The Pixel Packing Problem: A Quick & (relatively) Painless Introduction to Video Codecs
From the dawn of time man has sought to compress video in the highest possible quality with the lowest possible size. Great mathematical minds from Archimedes to Albert Einstein have all tackled this age-old pixel packing problem. Well, all right, not those guys, but engineers and mathematicians from all over the world have worked out various codecs to store video and audio digitally. Perhaps it’s the simplicity in the term “compression” but it is easy to gloss over the fact that in making your file smaller, the codec is utilizing some tricks and schemes to get the job done. The world of compression is vast, with various companies and individuals trying out different things. As such it can get technical quickly, but knowing the basics can help your end product have the best balance of quality and file size for it’s end use.
Perhaps the best starting point is defining “lossless” and “lossy” codecs. Lossless makes the task easy, those codec preserve the most information, often losing nothing in the process. These codecs are beefier in size with their algorithms allowing the exact file to be extracted. While it’s great to be able to transfer a smaller file, sometimes Lossless is just too much. It’s defiantly a way to go to archive but the compression algorithms can be slow to encode and decode, and jerky playback is never fun. This is where “lossy” comes in.
Lossy codecs are where the engineers and mathematician types get as creative as motionographers. The bulk of the compression work comes from separating the luma (brightness) and chroma (color) information. The human eye is more sensitive to brightness so lossy codecs can dump some color information to reduce size and still look similar to the original. Of course the more information you lose, the more compression artifacts can crop up. These range from ringing (patches of off color pixels around a figure) to banding (abrupt changes in a gradient). A major downside to this route other than immediate quality loss is that it wall also degrade generationally. This is why it’s best to archive a lossless version and output, based on where the video is going to end up.
I think it's a burger, how about you?
So where does this leave you when you need to choose a codec? I have, from the second I set my sights on codecs for an article, been hesitant to recommend any specifically. There are enough factors involved in choosing a codec that it would be impossible for anyone to categorically state anything involving codecs. Not to mention it’s impossible to know what’s coming in the future from the mad scientists of codecs. I will, however, say this: My main output is for the web, and I find H.264 is my most often chosen codec for a final. When I am working internally and sound isn’t needed, I most often go with a .TIFF or .TGA sequence.
While we have barely scratched the surface of codecs, hopefully you have a firm place to stand whenever you solve the riddle of “Which codec is best?” Just remember what you’re outputting for and possible limitations of a client who may not have the most up to date equipment for playback. The bottom line is, you’ve worked hard creating that masterpiece so make sure your settings for the final step are the best they can be.
Jeff Saunders has a Bachelors Degree in Visual Effects and Motion Graphics from the Art Institute. He currently works in Minneapolis, MN as the Visual Art Director for Fitness on Request LLC. He has also over 3 years experience in broadcast television behind the scenes as produciton assistant, floor director, camera operator and has even moved the levels on soundboards.
Very nice overview, Jeff