03/16/91)
If we are talking video, why assume 24 frames per second?
CDO@ibm-b.rutherford.ac.UK (Chris Osland) (03/18/91)
On 15 Mar 91 18:39:11 GMT Lenny Tiefel said: > >The Eastman Kodak Research Laboratories will be generating >video movie sequences and providing these freely through >anonymous ftp within the next three months. (The details >will be given later.) > >So, to provide the world's research community with the most >useful images, consider the following questions: > > >What image format would be most useful? >Example: gif ISO standard 8632 (Computer Graphics Metafile) provides for all conventional graphical primitives, is self-contained and self-describing and we have used it for the production of videos that have been used by scientists and broadcasters for some years. CGM includes Cell Array, which may or may not map 1-to-1 onto a pixel array. > > >What image size (pixels, lines, bands) >would be most useful? >Example: 256x512x3 > If the pictures are stored as Cell Arrays in CGM, the intended aspect ratio and the resolution of the picture is stored as part of the CGM and so the picture can be correctly scaled onto any physical device. It can also be anisotropically stretched to fill a screen or to save resampling the image. > >Would monochrome be more useful than color? > No. > >How many bits per pixel would be most useful? >Example: 12 > 8 bits/pixel/colour is easiest to process if representing RGB; standard digital TV (in CCIR 601 4:2:2 coding) is 8 bits for each component (Y, U, V in UK; possibly called Y, I, J in US?) but with colour difference only stored for every other pixel. If the suggestion of 12 is bits/pixel (so 4 bits/pixel/component) I would say that is much too few - coarse colour changes will result. Minimum 6 per component. > >Would computer generated motion drawing >be more useful than actual video movies? > No comment. > >What sort of scenes would be most useful? >Example: a moving car > For the research community I would have thought a mixture of ordinary images (as per moving car) and scientifically oriented images (eg views of a sculpture rotating on a pedestal, rocket motor firing) would both be useful. > >How many frames per scene would be most useful? >(Assume 24 frames per second.) >Example: 100 I am intrigued that the suggestion is that of a film, not a video, frame rate. 100 frames seems fine. CGM (see long way above) provides run length encoding as standard for call arrays and this can often provide a factor of 20 or so compression on computer generated pictures. Because of noise, compression of camera-captured pictures is not so good. Chris Osland Headof Computer Graphics Rutherford Appleton Laboratory Chilton, DIDCOT, Oxfordshire, UK OX110QX Tel +44 235 44 6565 Fax +44 235 44 5808
okeefe@cs.Buffalo.EDU (Paul O'Keefe) (03/20/91)
In article <9103190446.AA14932@psi.rutgers.edu>, CDO@ibm-b.rutherford.ac.UK (Chris Osland) writes: |> On 15 Mar 91 18:39:11 GMT Lenny Tiefel said: |> > |> > The Eastman Kodak Research Laboratories will be generating |> > video movie sequences and providing these freely through |> > anonymous ftp within the next three months. |> > |> > What image format would be most useful? |> |> ISO standard 8632 (Computer Graphics Metafile) provides for |> all conventional graphical primitives, is self-contained and |> self-describing and we have used it for the production of |> videos that have been used by scientists and broadcasters |> for some years. CGM includes Cell Array, which may or may not |> map 1-to-1 onto a pixel array. |> |> Chris Osland |> Headof Computer Graphics |> Rutherford Appleton Laboratory CGM may be a standard, but is not widely supported and were designed for GKS-sytle 2D primatives, not image sequences. Since these are sequences, they should be in a format which supports the animation of image sequences and has a public-domain sequence animator. The NCSA (Natl. Center for SuperComputer Applications, U. of Illinois) HDF image format fits the bill. The NCSA HDF image format and image viewers: 1) support the animation of color image sequences 2) are available in source and binary for numerous system types (mac, unix, PC, ...) 3) are maintained and improved regularly 4) are public-domain 5) are available via anonymous ftp at ftp.ncsa.uiuc.edu (128.174.20.50) Paul O'Keefe okeefe@cs.buffalo.edu
rthomson@mesa.dsd.es.com (Rich Thomson) (03/20/91)
