chris@castlab.engr.wisc.edu (Christian Rohrmeier) (06/30/91)
Hello all, I would like some info on the NeXT dimension board: has it got JPEG? If so, will it work ok for motion instead of still video? Can you get MPEG on a ND? DO I want MPEG??? Thanks for answering these rather broad questions... I have in mind using the ND for live video, and was wondering if the ND would do the job (digitize right of off cable, and manipulate the images.) -- Christian Rohrmeier chris@castlab.engr.wisc.edu
lacsap@media.mit.edu (Pascal Chesnais) (06/30/91)
In article <1991Jun30.002650.11570@doug.cae.wisc.edu> chris@castlab.engr.wisc.edu (Christian Rohrmeier) writes: > I would like some info on the NeXT dimension board: has it got > JPEG? If so, will it work ok for motion instead of still video? NeXT 2.x supports software JPEG, and if hardware is present (and obvious it has to announce itself) it will take advantage of it. This was explained when people started noticing a new process -imageserver appearing. This little daemon figures what is around to do JPEG. Currently all the NeXTdimensions we have seen do _not_ have hardware circuits for doing JPEG decompression. No surprise, C-Cube was unable to produce a chip that could do full 640x480 30fps JPEG compression. There is no socket for a single chip on the board since C-Cube could not guarantee the pinout of a chip able to do full motion full resolution compression. There is an undocumented connector on the NextDimension board. No one at NeXT will tell you what it is for at present, which is fine, why get people's hopes up. MPEG is an interesting point, JPEG is an intra-frame coder, which means it does not benefit from information present in other frames. MPEG is a interframe coding scheme, but does not do as well as JPEG for stills. MPEG was also not finalized when NeXT commited to JPEG for the ND. Still images are pretty important in the NeXT community. To support a motion coder such as MPEG, NeXT would have to add more hardware to the board, while with the C-Cube it was believed it would do an adequate job at compressing video (adquate meaning that people are willing to put up with 6 hour mode VHS with tracking errors will think the video looks cool on the NeXT....) All video compression schemes suffer from visible artifacts at low bandwidths, which are more offensive seems to be a matter of taste. One point not addressed by most "multi-media" platforms is synchronized sound and video... how to accomplish that in asynchronus environments such as unix/X or mach/NeXTstep is quite a challenge. pasc -- Pascal Chesnais, Research Specialist, Electronic Publishing Group Media Laboratory, E15-351, 20 Ames Street, Cambridge, Ma, 02139 (617) 253-0311 email: lacsap@plethora.media.mit.edu (NeXT)