The following are draft requirements for a potential Enhanced-Performance
Video Coding ("EPVC") video coding standardization project proposed for
standardization as joint collaborative team (JCT) effort of the ITU-T SG 16
Question 6 Video Coding Experts Group (VCEG) and the ISO/IEC JTC 1/SC 29/WG 11
Moving Picture Experts Group (MPEG). The "EPVC" project will develop a new
Recommendation | International Standard (or an extension of ITU-T Rec.
H.264 | ISO/IEC 14496-10, depending on which form of standardization is
determined to be appropriate for the technology design).
- EPVC should be capable of serving the needs of the following
- Digital cinema and large-screen digital imagery.
- Low-delay interactive communication.
- Mobile entertainment.
- Storage-based video applications (camcorders, camera phones, DSLRs,
computer files, disc media, local and network DVRs, download-and-play,
- Video on demand.
2. Compression Capability
- EPVC should be capable of providing a bit rate reduction of
approximately 50% at the same subjective quality compared to
H.264/14496-10 as used in these applications.
- EPVC complexity should allow for feasible implementation within the
constraints of the available technology at the expected time of usage.
- EPVC should be capable of trading off complexity and compression
capability, by enabling one or more operating configurations having
substantially lower complexity than is necessary for implementation of
the full set of design capabilities while retaining a compression
capability substantially beyond that provided by H.264/14496-10 as used
in the applications [Note: The potential need for further detail
regarding this aspect remains an open issue to be resolved.]
"Complexity" refers primarily to decoder computational resource
consumption (in terms of power consumption, computing cycles, memory
capacity, memory bandwidth, etc., including taking into account
opportunities for parallel processing in particular). Anticipated encoder
complexity will also be taken into consideration. Where relevant, the EPVC
encoder should use comparable tools with similar degrees of optimization to
the respective tools in the H.264/14496-10 comparison point.
4. Loss/Error robustness
- EPVC should be capable of robust operation in the presence of packet losses.
For transmission over error-prone networks (wireless, Internet), EPVC
should enable the ability to provide acceptable decoded video quality under
a variety of types of error patterns (burst, independent, uniformly
distributed, etc.). Aspects of the system-level support, encoder
optimization and decoder concealment behaviour necessary to achieve this may
not be within the normative scope of EPVC. Robustness capability similar to
that of H.264/14496-10 High Profile is considered acceptable.
5. End-to-end delay
- EPVC should be capable of operating efficiently in an interactive,
conversational, real-time, low-delay environment.
EPVC should be capable of trading off computational complexity,
compression capability, and loss robustness with delay characteristics.
6. Random access and "Trick mode" support
- EPVC should be capable of providing random access points in the video
bitstream for such functionality as channel switching and program
- Pause, fast forward, normal speed reverse, and fast reverse access to
a stored video bitstream should be enabled.
7. Interface to systems layer
EPVC should be designed to permit efficient adaptation and integration
with a broad variety of system and delivery layers relevant to the intended
applications. Buffer models, including hypothetical reference decoders (HRDs),
should be specified as necessary for target applications.
8. Bit depth and colour sampling
- EPVC shall be capable of a range of bit depths and colour samplings,
ranging from 8-bit 4:2:0 to 12-bit 4:4:4.
9. Resolutions, scanning methods, and frame rates
- The compression capability targets of EPVC are to apply to
progressive scan resolutions of QVGA (320x240) to 8K x 4K at frame rates
of 12.5 Hz and higher.
- Compression capability for progressive scan material outside of the
above range of resolutions and frame rates should be better than
H.264/14496-10 as used in the application.
- Picture formats of arbitrary size should also be supported, within
limits specific to each operating configuration. EPVC shall support at
least the same range of progressive scan picture formats and frame rates
as supported by the H.264/14496-10 syntax.
10. Picture fidelity range
- EPVC should be capable of being operated in a quality range from low
fidelities up to subjectively visually lossless, and support of
mathematically lossless capability is desirable.
11. Hierarchical/layered coding
- EPVC should support extraction of decodable subsets of a coded bit stream.
For example, it should be possible to reduce the frame rate of a coded
sequence by dropping frames without re-coding the sequence.
Effects of hierarchical coding on characteristics such as buffering,
memory, and end-to-end delay must be taken into consideration, and EPVC
should be capable of providing tradeoffs between these.
12. Possible extensions
- EPVC should be capable of coding 3D video.
- EPVC should be capable of bitstream scalability in addition to frame
- EPVC should be capable of monochrome operation.
- EPVC should be capable of multi-channel operation, including alpha
- EPVC should be capable of supporting bit depths beyond 12 bits.
- EPVC should be capable of distributed processing operation.