Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The document IEC 14496-2-05 (Cor1-2005) serves as the official technical corrigendum to ISO/IEC 14496-2:2004, commonly known as MPEG-4 Visual (Part 2 of the MPEG-4 suite). Published in 2005, this corrigendum addresses a number of editorial errors, ambiguous descriptions, and minor technical inconsistencies discovered after the release of the 2004 edition. Its primary purpose is to align the standard text with the intended behavior of MPEG-4 Visual decoders and to eliminate sources of misinterpretation that could lead to interoperability issues between encoders and decoders.
The corrigendum covers corrections in syntax tables, semantics, and decoding processes, including the handling of video object plane (VOP) types, B‑frame referencing, and the decoder complexity model. While no fundamental architectural changes are introduced, these corrections are essential for implementers who seek strict conformance to the MPEG-4 Visual specification.
IEC 14496-2-05 (Cor1-2005) modifies several sections of the parent standard. The following table summarizes the most significant technical corrections and their impact on implementations.
| Section | Original Issue | Correction Applied | Impact on Implementation |
|---|---|---|---|
| 6.2.2 – VOP Coding Type | Ambiguity in the decoding table for vop_coding_type when short video header is used. | Explicit constraints added for the allowed values of vop_coding_type in short video header streams. | Decoders must reject or properly handle invalid coding types; prevents crashes on malformed streams. |
| 7.3.1 – B‑Frame Motion Compensation | Equation for direct mode motion vectors contained an erroneous weighting factor. | Weighting factor corrected from 1/256 to 1/128 in the forward vector computation. | Affects reconstruction of B‑frames; without correction, visual artifacts and mismatched references occur. |
| 7.4.5 – Quantizer Parameter (QP) Range | Clamping condition for QP values during intra VOPs was inconsistently described. | Text clarified that QP must be clamped between 1 and 31 for all VOP types, with a note for baseline profile. | Ensures quantizer stays within valid range; improves conformance for intra‑frame encoding. |
| Annex E – Complexity Model | Example parameters for the video complexity verifier (VCV) contained numerical typographical errors. | Tables E.1–E.3 updated with correct cycle counts and buffer limits for each profile and level. | Correct resource budgeting important for decoder hardware design and conformance testing. |
Adopting IEC 14496-2-05 (Cor1-2005) requires a systematic review of both encoder and decoder reference software. The following points are critical for a successful implementation update:
The corrigendum clarifies the syntax for several header extensions, such as the video_object_layer shape extension. If your parser relies on the original 2004 wording, it may misinterpret certain reserved bits. Update the bitstream parsing logic to match the revised semantics, especially for short header mode.
The correction to B‑frame direct mode motion vectors (Section 7.3.1) directly affects picture quality. Any decoder that has not incorporated this change will produce incorrect predictions for B‑frames encoded with direct mode. Verification tests using the conformance bitstreams provided in the corrigendum are strongly recommended.
The revised complexity model in Annex E corrects the maximum number of macroblock cycles per frame for High and Advanced Simple profiles. Designs targeting these profiles must be rechecked against the updated tables to avoid buffer under‑ or overflow during compliance testing.
For organizations seeking formal compliance with the ISO/IEC 14496-2 standard, adoption of IEC 14496-2-05 (Cor1-2005) is mandatory. Testing laboratories accredited by ISO/IEC 17025 typically require that the corrigendum be applied before conducting conformance tests.
Furthermore, the corrigendum is referenced in subsequent profiles (e.g., Simple Profile Level 3) and is essential for backward compatibility with earlier MPEG‑4 visual streams. The corrections do not affect the bitstream syntax in a way that breaks existing content; rather, they ensure that decoders handle edge cases consistently.