Can Pulsent Technology be applied to Xvid?

[kastro68]

http://www.pulsent.com/press/pressrelease032502.html


Pulsent’s technology provides a 400 percent improvement in bandwidth and storage efficiency over existing block-based video compression schemes like MPEG-2


I am just wondering if this claimed breakthrough can be applied to Xvid to make it better.


I don't specialise in Video Compression so just ignore this message if it is a stupid question.

[-h]

Most of the claims are hogwash - the biggest "gains" are found via blurring/quantizing areas of the picture that "people don't look at", not the data format itself.

XviD will incorporate more and more such psychovisual effects in the future, but 400% is a crazy figure.

-h

[avih]

>"but 400% is a crazy figure"

not to mention *impossible*

how about 75% (i.e. 4M/min -> 1M/min is 75% less)

:)

[MfA]

I dont agree, better quantization always helps ... but there are still sources of real redundancy left. Temporal redundancy in motion remains completely untapped, and multi-reference motion compensation can give some good improvements too (B-frames help a little, but compared to some of the experiments in H26L its childs play).

That aside though, wether use of dense motion (which is essentially what object/region based coding gives you, they are the means to that cause IMO) can give appreciable gains over block based methods is open for discussion. But stuff like this (http://vision.ai.uiuc.edu/scyoon/research/proj_video_compression.html) and recent publications from Christine Podilchuk and Soo-Chul Han look very promising.

[Kornylius]

4 Mb ->1Mb may be 75% less, but it is also 25% of 4 Mb. 4 Mb is 400% as much as 1 Mb :D

I don't see the groundbreaking performance in compressing a 4 Mbps stream down to a 1 Mbps stream. Look at what DivX or XviD do: They take a 4.7 GB (or more) MPEG-2 DVD file and compress it down to 700-1400 MB. on the lower end that is much more than 4fold compression without drastic quality loss.

This press release reeks of a bunch of posers showing off without really having something to offer. What was that marketing lingo? "Stealth group of scientists"? Gimme a break :)

[nikthebak]

Originally posted by Kornylius

This press release reeks of a bunch of posers showing off without really having something to offer. What was that marketing lingo? "Stealth group of scientists"? Gimme a break :)

Yes, exactly. These kind of wild claims are presented every now and then and the term "breakthrough" seems to dominate press releases :D.

Some while ago there was a similar claim in lossless audio compression. The "press statement" was almost identical with this Pulsent one, and unsurprisingly this lossless audio compression technology was never heard of ever after.

Sometimes I wonder if there is a single person behind all these "scientific breakthroughs" who tries to collect as many license fees as possible before disappearing into Cayman islands :p.

I might have came off too negative towards this matter, but until this supposed Pulsent company provides a real encoder and until there are real proofs of it's superiority, I consider it a hoax.

[Doom9]

we'll see how it performs when I get a demo copy. I asked for one yesterday.. let's see if they are willing to face some serious testing instead of just making statements via press release. Naturally I expect them to ignore my request, especially when reading how I try to bring out the worst in each compression method.

Only time will tell.