[Gooberpatrol66]

I just attended a brilliant IEEE presentation in Austin by a man who is trying to supplant MPEG. His algorithm, called ZPEG, can supposedly deliver visually perfect (not lossless, but difference less than quantization error, making it invisible) video at 1Mb/s, and lossless compression at a 10:1 compression ratio. Since the algorithm is symmetric, any device that can decode a given ZPEG video in realtime (which would includine things like cellphones) can also encode it in realtime, and it takes only a fraction of the computation required by MPEG. Website's at zpeg.com. They have a downloadable demo that you can mess with to make your own videos and see the video quality and the bitrate of the codec for yourself. He mentioned he was having trouble getting support in the industry so I thought I might raise some interest among multimedia enthusiasts.


JPEG and MPEG both perform 2-dimensional transforms. ZPEG works by performing a 3-dimensional (width, height, and time) Fourier transform, which is a lot more efficient because an entire extra dimension is being compressed. So instead of each frame containing its own grid of macroblocks, each macroblock spans several frames. Also does away with MPEG macroblock search algorithm, resulting in a massive speedup.


What do you guys think? Are you interested? Subjectively perfect video at 1Mb/s that you can encode in at least realtime on your iPhone? I'm very excited.

This is totally legit, by the way. The guy who gave the presentation, Raymond Westwater, is a PhD and the company that owns ZPEG, Futureware, is the same one that made the AVI specification.


[Tell me if this isn't the right forum to talk about this.]

 

[Richard Paul]



Quote:
Originally Posted by Gooberpatrol66  /t/1491307/anyone-heard-of-zpeg#post_23750069


I just attended a brilliant IEEE presentation in Austin by a man who is trying to supplant MPEG. His algorithm, called ZPEG, can supposedly deliver visually perfect (not lossless, but difference less than quantization error, making it invisible) video at 1Mb/s, and lossless compression at a 10:1 compression ratio.

I am very skeptical and the YouTube videos he made were posted almost a year ago and yet I can't find any independent tests for ZPEG.

Quote:
Originally Posted by Gooberpatrol66  /t/1491307/anyone-heard-of-zpeg#post_23750069


Website's at zpeg.com. They have a downloadable demo that you can mess with to make your own videos and see the video quality and the bitrate of the codec for yourself.

That encoder is only for real time demonstrations and what is needed is an encoder that can compress video to a file.

Quote:
Originally Posted by Gooberpatrol66  /t/1491307/anyone-heard-of-zpeg#post_23750069


He mentioned he was having trouble getting support in the industry so I thought I might raise some interest among multimedia enthusiasts.

Even for a multi-billion dollar company it is very hard to make a video compression format successful. For example VP8 has had limited success and it is supported by both Google and open source advocates.

Quote:
Originally Posted by Gooberpatrol66  /t/1491307/anyone-heard-of-zpeg#post_23750069


This is totally legit, by the way. The guy who gave the presentation, Raymond Westwater, is a PhD and the company that owns ZPEG, Futureware, is the same one that made the AVI specification.

The company helped develop the original AVI container but without independent testing I wouldn't put much hope in ZPEG.

 

[Gooberpatrol66]



Quote:
I am very skeptical and the YouTube videos he made were posted almost a year ago and yet I can't find any independent tests for ZPEG.

Neither can I, and that's why I wanted to make this post. Virtually no one knows about it. I want to see some third-party opinions from people who are better experts than me, or reviews, or discussions, or tests, or debunks, or just something. This is the first time I've been interested in a technology so obscure that a Google search can't tell me more about it. I didn't believe the claims at first either.

Quote:
That encoder is only for real time demonstrations and what is needed is an encoder that can compress video to a file.

You don't need a file to measure the compression ratio, right? You can just monitor the amount of data sent over IP? (That's assuming you don't trust what the program's telling you.) At least you can check out the VQ and the CPU usage. But you're right, it's not exactly very useful to the average member of this forum until you can save it as a file.

Quote:
Even for a multi-billion dollar company it is very hard to make a video compression format successful. For example VP8 has had limited success and it is supported by both Google and open source advocates.

