[Spliff]

http://www.extremetech.com/print_art...=121163,00.asp

That's depressing, especially considering that Windows Media Player for Mac is a piece of junk.

At least we have DivX and Sorenson, but that won't help when Microsoft's codec finally wins out.

[voodoo]

1Mbit per sec or about 128 KByte per sec.

The QT and MPEG4 codecs were by far the fastest. I suspect the reviewers didn't quite know how to encode..

[Spliff]

Originally posted by voodoo:

1Mbit per sec or about 128 KByte per sec.

The QT and MPEG4 codecs were by far the fastest. I suspect the reviewers didn't quite know how to encode..

What we need is some other sites to repeat the tests and see if they get the same results.

There's a discussion at Slashdot.org about the article now. Here's one comment:

Limited value
by Dr.Knackerator on Thursday March 11, @01:15PM

Does it tell you which codec is best? Maybe but only for recompressing MPEG-2 footage. They *should* have tested against DV output as the standard consumer format, and uncompressed video. Plus looking at snapshots of compressed movies is of limited value, there is a big difference between what detail we can determine through a still image and a moving one. If you were to freeze a tv picture (or look at a captured frame) which includes something moving you would see a combing effect of the interlaced video. It doesn't look like that when you view it though.

[Powaqqatsi]

Seriously, when I encode stuff into QT/MPEG4 it doens't look that shitty at all

[Big Mac]

Isn't extremetech that rabidly anti-Mac rag? I would think any source with the word extreme in its title is going to be biased against the Mac.

[Thinine]

They should have also tested the 3ivx encoder. It's one of the highest quality encoders in existence.

[Horsepoo!!!]

QuickTime *is* getting old. If Apple keeps doing the bare minimum in regards to QT updates, they'll lose. Apple will have to revamp or rewrite the damn thing soon and provide higher quality MPEG4 codecs than their current one because WM9 is wiping the floor in many areas.

The framework itself is very flexible and functional but it's a huge bloated mess.

The QuickTime Player in particular is as old as it gets. It essentially hasn't changed since QT4 when we first had a glimpse of a brushed-metal interface. Apple has really neglected this one...and possibly right up there with the Bad-Carbon-Ports™...close 2nd behind AppleWorks.

I can only hope a they have been slowly rewriting QT while simultaneously upgrading the current one. If we're still stuck with bad codecs and bad carbon ports beyond 10.4, I think it's over for QT.

[Uncle Skeleton]

a few thoughts:

• Apple makes computers. On the side they make a kick-ass operating system. On the side of that they make QuickTime which was once upon a time the unrivalled leader of the computer video industry in every imaginable way, but at this point hasn't really seen a major update in over 5 years. The QuickTime group makes a video codec that is fully standards compliant and still comparable to the latest cutting edge codecs. Who ever expected the best codec ever to come from this pedigree? Looking at the other side, there's Divx.com, who spends all day every day working on video compression. Of course they're going to be better at it. But it's not cheaper (you don't need to buy QTPro to encode with Apple's MPEG-4 codec, but you do need to buy Divx Pro to encode with the features that make it what it is), and it's not officially compliant with anything but itself (and their new policy is to diverge from standards compliancy over time).

• Whoever did this test was kind of an idiot. They didn't crop the black borders, which spoils some of the compression algorithms of all modern codecs. They converted from MPEG-2 (lossy) to uncompressed (lossless) to Indeo5 (lossy) to the distribution format (lossy). Stupid (even though a human observer might not see the artifacts introduced by multiple lossy conversions, the encoder at the end can see them and tries to preserve them as if they were part of the source, because now they are). And at the end they used subjective opinion to judge the difference (and probably not a blind trial), instead of some kind of scale like PSNR.

• Of course, they may be exaggerating on the degree, but Apple's codec really isn't as good as DivX or WMV3 when you're talking just about compression quality and not interoperability or price.

[eklipse]

Where's 3ivx?

