Panasonic AJ-HPX2700 - Manual (Page 9)

8
A V C - I N T R A T e C H N O L O g y
Intra-Frame (I-Only) Compression Superiority
Motion-image compression can be divided roughly into two
methods: I-Only compression, which completes all processing
within each frame, and Long GOP compression, which processes
across multiple frames. AVC-Intra and DVCPRO HD use I-Only
compression, while HDV uses Long GOP compression.
The MPEG-4 AVC/H.264 standard encompasses both methods.
When the images of adjacent frames are similar, Long GOP
compression achieves an advantageously low bit rate. However,
this trait is not often seen in broadcasts like flash-filled press
conferences, fast-action sports, and music shows with confetti
and electronic displays.
Also, because processing is performed frame-by-frame in I-Only,
new-generation multi-core CPUs offer high-speed parallel
processing. This makes I-Only compression more suitable for
Non-linear editing than Long GOP, for which parallel processing is
difficult due to its inter-frame dependence.
With its I-Only compression, AVC-Intra produces remarkably
stable images that are unaffected by adjacent frames, and meets
professional needs in virtually all situations and workflows.
Twice the Compression Efficiency of MPEG-2
By selecting the most effective compression techniques from
among those in compliance with the H.264 standard, AVC-Intra
has doubled the compression efficiently over earlier Intra frame
compression schemes. Its intraframe predictive and context-
adaptive entropy coding are particularly effective methods for
boosting compression efficiency.
Intraframe predictive coding (intra prediction)
This process generates predictive images based on adjacent
blocks of 8 x 8 pixels. Selecting the most suitable predictive
mode from among nine luminance signal modes (see illustration)
and four color signal modes, it generates accurate predictive
images. The residual data (obtained by subtracting a predictive
image from the original input image) is recorded together with the
predictive image. Because the prediction accuracy is high,
there's minimal residual data, and thus high compression is
achieved. This process is conducted within the frame, so
prediction accuracy remains high even with fast-motion images.
Context-adaptive entropy coding
The entropy coding process used in MPEG-4 AVC/H.264 utilizes
CAVLC (Context Adaptive VLC) and CAVBAC (Context Adaptive
Binary Arithmetic Coding), both of which are context adaptive.
MPEG-2 uses a fixed table when performing the VLC coding,
with the result that compression efficiency is low with some types
of images. In context-adaptive coding, on the other hand,
operation varies with different kinds of images and high
compression efficiency is maintained at all times.
Conceptual Illustration of Luminance Signal (Y) Intra Prediction Mode
This process generates predictive images based on adjacent blocks of 8 x 8
pixels. Selecting the most suitable predictive mode from among nine luminance
signal modes (see illustration) and four color signal modes, it generates
accurate predictive images.
Left: Original image Center: Intra-frame predictive image Right: Difference image obtained from subtracting the intra-frame predictive image from the original image.
This shows the high accuracy of intra prediction.
Sample Images of Intraframe Preduction
mode 0 mode 1 mode 2 mode 3
mode 4 mode 5 mode 6 mode 7
mode 8
Copy of below pixels
Predictive pixels
Adjacent pixels
Average