Getting the most from online video encoding
July 28, 2008It can sometimes feel like there is a dark art at work when it comes to encoding video for the web and getting good visual results.
The visual quality of heavily compressed video has a wide variety of factors at play. The first is of course the bitrate of the compression, how much data is available, but this is not the only factor. How much the image moves is also a major element; the more kinetic the motion of the subject the harder it is to compress well as video codecs rely on grouping the similarity between frames. Fast moving subjects have less consistency between frames and so compress more poorly. As opposed to still subjects like documentary talking heads and dialogue scenes which can compress very well because little changes from frame to frame.
There is another factor that contributes to encoding quality that is often misunderstood or even not widely known; that is encoding blocks.
Contemporary video codecs use fixed blocks to group the image data so the best results come from exporting to frame dimensions that are equally divisible into data blocks the codec uses.
Sound like a head trip? Its really quite simple. Adobe’s FLV format using the On2 codec uses blocks of image data in groups of 16×16, so to get the best image results you should export your video to match these blocks. Adobe have published a table with frame dimensions to best utilize the data blocks in FLV encoded files.
The good folks at FreshDV have also pointed to a handy FLV bitrate calculator for working out before hand the best size for quality ratio.
Let online video encoding be a dark-art no more….
