In March 2023, Bitmovin announced the launch of their “Smart Chunking” feature, an improvement and an extension of their “Split & Stitch” transcoding logic.
OTTVerse caught up with Markus Hafellner, Senior Product Manager, VOD Encoding, at NAB 2023 to understand more about Smart Chunking.
Hi Markus, thank you for sitting down for this interview. We were intrigued by the announcement of Smart Chunking and would like to learn more about it from you.
To get started, can you explain Smart Chunking to our readers and how it builds on Bitmovin’s “split and stitch” algorithm?
Smart Chunking is an important update to our VOD Encoding solution and evolution of the split and stitch algorithm. If you are unfamiliar with the split and stitch algorithm, it splits the encoding job into multiple parallel encodings or segments, accelerating the entire process.
However, any provider doing split and stitch also faces the downside of potential quality drops when using fixed GOPs and segments, degrading the overall visual quality for viewers.
Smart Chunking solves these issues by optimizing chunk lengths and bitrate distribution, which improves visual quality throughout the whole asset at a faster pace than before.
How does Smart Chunking impact quality, bandwidth usage, and video delivery costs?
Smart Chunking provides better quality at the same bitrate.
This can be used in two ways,
- increase the quality at the same bandwidth and delivery cost, or,
- reduce bandwidth and delivery cost at the same quality.
Video quality is measured currently with VMAF, an objective quality metric created by Netflix, and it’s one of the most widely used metrics in the video streaming industry to benchmark video quality.
We used VMAF when benchmarking the image quality of a highly complex video asset processed with the standard split and stitch approach compared to Smart Chunking.
The results showed the lowest quality 1% frames increased by 6 VMAF points (6 VMAF is the exact noticeable difference for the human eye). There was also an impressive 22 VMAF points increase in the lowest quality 0.1% frames, and the worst frame saw an increase of an astonishing 60 VMAF points.
These are the main benefits of Smart Chunking, which speaks to our customers’ ambitions of optimizing the quality of experience while simultaneously reducing bandwidth and video delivery costs.
Does Smart Chunking only affect transcoding, or does it also impact segment durations used for HLS & DASH?
Segment durations used for HLS and DASH are kept as specified by our customers. This ensures playback compatibility stays the same.
Segment duration is part of the muxing definition, while chunk length is part of the logic used in the horizontal scaling of transcoding with split and stitch.
In certain codec implementations, we’ve noticed “beating” artifacts around IDR frames (where the quality abruptly changes at GOP boundaries). Does Smart Chunking help mitigate this issue?
Yes, one of the main benefits of Smart Chunking is that those kinds of artifacts are reduced. Smart Chunking helps to create a more uniform quality distribution across the asset, reducing low frames, which mostly happen at the end of a closed GOP.
Smart Chunking leads to different chunk sizes for different movies. How does it work alongside content-aware or per-title encoding, and does it impact the performance?
The content type alone does not determine chunk duration. Also, codec and codec configurations have an impact on chunk duration selection.
It is fully compatible with other advanced video optimization techniques, and the positive effect on performance image quality and optimizing bitrate distribution at a faster pace remains.
How does Smart Chunking affect the transcoding throughput of your existing workflows? Does it require an n-pass algorithm or several seconds of look-ahead to assess video quality?
Smart Chunking helps accelerate turnaround times, especially for long-form content. We are still optimizing it for short-form content.
Are there any enhancements planned for Smart Chunking? Livestreaming, for example.
We will continue to develop and optimize Smart Chunking for more use cases later this year. We don’t expect to utilize it for Live encoding at the moment.
That brings us to the end of this interview. Thank you, Markus Hafellner, for answering our questions and educating our readers about your latest innovation.
Krishna Rao Vijayanagar
Krishna Rao Vijayanagar, Ph.D. is the Editor-in-Chief of OTTVerse, a news portal covering technological and business news in the OTT space. With extensive experience in video compression, ABR streaming, video analytics, monetization, and more, Krishna has held multiple roles in R&D, Engineering, and Product ownership at companies such as Harmonic Inc., MediaMelon, and Airtel Digital. Krishna has published numerous articles and research papers on the latest trends in OTT and frequently speaks at industry events to share his insights and perspectives on the fundamentals and the future of OTT streaming.