Browse Prior Art Database

High Quality High Compression 8-Bit Digital Video

IP.com Disclosure Number: IPCOM000105146D
Original Publication Date: 1993-Jun-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 145K

Publishing Venue

IBM

Related People

Butterfield, DJ: AUTHOR

Abstract

This article describes presenting high quality digital video on an 8 bit playback medium with high compression ratios.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 36% of the total text.

High Quality High Compression 8-Bit Digital Video

      This article describes presenting high quality digital video on
an 8 bit playback medium with high compression ratios.

      Most digital video technologies (such as MPEG) assume and are
optimized for a 24 bit playback medium.  On desktop machines, 8 bit
(e.g., 256 colors selected from a 24 bit or approximately 24 bit
medium) mediums are by far more common.  Playing back 24 bit video on
an 8 bit medium requires "on the fly" run time mapping of 24 bit
video to the 8 bit medium, which typically results in a loss of
quality.

      One commonly applied solution to this problem is to select a
single fixed 256 color palette for the 8 bit medium, and map the 24
bit digital video to this single palette for the entire video (either
at video compilation time or at run time).  This inevitably results
in loss of quality for some sections of video, because no single 8
bit palette is optimum for all 24 bit frames.  Depending on the video
in question, severe degradation can occur, especially if many frames
with finely graduated shades are present (which is common in the two
primary sources of 24 bit digital video, artificially rendered three
dimensional scenes, and captured analog video).

      Another less common solution to this problem (less common
because it is much more cumbersome) is to generate a unique 8 bit
palette for each 24 bit frame, thus allowing the palettes to be
optimized for each frame, resulting in maximum quality playback in
the 8 bit medium.  Currently, this technique is really only
applicable at compilation time, as the process of selecting an
optimum 8 bit palette for a 24 bit frame, and then mapping that frame
into that palette is a computationally intensive one (not suitable
for real time applications in desktop machines without dedicated
hardware).

      Unfortunately, the most common and effective compression
technique used for 8 bit video is "delta compression," a technique
whereby only differences or changes between frames are preserved,
thus greatly reducing the overall quantity of information that needs
to be preserved in a video (since most videos are comprised of a
series of frames which change little from one frame to the next).
This compression technique relies greatly on the coherence from frame
to frame of the underlying 8 bit video data.  Changing the palette
used from one frame to the next essentially changes the "data space"
in which the underlying 8 bit video data exists, and as a result,
eliminates any coherence from one frame to the next of that
underlying data.  This renders the delta compression technique
useless (i.e. it does not compress the data with any efficiency), and
in doing so removes the possibility of using this most effective
compression technique for 8 bit video.

      This quandary of having to choose between either low quality
video with a high compression ratio, or high quality video with a low
compression...