Forums [Archives] Forums [Archives] (
-   Computers (
-   -   The Future Of Video Compression? (

Jellygoose 02-05-2005 01:19 PM

The Future Of Video Compression?
From a Documentary in german television, about a guy called "Uwe Prochnow" who wants to revolutionize Video Compression... Interesting stuff. Translated through Google from :


The technical compression standards so far the internationally recognized Motion Picture Expert Group set - briefly for MPEG mentioned.  But now the procedure of reason wants to revolutionize on.  Uwe Prochnow wants to compress videos in such a way in television quality that they fit without considerable quality loss by each usual voice grade channel.  This was considered to experts so far as impossible.  If Prochnows vision should carry out itself, this would revolutionize the transmission of video data with lasting consequences for the billions-worth market approximately around the MPEG licenses.  In his company Atvisican develops Uwe Prochnow for four years with its coworkers a new image processing system.  The 31-jaehrige a video compression procedure, which with substantially smaller data rates than so far never since been image quality transfers, promises inventors made of Saxonia-Anhalt.  With fewer bits a better picture.  A part of this new image processing system is contained in a Multimedia box, which developed Atvisican.  Uwe Prochnow blocked therein only the best and most expensive parts.  Whether video, television, telephone, stereo or computer - the box is everything in one and is called therefore also "Allcanview".  For 12.800 euro one is to receive thereby the at present best DVD Player, combined with the only HDTV receiver in Germany - one, we says could the "Allcanview" not even test.  But the Multimedia box is not Prochnows vision, it is - although in small numbers of items already reality.  The vision exists in a video compression procedure, which differs in principle from the pervasive MPEG standard.  Prochnow:  With us the entire procedure refers to no more individual cluster blocks, but the entire picture is also again developed object moderate and for color processes and background vectorially seized and in these completeness and again given, and not the calculation and lining up of individual pixel cluster blocks.  And I hope that I this procedure, because it is technologically also more fastidious and better against MPEG will also succeed.  Without data reduction, thus without compression a film consists like a thumb cinema of frames.  In a compressed film the changing picture parts will only transfer in a screen sequence.  Picture parts, which do not change, do not have again to be transferred also.  A MPEG video is accurately specified with the compression in his future dissolution and quality.  Tom Horton:  I believe, it give during the MPEG coding two problems.  First they work with fixed algorithms.  There some people looked at themselves a few pictures and said, this are the average of the pictures, with which we will have to do it with coding.  And then we build these algorithms, which also compress these picture types.  But the real world does not fit these picture types.  Things pass simply - a car comes us against, water squirts - each picture are singular.  And MPEG has problems to seize this.  If MPEG is confronted with a singular situation, the picture worsens considerably, or the file becomes simply too large.  Prochnow:  Now there are different interpretations and remarks of the MPEG codecs, but if we orient ourselves there at the front runners, and those are quite at present the Windows Media 9 codecs and on the other hand the entire number of the Divx codecs, can we today still say that we achieve a twenty-way higher compression to the leading product, under same capability characteristics as the raw material to it are.  twenty-way better than MPEG?  That makes Atvisican and Uwe Prochnow interesting not only for the advertising industry, but also for the bank trade.  John Ryan international sells world-wide technical services approximately around the bank trade.  Gary Madgwick is responsible for the entire technical conversion.  In a world-wide network thousands of branch banks are supplied with information - also with pictures and videos.  Uwe Prochnow:  Since the topic object recognition was so far a mainly military topic, however now due to the rising safety requirements ever more a civilian topic will, were naturally the first training objects military things, e.g. night vision photographs should be accordingly analyzed, or different recognitions of vehicles, types identification, which is very much, very important in military situations that one can accomplish thus a completely clear enemy and friend separation.  And that were already first principles and basis beginnings, in order to really work with the topic object recognition from knowledge conditions.  That is not at all carry outable with the conventional MPEG procedure.  I think, in order to summarize in substantial points this object recognition and the compression system, must one only once know that this procedure differs in principle from the past video compression procedures, because the picture in individual blocks am not analyzed here, or am not taken apart, but it is collectively seized.  It is possible to pull and even in a data base of the decoder for several Frames store objects out of these pictures and only with modification information feed.  On the basis sharpness differences, positions in the picture and independent movement, are to be assigned to objects positions.  Objects with high priority are computed therefore more exactly.  Complex color processes, e.g. in would background are likewise analyzed and mathematically are described.  In contrast to the MPEG standard, with which all pixels are equivalent and concomitantly in only one passage are compressed, is to make the Atvisican available compression procedure for objects with different priorities also different procedures.  A picture will therefore go through many different compression stages and thus will be supposed extreme data reductions to become possible.  30 seconds a Werbespot is large in its poorest resolution still 25 MT.  Uwe Prochnow succeeded it to compress for Tom Horton a Pampers Spot on only 1.5 MT.  David Feldt:  I was impressed above all of the fact that the procedure akkurat also under extreme load gracefully and worked and that the picture could be scaled.  One of the terrible problems, which we have with the conversion for large displays, is that from small pixels large, fat, ugly pixels becomes.  There one gets stairway steps and artifacts and a very unattractive picture, if one does not regard it from very large distance.  And, O miracle, I got a picture, which could be scaled apparently on completely unbelievable sizes.  And also still with an extreme quality when high resolutions.  First I could not introduce myself, how he had made that.  We expected that this corresponded not to the truth and would not function.  But after we were in meals, Uwe it met, which looked at development processes and which could also test product, became very fast clearly that it functioned nevertheless.
Uwe Prochnows company webpage :

Have fun reading ;)

Prodater64 02-06-2005 12:47 PM

Could you explain it in a short way?

All times are GMT -5. The time now is 11:19 AM  —  vBulletin Jelsoft Enterprises Ltd

Site design, images and content © 2002-2019 The Digital FAQ,
Forum Software by vBulletin · Copyright © 2019 Jelsoft Enterprises Ltd.