In article <66302@eerie.acsu.Buffalo.EDU> okeefe@cs.Buffalo.EDU (Paul O'Keefe) writes: >CGM may be a standard, but is not widely supported and were >designed for GKS-sytle 2D primatives, not image sequences. I agree with Paul about the unsuitability of CGM. >Since these are sequences, they should be in a format which supports >the animation of image sequences and has a public-domain sequence animator. >The NCSA (Natl. Center for SuperComputer Applications, U. of Illinois) >HDF image format fits the bill. The Utah Raster Toolkit also fits the bill. A sequence of images can be kept in a single file and viewed as a movie with getx11 (with the -m flag). The URT is publically available, see the frequently asked questions posting in either comp.graphics or alt.graphics.pixutils. I haven't seen many conversion tools that understand HDF, although most image manipulation tools I've seen understand the Utah Raster Toolkit RLE format, though. Of course, I've not seen everything ;-}. -- Rich -- ``Read my MIPS -- no new VAXes!!'' -- George Bush after sniffing freon Disclaimer: I speak for myself, except as noted. UUCP: ...!uunet!dsd.es.com!rthomson Rich Thomson ARPA: rthomson@dsd.es.com PEXt Programmer
cn09+@andrew.cmu.edu (Christopher Kalevi Nuuja) (03/21/91)
Hi, I'm forwarding this message from a co-worker: I'd just like to point out that cgm allows multiple frames to be put in one file, and therefore supports "animation of image sequences". Date: Wed, 20 Mar 1991 12:30:12 EST From: ANDREWS@CPWSCA.PSC.EDU (Phil) Message-Id: <910320123012.25003255@CPWSCA.PSC.EDU> Subject: RE: HDF vs CGM To: NUUJA@a.psc.edu, WELLING@SEURAT.PSC.EDU X-Vmsmail-To: SMTP%"NUUJA@CPWSCA.PSC.EDU" At the Pittsburgh Supercomputing Center we have been using CGM successfully for movie sequences of all possible lengths for several years now. In fact all of our two-dimensional graphics is in CGM format. We view CGM files with our locally developed CGM interpreter, GPLOT which is freely available via anonymous FTP, as are libraries for the production of CGM files. The present official release of GPLOT (4.3) runs on UNIX (including UNICOS) and VMS and supports many output devices and two animation controllers: the pc-based Diaquest and the SONY LVR video disk recorder. A new version, in beta test now, is written in C++ and runs on UNIX and MAC systems, the UNIX version supporting animations on X-servers via a MOTIF interface. We hope to release a PC windows-based system in a few months and would much prefer that publicly available image sequences be in CGM format rather than HDF. Although both source and binaries are available via anonymus ftp we prefer that you first send mail to Anjana Kar (kar@psc.edu) for instructions and addition to our mailing list. -Phil Andrews (andrews@psc.edu)
kas@cs.aber.ac.uk (Kevin Sharp) (03/23/91)
>>>>> On 15 Mar 91 18:39:11 GMT Lenny Tiefel said: Lenny>How many frames per scene would be most useful? Lenny>(Assume 24 frames per second.) >>>>> On 18 Mar 91 11:49:10 GMT, CDO@ibm-b.rutherford.ac.UK (Chris Osland) said: Chris> I am intrigued that the suggestion is that of a film, not a video, Chris> frame rate. 100 frames seems fine. The frame rate should be match screen refresh rate of the monitor on which the movie is being played ie. if the refresh rate is 50Hz then the frame rate should be 50 fps, 25 fps, or 16.6 fps etc. Unfortunately there's no standard for this. UK systems are often 50Hz. US systems are usually 60Hz. Some flicker free systems offer 80Hz etc, etc, etc. Attempts at converting between frame rates are invariably unsatisfactory --- usually resulting in jerky motion. If an object based animation was used, this could be overcome. The frames can be generated at the required frame rate at `view time'. However, this is very restrictive. I suggested in my earlier posting that you should really store successive frames as `diffs'. What might work is to generate the movies at a very high frame rate. This way the `diffs' will be fairly small and the overall increase in data may be moderate. A better result should be obtained by converting from a high frame rate. -- -- Kevin Sharp, UUCP : {WALES}!ukc!aber-cs!kas Research Associate, JANET: kas@uk.ac.aber.cs AI and Robotics Research Group, PHONE: +44 970 622450 Department of Computer Science, University College of Wales, Aberystwyth, Dyfed, UK. SY23 3BZ
CDO@ibm-b.rutherford.ac.UK (Chris Osland) (03/25/91)
Thanks for the reaction re videos. I notice that there was an unfortunate juxtaposition in my original note. Having talked about frame rates (24, 25 30 etc..) I added a comment about the length of the examples saying '100 frames would be fine'. Reading it back, this might seem to indicate '100 frames/sec would be fine'! That was not my intention. Chris Osland
gilge@ICSI.Berkeley.EDU (Michael Gilge) (03/27/91)
To make the list complete, I propose using the KHOROS package. It is PD software and the file format supports movies. There is actually a tool called 'animate' which lets you display an image sequence in real-time on your workstation. For more information on KHOROS you can send mail to khoros-request@chama.unm.edu Michael Gilge, gilge@icsi.berkeley.edu International Computer Science Institute 1947 Center Street, Suite 600 Berkeley, California 94704 Phone: 1-415-642-4274-134 Fax: 1-415-643-7684