In my experience it spanks all 4 of the tested codecs in terms of speed and quality.

[dru]

Originally posted by Spliff:
http://www.extremetech.com/print_art...=121163,00.asp

That's depressing, especially considering that Windows Media Player for Mac is a piece of junk.

At least we have DivX and Sorenson, but that won't help when Microsoft's codec finally wins out.

Where are the full motion clips? Did anyone notice the frame in WMP Matrix bench shot is not the same frame as the other three codecs? This tested re-encoding from MPEG-2 which is common for folks who want to drag films around with them but that isn't necessarily going to give you the results comparable to encoding uncompressed DV.

Who is surprised that taking TWICE as long to encode is going to improve output quality dramatically? This is a crap "test" designed to slag QuickTime.

I'd like to see this repeated elsewhere. I typically use DiVX or 3vix, fwiw.

[Graymalkin]

Rather, this is an article for those who want to compress video for home use: to e-mail to family members, put up on a Web site, burn onto a CD, re-compress to fit on a PDA or portable video player, or just archive for later use. Our focus is on middle-of-the-road bitrates suitable for download or CD archives, not extremely low bitrates for streaming over the Internet or very high bitrates for DVD-ROM based packaged media.

It is very ridiculous to have your article's premise be testing codecs for personal use - sending video in e-mail or putting video on a website - and then testing with source material that has already been compressed and mastered for commercial distribution. This is top notch crap reporting from Ziff-Davis.

The first major problem is the source video set up. Ripping from MPEG-2 to uncompressed video to Indeo not only adds a whole new level or artifacts but isn't exactly something you would do when sending videos to grandma. There's no reason why they couldn't rustle up enough money to rent a good DV camcorder for an afternoon and shoot some video. They could have hired someone to shoot it for them that would know what they were doing and properly light and expose the video. A baby's first steps are more likely to be DV video clips rather than MPEG-2 DVD rips, unless you're sending grandma video of someone else's baby.

The second major flaw is the presence of the horizontal bars in the video frames. Having the frames in the video really messes with the encoding algorithms. The black bars will make even a production quality encoder give craptacular output. The disclaimer that you shouldn't judge the codec by still shots should make everyone blink. If still shots don't do the codec justice, why even use them?

The reporting on the encoding time is likewise pretty absurd. The author was more concerned with overall encoding time than he was with the encoding rate. A more meaningful speed measurement would have been the average fps during the encoding process. A video clip of x number of seconds will have about x * 30 frames. At a particular fps rate it is pretty simple to figure how long the encoding will take. A measurement of the total encoding time is useful for a quick reference but isn't necessarily indicitive of the codec's overall speed.

What this article reads like is a guide to the best free/cheap encoder to rip DVDs to distribute them via P2P or some other means. The test cases were not congruent with the alleged intentions of the article. Grandma probably isn't interested in rips from SpiderMan and the Matrix but 31337d00d666@aol.com probably is. Ah, the drivel that is Ziff-Davis.

[Axel]

Originally posted by Big Mac:
Isn't extremetech that rabidly anti-Mac rag? I would think any source with the word extreme in its title is going to be biased against the Mac.

All their tests were made on one single PC. They don't mention Macs anywhere in the article.
Remember, Quicktime != Mac.

Besides, judging from the results of their tests, you really don't need to be biased against the mac to admit Quicktime's MPEG4 codec performed poorly compared to the others.

[crayz]

Graymalkin: Nice post. Good summary of the problems with the article. I still think Apple MPEG-4 is crap though. Xvid is my current fave

[Graymalkin]

Apple's MPEG-4 encoder is really disappointing. It works well enough to work but doesn't quite work well enough to do really nice work. I tend to believe that the encoder is less than stellar in order to keep the market open for Apple's friends at Sorenson and Discrete and even their own Compressor product in FCP. If Quicktime encoded high quality video for $30 there's be little incentive for even prosumer video folks to pay centibucks for Squeeze, FCP, or Cleaner.

