Open-Source Framework for Reduced-Complexity Multi-Rate HEVC Encoding

Adaptive bitrate streaming (ABR) is a key enabler for large-scale video distribution over the Internet. It relies on offering multiple representations of the same content at different bitrates (the "bitrate ladder") at the content generation stage, and on the client ability for the real-time adaptation to varying network conditions by selecting and downloading a representation it can sustain. In turn, this leads to robust video distribution as a trade-off of relatively high computational resources: compared to the traditional cable/IPTV distribution, which typically requires a single SD and a single HD representation, the typical bitrate ladder scheme requires significantly more representations, thereby taking a heavy toll on computational resources. In this work, an x265-based implementation of a multi-resolution encoding framework is presented, which adaptively identifies the most appropriate settings and corresponding analysis information for the efficient encoding with a variety of bitrates and resolutions. The x265 encoder is currently the most popular open-source video encoder, which is based on the HEVC (H.265/MPEG-H) video coding standard, and as a result, the x265-based implementation proposed in this work enables achieving an optimal trade-off between the encoder performance and coding efficiency. According to the extensive experimental results, the proposed x265-based encoding framework achieves up to about 95% computational efficiency improvement, as compared to independent)standalone) encoding per each representation, in terms of the turnaround time, at a negligible visual quality loss.