ExtremeTech Wages War of the Codecs 356
prostoalex writes "ExtremeTech tested Windows Media, DivX, QuickTime/Sorenson and QuickTime/MPEG4 codecs. They encoded clips from Matrix Reloaded, Monsters, Inc., X2 and Spider-Man. QuickTime/Sorenson won the encoding speed contest, for the quality tests read the entire review, as each movie sample was encoded with 500KB and 1MB bitrates. Video samples provided on the site as well, so see for yourself."
But no Xvid? (Score:5, Interesting)
Re:These encoding algorithms... (Score:3, Interesting)
Made on a Mac? (Score:3, Interesting)
M$ does something right (Score:1, Interesting)
Re:stills vs. motion... (Score:5, Interesting)
Xvid rules the scene (Score:4, Interesting)
Ogg Vorbis audio (Score:1, Interesting)
What really matters. (Score:4, Interesting)
Give it some time (Score:4, Interesting)
Regardless, Apple has been one of the biggest supporters of MPEG-4, and I thank them for that.
Re:Terrible reporting - used wrong programs to enc (Score:3, Interesting)
Several codecs may be used to produce a single movie trailer, with different codecs being employed where their relative strengths are required: low motion versus action versus bright scenes versus dark scenes.
These guys are WAY more sophisticated in their technique than any home user will ever be.
Lesson: Admire Apple's movie trailers but don't think you're going to reproduce their quality.
--Richard
I call shenanigans (Score:5, Interesting)
Very Strange Results -- Artifact of Methodology? (Score:3, Interesting)
What's particularly suspicious is that Apple's MPEG-4 came out so poorly, though WMV9, and DivX are nothing more than early MPEG-4 codecs. Sorensen3 is the only substantially different algorithm used. And why use MPEG-4? It was originally designed for low-resolution low-bitrate applications (PDAs, cell-phones, etc.)
Why so slow? I do most of my video transcoding under Linux, but they aren't getting much better throughput than I do, and their machine's at least 4 times as fast as mine? I suppose it's got to do with using Indeo (my source is DV), so there's an extra decode step, but it's still quite slow.
I've distributed a number of my own videos in the MPEG-4 format, and don't see the sort of horrible results they demonstrated in their examples -- but then again, perhaps I do preprocessing (quantization, denoising, etc.) that they don't include in their process.
Regardless, my personal experience is that at high or low bitrates, most of the codecs are interchangable. Perhaps you need to fiddle with the encoding parameters, but you can almost always get results close enough to identical as not to matter. It becomes more difficult with mid-range bitrates (2-3Mbps@720x480x29.97) that some codecs show strengths over the others. In that department, I almost always go with MPEG-2 with custom quantization matrices...
Well it might be licensing (Score:5, Interesting)
Limited value (Score:5, Interesting)
Re:What's up with MPEG4? (Score:3, Interesting)
I've never seen it used in practice. The 3ivx [3ivx.com] codec, for example, adjusts it's post processing according to the available CPU which seems more sensible.
Re:What's up with MPEG4? (Score:5, Interesting)
Referring about Sorenson, keep in mind this is the FREE codec that comes with Quicktime Pro. Professionals use a several hundred dollar 'Developers Edition' with Cleaner (two pass VBR encoding, which makes Sorenson rock). I know this is a for-user comparison, but in the professional scene, Sorenson can be even better then third/second place in quality.
All fail miserably at making good encoding easy (Score:5, Interesting)
Real Media (Score:2, Interesting)
I mean I can fit 5 seasons of Daria on a cd and it's watchable. And there is tonnes of old anime kicking around in it.
Re:Fair and yet unfair comparison (Score:3, Interesting)
I would also be curious to see a comparison of codecs based on using 1:5 compression consumer level DV souce material.
Re:Not the best evidence. (Score:3, Interesting)
Actually, considering that WMV9 supposedly uses the MPEG4 AVC (while the rest use the ASP), I would say that by more-or-less tying with DivX, MS lost, big-time. It took twice as long (damned good, for an AVC implementation) to produce comparable results (damned bad, for an AVC implementation). I had seriously expected WMV9 to utterly crush the competition in this test, before reading the article, yet it did not.
This may mean nothing to someone trolling as an AC, but to those of us likely to actually use one of these codecs based on their technological merits, this should send up a few huge red flags regarding WMV9. If MS had an actually decent AVC encoder, I'd use it in a heartbeat (though only because, that I know of, no other real (as in non-pre-alpha) and almost-free (as in beer) implementations of the AVC profile exist)... AVC has the potential to yield the same quality at a quarter of the bitrate of the ASP. To only perform comparably... Poor showing indeed.
So complain that I just want to cry about MS winning, but if you believe that, you clearly don't understand the encoding methods involved here.
However, as I said in my original post, this test has absolutely no validity. I would like to see a fair test, for the exact reason I mention here (ie, does MS's AVC really perform that poorly?), but the present comparison can at best make me raise an eyebrow at such an unexpected result.
me neither (Score:3, Interesting)
http://xinehq.de/ [xinehq.de]
Re:But no Xvid? (Score:4, Interesting)
I do WMV streaming at work- and it works great. Previously, right here, on Slashdot of all places, when I mentioned it, people would tell me over and over that the quality sucked.
So, if they really cared what the slashdot kiddies thought, they would have have done something to skew the results. But this didn't happen.
They did mention that the encoding took far longer for Windows Media, and that is true. (But their hardware was crap for media encoding- a single processor? If you are doing video encoding, you probably have a lot more hardware than that to throw at the problem). But when it comes to ease of hosting, and getting the users to actually view the thing, nothing beats Windows Media 9.
I did try Divx for a while- and after the 9,000th complaint in about 2 days, I finally relented, and put it up in a
Josh and Trish America want the video to play with the click of a link- which generally means Quicktime or Windows Media. I'll stick with Windows Media.
Honestly- very few of the people here on Slashdot want to watch the movies I serve up- but your parents do. Now do you really want to explain to them how to play an Xvid file?
Re:Becauses it's a Mac of course! (Score:4, Interesting)
Re:Of course people care (Score:2, Interesting)
If you want to keep the grain only because you don't want to lose detail, I recommend you take a look at Avisynth's undot and pixiedust filters. These do wonders on removing the grain and keeping detail- the output of them usually looks better than the dvd itself, and compresses much easier.
The only tradeoff is, pixiedust is slow as hell. Process everything to a huffyuv avi then do a two-pass of xvid.
Re:But no Xvid? (Score:4, Interesting)
Maybe that's why they didn't review it.
Re:But no Xvid? (Score:3, Interesting)