On2 VP6 Technical Data & Applications
Aug 25th, 2007 by Leo A. Geis
H.263, a progressive stage of .mpg-1 and .mpg-2, was released in Flash MX. H.263 itself was further improved only to be superseded by H.264. Some relevant site links are:
- MPEG, the Motion Pictures Expert Group
- ITU-T, the International Telecommunication Union (Telecommunication Standardization Sector). The VCEG (Video Coding Experts Group) operates within ITU-T under the title of “Question 6, Study Group 16.” I’m not making this up
- SPIE, the International Society for Optical Engineering (the acronym survived a few name changes).
Contrary to popular belief, On2 VP6 does not use H.264: The H.264 available within Flash is the Sorenson option. H.264 is tremendously sophisticated, to be sure, and includes the following features:
- Variable block sizes. You’ll see this manifest in comparative frame captures. Very smart!
- A further-refined deblocking filter (H.264 is still a Discrete Cosine Transform/DCT compression)
- .25 Pixel accuracy. This is an issue when displaying at 72 or 96 dpi, antialiasing, etc. .05 Pixels is termed a “Twip” and is the smallest unit of spatial metering in Flash (yes, inherited from Typography)
- Many wonderful acronyms such as CABAC, SEI, and CAVLC. Acronyms are chick magnets. Everyone knows that.
Here is an On2 White Paper on VP6 that is more technically satisfying than that posted in my previous VP6 piece. A number of very interesting points from it, and other resources on the On2 site:
- Cursory comparisons certainly do favor VP6 over H.264, though both are either being or have been superseded: VP7 is here, H.265 is expected around 2009. There may be a lesson in this about private enterprise vs. international consortiums and their relative efficiencies
- VP6 is for Flash Player 8 and 9 only. With Adobe’s Version Checking, it’s nearly a moot point in North America, Europe, Oz and parts of Asia
- On2 VP6.2 contains a relatively rudimentary sharpening capability
- Flix adds Vector Video and Chroma-Keying to .flv encoding in the IDE and the stand-alone encoder
- “D1″ is a television standard that is rather loose, but can be approximated to NTSC or PAL DVD-V standards.
On2′s White Paper on VP7. For your convenience:
- Single or double-pass encoding
- A 4×4 DCT
- The White Paper compares VP6 and VP7 animation frames
- The White Paper references SpongeBob SquarePants
- I didn’t know Nero Digital H.264 had a “Psycho-Visual Enhancement” control that sets to “High.” I really cannot wait to run across that in a trial.
So, what does it all mean?
In the context of increasing bandwidth and processing capabilities of mobile devices, it means an awful lot for folks who want to communicate within ubiquitous, interactive, contemporary (here meaning “immediate” or “real time”), defensible formats:
- Very few compression codecs are surviving the cut. Fewer and more refined codecs will evolve, easing device and application development
- Applications demanding accuracy and administrative overhead (such as archiving) will enter their adolescence
- Remote medical diagnosis and management, time-sensitive (e.g. mealtime) and GPS/position-sensitive (e.g. rush hour) marketing, and personal communications will benefit beyond my capacity to imagine.
L


