Technical specifications
APV encodes each frame as a self-contained still image, similarly to Motion JPEG and Apple ProRes. Compared to distribution-oriented codecs such as H.265 and AV1, this approach allows faster encoding and decoding, especially on multi-core CPUs where individual frames can be processed in parallel, but at the cost of compression ratio.
Each frame is divided into macroblocks. Each macroblock is encoded independently and the bitstream includes the offset for each. This reduces compression ratio but enables parallel encoding and decoding within a single frame, improving decoding speed on multi-core CPUs.
Each macroblock is further subdivided into 8x8 blocks, which are encoded using DCT followed by entropy coding, similarly to Motion JPEG and Apple ProRes.
Intra prediction occurs only within a macroblock to allow for parallel processing of macroblocks. Unconventionally for video codecs, intra prediction operates on DC coefficients instead of pixels, which reduces computational complexity.
APV also uses a deblocking filter to reduce artifacts at DC block boundaries.
Samsung claims that APV achieves a 20% improvement in compression at the same visual quality compared to an undisclosed professional video codec implementation. They also claim that APV allows for higher quality encoding compared to the I-frame-only mode of H.265 on smartphones, where the computational complexity of H.265 makes it impossible to achieve high bitrates. However, this quality improvement comes at the cost of much larger file size. [2]
This page is based on this
Wikipedia article Text is available under the
CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.
