Typically, MainConcept sits on the receiving end of specifications. Our mission is to convert industry standards into commercial-grade software modules, which can easily be integrated into applications and workflows used by professionals across multiple verticals. So, it was a very welcome change to be invited to the “Workshop on Future Video Coding” at ITU headquarters in Geneva on January 17th, 2025.
This workshop was embedded into the ITU SG21 meeting which was held to go into detail on next generation video applications and set the stage for upcoming standards. In three sessions, panelists were invited to provide industry input which helps shape the requirements for future video codecs.
In addition to attending, I was invited to speak during the first panel, entitled “Requirements and Use Cases: Voices from Industry and Users”. Alongside representatives from Samsung, Amazon and China Mobile, and under moderation of Prof. Dr-Ing. Jörn Ostermann (Leibnitz University, Hannover) and Yuan Zhang (China Telecom Research Institute), we highlighted challenges and opportunities for future standardization.
All panelists agreed that keeping the focus mostly on an “additional 30-50% bitrate reduction” might not be the right Mantra for “MPEG-next” anymore. Instead, the panelists suggested attention be on detail improvements, subtle changes to optimize encoding at medium resolutions, and keeping the complexity of a next-gen codec under control.
With VVC/H.266 just growing out of the “initial implementation” phase and entering a mode in which practical applications for the codec are sought and evaluated, it seemed very important to point out that any next generation codec will have an initial ramp up of at least 10 years until it achieves meaningful market adoption. So, anything discussed on this particular Friday in January 2025 will have a commercial impact some time in 2035.
Also read: The Codec Lifecycle
Under this light, the following discussions which included the use of Machine Learning for future video compression complemented the discussion and set a different focus on the media landscape we will see in the next decade.
The second session, “Towards H.267/MPEG-Next: Current Status in JVET and Future Time Horizon”, focused on research results which have been achieved since the standardization of VVC/H.266. The third session “Practicalities: Hardware Capabilities and Software Implementations” with presentations from MediaKind, Qualcomm and Spin Digital, picked up right at the point where the first session ended. All panelists presented lessons learnt from real-world codec implementations, which echoed and deepened the sentiment that was expressed in the first session.
Given the unexpected level of agreement across panelists and even panels, we can assume that “MPEG-Next” will likely have a different goal than previous MPEG-based codec generations. What will that be? Nobody knows today, but I am excited to play a role in its early definition.
It was a great pleasure speaking in front of 200+ people in the room (with 300+ additional people online) in the honorable halls of ITU and it was a great opportunity for MainConcept to raise a “well-grounded industry voice” (as one of the participants said afterwards) to this great audience.
What are your thoughts on this weighty topic? Send me an email if you have an opinion: nicolai.otto@mainconcept.com.
You can view the full Friday afternoon session on demand here.
Images courtesy of ITU Pictures