Linked by Thom Holwerda on Mon 25th Mar 2013 21:09 UTC
Legal Late last week, Nokia dropped what many consider to be a bomb on the WebM project: a list of patents that VP8 supposedly infringes in the form of an IETF IPR declaration. The list has made the rounds around the web, often reported as proof that VP8 infringes upon Nokia's patents. All this stuff rang a bell. Haven't we been here before? Yup, we have, with another open source codec called Opus. Qualcomm and Huawei made the same claims as Nokia did, but they turned out to be complete bogus. As it turns out, this is standard practice in the dirty business of the patent licensing industry.
Thread beginning with comment 556908
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[13]: Big picture...
by saynte on Thu 28th Mar 2013 08:42 UTC in reply to "RE[12]: Big picture..."
saynte
Member since:
2007-12-10

"Saying someone is wrong is not an attack; and I didn't tell you to shut up, I said that you should stop discussing until you can base it on data, because it's not a useful discussion without data.

I accepted the contrary evidence supplied by Radio, contrary data can exist at the same time, because if you read the details of the IETF data, it's for WebRTC, and uses the baseline profile. This is a legitimate use-case, I said that already. Where do you find bias?


I find bias in your lack of belief that VP8 was competitive with h.264, when by the very measures you use to "support" your bias, VP8 actually performs better than h.264 in about a third of cases, the reverse is true for about a third of cases, and there is very little difference between the two in the remaining third of cases.
"

Wrong, I didn't say it wasn't competitive, I actually said that BOTH were acceptable in a sibling thread!

What I said was that x264 tends to produce better results on objective metrics, and supplied some data to show this for a given profile setting.

How is this worse than what you did: claiming VP8 was better with zero evidence? So it's bias in the case where I provide data, but not bias where you provide none?


The actual truth is that these two codecs by objective measurements (e.g. PSNR, SSIM) have very little difference.

Define "very little"? Some of these metrics are based a logarithmic scale, so it's not straight forward to relate the differences.




http://www2.tkn.tu-berlin.de/research/evalvid/EvalVid/vp8_versus_x2...

Here is a comparison on the metric you wanted, x264 "wins" at least 85% of the time (lower is better).


By subjective measurements I believe that VP8 is the (slightly) preferred codec, but most people cannot really tell the difference.

You're free to believe that.


How is it not bias to rant and rave when someone questions your dubious (and provably false) claim that h.264 was better by every performance measure?

I didn't claim it was better by every performance measure, I just presented evidence that showed x264 was better (for the defined experiment). I don't think there was any ranting or raving involved.

Reply Parent Score: 3

RE[14]: Big picture...
by lemur2 on Thu 28th Mar 2013 09:35 in reply to "RE[13]: Big picture..."
lemur2 Member since:
2007-02-17

Wrong, I didn't say it wasn't competitive, I actually said that BOTH were acceptable in a sibling thread!

What I said was that x264 tends to produce better results on objective metrics, and supplied some data to show this for a given profile setting.

How is this worse than what you did: claiming VP8 was better with zero evidence? So it's bias in the case where I provide data, but not bias where you provide none?


No, it is bias in the case where you say you did not claim h.264 was better when you did, and you also saying that I claimed VP8 was better when I did not. I said multiple times that one could get VP8 to perform as well as h.624 ... claiming "as well as" is not claiming "better".

"The actual truth is that these two codecs by objective measurements (e.g. PSNR, SSIM) have very little difference.

Define "very little"? Some of these metrics are based a logarithmic scale
"

"Very little difference" would be the words of some people who have performed test, and others involved in blind testing of subjective quality.

http://www2.tkn.tu-berlin.de/research/evalvid/EvalVid/vp8_versus_x2...

Here is a comparison on the metric you wanted, x264 "wins" at least 85% of the time (lower is better).


Using codec: Google: VP8 0.9.0-13-g6be1d93 from WebM

http://code.google.com/p/webm/downloads/detail?name=libvpx-0.9.0.zi...

Version 0.9.0 was current May 18, 2010. That would have been the original version. Since then there has been The following releases:

Thursday, October 28, 2010: VP8 Codec SDK "Aylesbury" Release
http://blog.webmproject.org/2010/10/vp8-codec-sdk-aylesbury-release...

Tuesday, March 8, 2011: VP8 Codec SDK "Bali" Released
http://blog.webmproject.org/2011/03/vp8-codec-sdk-bali-released.htm...

Thursday, August 4, 2011: VP8 Codec SDK "Cayuga" Released
http://blog.webmproject.org/2011/08/vp8-codec-sdk-cayuga-released.h...

Friday, January 27, 2012: VP8 Codec SDK "Duclair" Released
http://blog.webmproject.org/2012/01/vp8-codec-sdk-duclair-released....

Friday, May 11, 2012: VP8 Codec SDK "Eider" Released
http://blog.webmproject.org/2012/05/vp8-codec-sdk-eider-released.ht...

Each release has brought improvements over the previous version. We are now quite a way ahead of the original release, and the improvements since have been substantial.


" By subjective measurements I believe that VP8 is the (slightly) preferred codec, but most people cannot really tell the difference.

You're free to believe that.
"

Of course I am. I have good reason to believe it too, since the original VP8 release was thought to be better (subjectively) than h.264 in 15% of cases, and it has been substantially improved five times since then.

" How is it not bias to rant and rave when someone questions your dubious (and provably false) claim that h.264 was better by every performance measure?

I didn't claim it was better by every performance measure, I just presented evidence that showed x264 was better (for the defined experiment). I don't think there was any ranting or raving involved.
"

I think we are simply going to have to differ there.

Edited 2013-03-28 09:41 UTC

Reply Parent Score: 1

RE[15]: Big picture...
by saynte on Thu 28th Mar 2013 12:13 in reply to "RE[14]: Big picture..."
saynte Member since:
2007-12-10

This whole thing started because you stated:


VP8 is not inferior in performance to h.264 except for just one factor: encoding speed. In every other respect VP8 can match or exceed h.264 performance.


Then I showed some relevant data where x264 outperformed VP8.

I'll state why I feel x264 has a higher maximum quality/bit than VP8 (Google's encoder)

- the comparison of x264-baseline and VP8 for WebRTC uses the baseline profile, which was written about 10 years ago. They also didn't ask x264 to optimize for PSNR (what they measured), slanting the results away from x264. The test was done by a Googler (apparently) and the methodology ripped apart on the x264 mailing list http://mailman.videolan.org/pipermail/x264-devel/2013-March/009913....

- the study done at MSU which puts x264 above VP8 on SSIM at least on the high profile setting (I think it does well/better on the others as well, would have to check).

- the study from TUB which also puts x264 above VP8 on VQM, although it is from the initial public release of VP8 (which still had some years of development within On2).

- using recent versions of both encoders, this screenshot comparison https://gist.github.com/Hupotronic/4645784, single screenshots such for comparison, but c'est la vie.

- another comparison on various video clips measure PSNR and SSIM http://blog.existentialize.com/tag/vp8.html

- the opinion of an expert on the topic, the author behind x264 and also a vp8 encoder

Reply Parent Score: 3