RV9 wins PC Magazine test

[karl_lillevold]

In the big "Play Anything" Media Player test in PC Magazine,
http://www.pcmag.com/article2/0,4149,805233,00.asp

RV9 wins over WMV9 in 6 out of 8 clips with (16-67% more approval), while it is pretty close for the 2 remaining clips (16% more approval for WMV9). ( http://www.pcmag.com/article2/0,4149,806414,00.asp )

Quicktime was also tested, but to quote the article "QuickTime uses the MPEG-4 standard, which is considered good, but it could not compete with Real and Windows Media." Now, 300 kbps, the highest tested bitrate, is pretty low, but keep in mind the resolution was from 360x136 to 320x240 for this bitrate.

For music ( http://www.pcmag.com/article2/0,4149,806392,00.asp ), MP3 and MP3Pro come out on top. RealAudio scores higher than WMA, especially for lower bitrates. Vorbis scores about the same as RealAudio and higher than WMA for low bitrates but much worse for higher bitrates.

RealOne was selected Editor's Choice for anyone who "wants to access a wealth of media" (along with Musicmatch if "your priority is strictly music")

Now I am sure you will ask what about other MPEG-4 encoders than Quicktime. I can not speak for the editors, but the focus was probably more on the player than the formats. How much you dislike RealPlayer you say, well, I guess you can always run Media Player Classic to play RealVideo.

[midiguy]

hey, you (real/helix) and Amir (MS) are really trying to win over this part of the market, eh?

[Teegedeck]

Don't read any computer mag except for c't.

The rest of them is the work of the devil.

:D

[AlphaDivxMovies]

British Computer Shopper is one of the most sober ones and for games the french Joystick is the best i know.
How could Vorbis be even equal to RA? They`ve must be kidding!
Vorbis is IMHO better than MP3 at every bitrate and the only way RA could surpass it would be in either features or at very very low bitrates. I am sorry but this does not make sense. Vorbis is better, i have used it and i changed from MP3 for a reason!

[Sirber]

At witch kbps a Vorbis is equal in quality with a 192kbps MP3 JS?

Look here: http://ekei.com/audio/

A great comparison.

[sam_b]

Sirber, this was pre Vorbis 1 final. Anyone who uses vorbis can tell you that the low-bitrate quality sky-rocketed between the last release candidate and the final. I have heard the difference myself. LAME has moved on since then too quite a bit I believe. I've never heard anyone claim such a difference between FhG and LAME in recent times.

I would suggest that anyone here who is actually interested in codec comparisons of any kind studiously ignore any articles/studies such as these and Make Your Own Test.

[eXistenZ]

Quoted from:
http://www.pcmag.com/article2/0,4149,806392,00.asp
"The tracks were encoded with constant bit-rate settings."
:confused:

IMHO, this is enough to doubt about its worth...

Bye, eXistenZ

[karl_lillevold]

that is the question. In this forum, a video codec is usually compared in VBR mode, which means it is running with a VBR rate control that in most cases is external to the core codec.

However, I would argue that since a good VBR rate control can be applied to almost all modern video codecs improving their performance in a similar manner, one gets a better comparison of the core codec performance when it is run in CBR mode, forcing the codec to use close to a certain number of bits for each frame and each scene.

However, with proprietary rate controls, they are not easily interchangeable in the real world, so of course, the VBR comparison is perfectly valid as well, but then includes another factor besides core codec compression performance.

[sam_b]

karl, I know what you are getting at, but the fact is that most modern codecs are inherently VBR. Rate control begins to be needed for CBR. Your core codec may well have very well integrated frame-size decisions. To suggest that a codec is far poorer for not having an integrated control like this seems unfairly biased.

Also for CBR, averaging periods change significantly, skewing results. A comparison must make note if a codec averages over a large period, which will help it.

A codec which can adaptively change it's setting according to compression required will perform better than one which requires the user to set the codec up (in CBR). Real A/V was designed for streaming from the start, and so it will have lots of cool stuff to allow it to adapt to the source. XviD simply does not really like CBR for the reasons above. Vorbis' CBR is not very good, but is capable of streaming ABR very well.

If this aticle said 'Comparison of various A/V codecs in CBR' or somthing similar, then fine. Vorbis is pants at that. To suggest that the vorbis codec is vastly inferior to everything else at 128kb is laughable.

I appreciate your posting of this article, as it's quite an interesting read, but ultimately this is a forum for DVD-backups, and CBR just plain doesn't matter.

Please don't take this as an accusation of trying to mislead us, it's just that it can happen, and I know you and Amir are normally very specific about CBR/ABR/VBR. It is still relavent when it is your codec that is being given a good rating.

[karl_lillevold]

Originally posted by sam_b
karl, I know what you are getting at, but the fact is that most modern codecs are inherently VBR. Rate control begins to be needed for CBR. Your core codec may well have very well integrated frame-size decisions. To suggest that a codec is far poorer for not having an integrated control like this seems unfairly biased
I apoligize if it sounded like I was saying anything about a codec being poorer for not having an integrated frame-size decision process. With this I assume you mean to change rate control parameters during a single frame, to better match the target frame-size for this frame. As it turns out, even for streaming CBR, this is not very important.

And, you are right in that most video codecs are indeed VBR, even in CBR mode, since CBR is just VBR over a shorter window. If you look at a graph of the bitrate for RealVideo CBR, you will indeed see this - even for CBR, our frame sizes vary quite a bit, but stay within the CBR average over a short window.

For audio, it's different, and I realize VBR / CBR has to be designed into the codec from the start. I can easily understand why for instance Vorbis sounds so much better in VBR mode. In fact, I don't know why PCMag used CBR for their music test, that just sounds wrong, and unfair to Vorbis.

[sam_b]

Cheers for the reply Karl. That is indeed what I meant. It's nice to see someone from a company whose profits come (I assume) primarily from CBR a/v solutions admitting that it is not the only way of doing things.

[tangent]

Originally posted by Sirber
At witch kbps a Vorbis is equal in quality with a 192kbps MP3 JS?

Look here: http://ekei.com/audio/

A great comparison.

You've got to be kidding! This review has been shot down all over hydrogenaudio (http://hydrogenaudio.org). Check out http://sjeng.sourceforge.net/audio/codecs.html for a review of this review.