[alex_kac]

First, remember that they didn't use Sorenson Pro. Only the built in Sorenson. Second, this wasn't a test on QuickTime at all...only the Sorenson codec. They could have done it on AVI Sorenson and it would've been exactly the same.

The fact is - Apple ships with no codec equal to WMP9 or DivX. But its not Quicktime's fault the quality was bad, it was the codec they chose.

[awaspaas]

I call shenanigans (Score:5, Interesting)
by awaspaas (663879) on Thursday March 11, @04:07PM (#8536167)

Okay, I just encoded some DVD-size video at 1mbit and 500kbit, 1/4 size, in QuickTime MPEG-4 and can barely see any artifacts in either. This dude seriously got some settings wrong in his MPEG-4 encodings, although I don't quite see how that's possible as settings are quality, framerate, keyframes, and data rate (and he said quality was set at best). I'll post some screenies later if I get a chance.

[lenox]

They really should have posted clips - really short clips of the 1mbps encodes, perhaps longer clips of the quarter-size ones. Does the article even mention how the different codecs use key frames and then compress the motion between them to fill in the blanks? Does the author even attempt to avoid accidentally taking a 'still' of one of these key frames?

Stills will never do a motion video codec justice.

[Laurence]

First off, the testing methods were terrible, however, that does not really matter. The reason a majority of people use these codecs is to rip DVDs and share them with friends. Starting from a m2v source is justifiable in this regard. I have done this myself. Rip a DVD to MPEG2, and then encode to fit on a CD. For this purpose, Apple's implementation of the MPEG4 standard is complete and utter crap. 3ivX, DivX, Xvid, openDivX, etc are all better. I have never done any of this type of thing on a PC, but I would assume from the tests that WMV is probably much better as well.

Apple really needs to update the MPEG4 encoding for QT7 so that 2 hours of video can fit in 700MB and look at least as good as DivX 5.1. If they fail to do this QT will lose out to the others in the next few years.

I'm not saying QT will disappear, but it will be relegated to a professional, behind the scenes, technology to be used for video editing applications in the high-end sector.

I have a Lite-on LVD-2001 DVD player which plays, MPEG1,2,4, DivX, XVid, etc, and I have done many tests. MPEG4 output from QT burned onto a DVD and played back on a nice widescreen TV looks terrible at 1000kbps, DivX, XVid, etc all look great, from 10-12 feet away I can't distinguish the output from a commercial DVD.

I realize that Apple doesn't want to support piracy of movies, and that is exactly what would happen if they improved their codec to a point where it was competative, but that is exactly what they must do to survive in the consumer space. This is only going to become more and more of an issue in the future.

[Turias]

Originally posted by Laurence:
Starting from a m2v source is justifiable in this regard.

But before trying out any of the real codecs, they took the original video and encoded it using the Indeo 5 format, making their entire test worthless! If they had gone from DVD directly to the tested formats, that would have been fine, but they took video, encoded it using a lossy method, and the re-encoded again with another lossy codec!

It's like encoding something into mp3, and then taking the mp3 and encoding that into aac and ogg to test out those two codecs.


Their entire paper is crap. Worthless.

[Hydra]

I read that comparison page and experimented with a few things. I use the latest versions of my apps ( QT, FCP 4.1.1 and Compressor 1.1) and the one thing that struck me as very very odd is that when I export from FCP to QT and just select the standard MPEG4 settings for a 320x200 video file the file looks great. Once I deviate from those standard settings and "tweak" to my liking the quality drops off noticeably (even if I go with higher bitrates and subsequent file size - which should NEVER happen). I fired up Compressor and figured with it's flexibility I could beat the look of the standard QT setting but nope. The standard settings ALWAYS look better. What gives? I am using a mix of video formats from DV, MPEG2, VGA rez screen capture movies and titles generated in LiveType all for variety and the results all are the same. The standard sets ALWAYS look better (very very good in fact). Just wondering what gives.

-Jerry C.