From what I read about it on Wikipedia, VP8 doesn't really seem to offer any clear advantages over competing formats besides being free, but that's besides the point. You're right. The developer himself told me all about how hard it was. I really don't believe making this forum post will have any impact on its success in the industry. Even if it does die without a chance, I could still see a small community of people using it for archival purposes due to its high VQ and compression ratio. I know I'll be using it for my own selfish purposes if they ever release a file format and/or codec for it.

Quote:
The company helped develop the original AVI container but without independent testing I wouldn't put much hope in ZPEG.

Well, that sucks.

I had hoped to get someone on this forum interested in doing some testing of their own, but I'm sure the type of testing you're referring to would be much more official, formal, scientific, academic, professional, and maybe even peer-reviewed.

 

[Richard Paul]



Quote:
Originally Posted by Gooberpatrol66  /t/1491307/anyone-heard-of-zpeg#post_23762992


You don't need a file to measure the compression ratio, right? You can just monitor the amount of data sent over IP? (That's assuming you don't trust what the program's telling you.) At least you can check out the VQ and the CPU usage. But you're right, it's not exactly very useful to the average member of this forum until you can save it as a file.

The ZPEG encoder was made so that it can only send live camera captured video between two computers. That makes it unsuitable for testing and to me that is suspicious. FutureWare has made plenty of claims about ZPEG and yet they have not released an encoder that could be used for testing it.

 

[raymondwestwater]

Hi, this is Raymond of that IEEE presentation. IN the three years that have passed, I have formed a company, changed the direction of the codec, and prepared an interactive online demonstration (which we have just released).
The company is ZPEG Inc, (www zpeg com). The new direction extracts the human visual model from the codec and uses it to pre-process video, making it more compressible. When the processed video is compressed by a motion estimation encoder (H.264, H.265, VP9), the result is visually indistinguishable - but 20% smaller for HD, 30% smaller for UHD.
The demo page includes an interactive demonstration where you can upload your own content (65M limit, though), and it will process and compress the content, and return the results.
We are anxious to hear what you may think!
Thanks,

- Raymond

 

[wuther]

The side-by-side 'demo' is a mp4 file and even though it is a crushed video off the bat the 'zpeg' side is clearly more mushy resulting in loss detail like in the (fake?) foliage and the reds take a noticeable detail drop. Reminds me of the 'special' jpeg 'compressors' that just crushed more.

You will get best results uploading a high-quality file (high bandwidth) and reducing the bandwidth by at least half. There is a file upload limit of 65MB

A 65MB limit??? ROFL. I worked with 100GB files. 'Give us a crushed video and we will show how great zpeg is.'

 

[Worf]

At this point in the game, I think the only true way is to give away the encoder and decoder. The primary reason is unless you're hoping for professional use only, the only reason people will even consider a new codec is if they can test it easily. And if it gets popular, then you can sell a hardware decoder, because a software only decoder that requires a high end PC has limited use and is great for experimentation. Which can then be used to create demand for a hardware decoder for phones, tablets, etc if it is better than the existing codecs.

Because right now, unless a hardware decoder exists, it won't succeed. And the only way to get a hardware decoder is to create demand for it, which can only be done by getting people to use it. And the only way to do that is giving away the codec. Because otherwise people will just use h.264 or h.265.

 

[Joe Bloggs]

At this point in the game, I think the only true way is to give away the encoder and decoder. The primary reason is unless you're hoping for professional use only, the only reason people will even consider a new codec is if they can test it easily. And if it gets popular, then you can sell a hardware decoder, because a software only decoder that requires a high end PC has limited use and is great for experimentation. Which can then be used to create demand for a hardware decoder for phones, tablets, etc if it is better than the existing codecs.

Because right now, unless a hardware decoder exists, it won't succeed. And the only way to get a hardware decoder is to create demand for it, which can only be done by getting people to use it. And the only way to do that is giving away the codec. Because otherwise people will just use h.264 or h.265.

See Raymond's post above. Their new direction for the company is to release a pre-processor, not an encoder/decoder. So the video gets encoded into existing codecs (H264/H265, mpeg2 etc.).