Unified Streaming Standardization Representative Mohamad Raad attended MPEG 151 online from June 30 to July 4, 2025.

Hosted in person in Daejeon, South Korea, with online participation, the meeting featured two significant topics: the gap analysis for media authenticity standards, and the next generation of video codecs.

Media authenticity is an important topic for MPEG. Members took time to compare what the current set of MPEG standards offers, with what is needed to meet the challenges of identifying trustable content. Then they discussed how to bridge that gap. 

C2PA, or Coalition for Content Provenance and Authenticity, is what the streaming industry currently associates with media authenticity, said Raad. Working groups within MPEG have approached the topic of media authenticity differently.

One such working group, JVET, or Joint Video Experts Team, approved a set of SEI messages which an encoder can produce to carry authenticity information. (SEI, or supplemental enhancement information, is additional data inserted into the bitstream to convey extra information.)

“Some of this work overlaps with C2PA,” said Raad, “and the audio working group has taken a similar approach.”

The systems working group is looking at what gaps exist in MPEG systems standards for the signaling of media authenticity information, and concluded that existing systems standards don’t meet all the requirements for the delivery of media authenticity information. The group started a new provisional work item to address these gaps.

“Some initial steps are being taken to standardize a systems-level technology in this domain,” said Raad.

Video codecs

As far as video codecs go, “these days it’s all about the deployment of AI in different fields,” said Raad. “MPEG has been working on machine learning-based codecs for some time.

“It looks like the next generation of video codecs will mostly be AI-based. It is early days yet. There have been results reported that show the potential compression gains over VVC. There is a call for evidence to formalize the process for collecting evidence that shows that VVC can be improved upon,” he said.

VVC means Versatile Video Coding (VVC), and it is also known as H.266, ISO/IEC 23090-3, and MPEG-I Part 3.

The submission of evidence for this work on VVC is set for October 2025, so that the evidence can be evaluated at the next meeting. 

Trends

When asked about shifts, Raad noted the magnetism of all things AI, a technology that has clearly staked its claim on people’s attention.

Another strong current is media authenticity and how it’s handled. Raad noted that MPEG seems to be taking smaller steps in this field compared to other standards development organizations, which are making a “more significant commitment” to the matter.

If you like niche terms, you may appreciate the theoretical trend toward the use of “Gaussian splatting” for realistic scene creation.

Gaussian splatting is a 3D rendering technique that represents scenes using a collection of 3D Gaussian distributions, or "splats." These splats allow for detailed and realistic 3D reconstructions from images.

But Gaussian splatting is also still in its infancy, said Raad, with workshops and other educational tools being used to increase awareness of the technique.

MPEG 152

Raad enjoyed the conference and looks forward to MPEG 152, which will be held October 7–11, 2025, in Geneva, Switzerland.

At that meeting, Raad expects that the two heavyweight topics again will be media authenticity and next-generation video compression.

“MPEG remains a major forum for the definition of multimedia standards,” he said. “And MPEG experts continue to work on standards that address industry needs.”