PDA

View Full Version : xpeg


pieter1976
13th November 2004, 20:10
I am allways looking for better compression algorithms.
This time I found the folowing interesting website.

Xpeg (http://www.yet2.com/app/list/techpak?id=29320&sid=350&abc=0&page=tpprint)

They say that it is 10 time better than JPEG.
I also red that nasa is using it.

another one

scrunch (http://www.remotesatellite.com/products/scrunch.htm)

akupenguin
13th November 2004, 21:33
Do we need a new section: "New A/V Formats - Hoaxes" ?

And they don't say that NASA is using xpeg. They say it's based on NASA's work on image postprocessing (not even compression).

pieter1976
13th November 2004, 21:43
you are right about the NASA thing.
I should read more Carefully.

But they do sell a webcam that does 30 fps.
See here solectronics (http://www.solectronics.com/)

It must use a compression.
But I suppose MPEG could be streamed over a USB 1.1 cable aswell.

akupenguin
13th November 2004, 23:13
USB 1.1 allows 12mbit/s. Uncompressed video at 352x288x30fps is only 36 mbit/s (YV12 or RGBG). It doesn't take a very sophisticated compressor to transmit that (conventional lossless huffyuv is at least 2x, more if you include some denoising or temporal prediction). Even if the camera works as specified, Solectronics would not be the first company to sell such cams, nor the best. There are plenty of 640x480 cams out there, though they do have to use either some sort of (stronger, conventional) lossy compression, or USB2.

Anyway, their hardware department doesn't change the fact that xpeg is an obvious hoax:
When used in conjunction with JPEG, MPEG, or any other compression algorithm, XPEG increases compression of images or video by a magnitude of ten or greater without loss of quality... Works with any compression including MPEG1, MPEG2, MPEG4, MPEG7, Cinepac, Indeo, Microsoft Media, Real, Quicktime, JPEG, H.320, H.323, H.324, 2.5G, 3G, PCS, etc... Does not require custom codecs or codec distribution.... XPEG performance is independent of motion within videos. You don't gain compression by layering one compressor over another. If they had an improvement, they would release a new codec, not claim some "multiplier". Which leads me back to my previous post: It's not a compression algorithm, it's a postprocessor. And since different codecs produce different artifacts, it can't even be a good postprocessor.

As for Scrunch: /me laughs at how they use JPEGs to demonstrate their lossless compressor.

For both, and any aspiring future hoaxes: No (real) company will ever announce a 10x improvement over the previous standard. They will go to market as soon as they can possibly claim a 2x improvement, and the actual gains will be less than 1.5x, often much less.

RadicalEd
14th November 2004, 00:23
Works with any compression including MPEG1, MPEG2, MPEG4, MPEG7, Cinepac, Indeo, Microsoft Media, Real, Quicktime, JPEG, H.320, H.323, H.324, 2.5G, 3G, PCS, etc... Does not require custom codecs or codec distribution.... XPEG performance is independent of motion within videos.

lol, half of those don't even deal with compression.