There are many different factors that contribute to visual quality when encoding compressed video for the web. One factor that is often overlooked has to do with encoding blocks and what dimensions you choose when encoding video. Choosing non-optimal dimensions can result in poor visual image quality and reduced frame rate.
Video codecs use fixed blocks to group image data. Therefore, the best results come from exporting video to frame dimensions that match these blocks.
With Adobe FLV format, the On2 VP6 and Sorenson Spark codecs perform best when the frame width and height use multiples of 16 because these codecs use blocks of image data in groups of 16x16.
For the best image quality and playback, you should always use width and height dimensions that use a multiple of 4 (good), 8 (better), or 16 (best). The following table from Adbobe provides guidance as to the best frame dimensions (both 4:3 and 16:9 aspect ratios) for FLV encoded files.
NOTE: Both the native encoding dimensions and the playback dimensions should be a multiple of 16. For example, if you encode your FLV file to use a frame dimension of 320×240, you should scale to a size such as 512×384.