«Esta claro que los mejores codecs son H264 y WMV»
(esta frase lo dice todo)

Lamento discordar pero el lugar del que el autor sacó el artículo en cuestión incluye comentarios, varios, como éste:

Yes, the article is really poor. There are already so many myths in this video codecs area and this article spreads yet another misinformation. It’s funny that the author tries to teach the readers about codecs when he doesn’t even know what a codec is. He confuses specifications (standards) with implementations (codecs) and like someone already said, for example the reverse-engineered x264 part is ridiculous – x264 is not «H.264’s open source equivalent» as the article says (which again proves that the author has no notion of the most basic concepts and terms – apparently he thinks H.264 is a software product), x264 is an implementation of the H.264 specification. Just like the DivX and Xvid (and FFmpeg MPEG-4, 3ivx etc.) codecs are implementations of MPEG-4 ASP. Codec is a piece of software or hardware that encodes/decodes data to/from some format. MPEG-1, MPEG-2 MPEG-4 ASP, MPEG-4 AVC (H.264) are specifications (standards, formats), the codec authors then read the specifications of these formats and develop software (or hardware) implementations that work with the formats. That’s why when a standard like MPEG-4 is publicly available, the codec authors don’t «reverse-engineer» anything.

Y como éste:

I completely agree!! As an electrical engineer I’ve actually studied and implemented many of these algorithms… The author clearly shows that he has zero understanding of many of the basic concepts and algorithms used in A/V encoding schemes.

At the very least, read wikipedia and try to understand these things. It looks a lot better when you just say that you have no idea what your talking about, rather than BS your way through and get everything wrong.

Firstly, DCT. It’s just a discrete single sided fourier transform. «Take half the number and throw away the remainder» What?! No! In this case, the fourier transform takes data from the spatial domain to the frequency domain. Because much of the information in audio & video is very low frequency, you can get away with only storing the low-frequency data and dropping the high freqs «low-pass filtering». etc,etc

Y uno mas, de entre los que hay:

The «theory» part of this article is very poor. It confuses fundamental concepts such as temporal compression (a generic term), delta compression (what the author apparently calls «temporal compression») and motion compensation (which the author doesn’t seem to understand at all). It treats several concepts present in MPEG-1 (ex., BMC) as «new MPEG-4 features», and does not explain why certain approaches (ex., lossless delta) can’t work with certain types of footage (ex., video).

The description of the DCT is so completely wrong that it’s offensive. You don’t «need a degree in computer science» to understand it (first of all, because Fourier transforms are a field of mathematics, they have nothing to do with computer science), all you need is an IQ above room temperature. The third link at the end of this message does a pretty good job of explaining it.

The codec comparison part of this article is interesting, but one has to wonder how meaningful the comments are, when the author apparently doesn’t understand the fundamental theory behind the main compression algorithms.

It’s really depressing to see articles so full of misconceptions and plain nonsense, when accurate information about these subjects is available for free on the web. I feel sorry for the people who read this and think they’ve gained some sort of insight into these subjects («ah, DCT, yes, I know what that is, it’s dividing numbers by two»… sigh).

I would strongly recommend that people read a more technical (and more accurate!) book or article about compression, before trying to form theories about why codex «X» handles some situations better than codec «Y».

These should get you started:

http://en.wikipedia.org/wiki/MPEG-2

http://en.wikipedia.org/wiki/Video_compression

http://postproduction.digitalmedianet.com/articles/viewarticle.jsp?id=109739

http://postproduction.digitalmedianet.com/articles/viewarticle.jsp?id=125089

Creo que antes de ser tan voluntarioso a la hora de ayudar a otros conviene dedicar un poco mas de tiempo a comprobar lo que se traduce no sea quye el autor original resulte que tiene poca idea de lo que habla.

Yo lo siento que esté en Inglés pero no tengo tiempo de traducir esas cosas y sí me tomo la molestia de avisar de que no estan fiable como pudiera parecer a quien no sabe.

Phil.