Intel’s Video Processing Library “libvpl” is out with a new version ahead of the holidays. The only major change with libvpl 2.16 is adding experimental APIs for AI-assisted video encode functionality.
The Intel libvpl library serves as a dispatcher for interfacing with either the Intel Media SDK or Intel oneVPL Intel VPL GPU Runtime components for the different implementations. The only listed change of tonight’s libvpl 2.16 release is:
“Intel® VPL API 2.16 support, including new AI-assisted encoder APIs and documentation updates”
AI-assisted encoder APIs, of course, piquing my interest. Curious about the new APIs and digging in, there is a new mfxExtAIEncCtrl interface added as an experimental feature for controlling AI-assisted encoder solutions. As part of it the AI-assisted encoder solutions added to the specification are for a saliency encoder solution and ML-based adaptive target usage solution.
No real details beyond that with libvpl just being the dispatch library. From the Intel VPL GPU Runtime I haven’t seen any associated commits yet around the actual implementation, unless these AI-assisted features are being reserved just for some binary-only components. (The only mfxExtAIEncCtrl references within the public vpl-gpu-rt open-source code were the header structures, some dump/trace functions, and assert code.)
Those wanting to check out the Intel libvpl 2.16 release can find it on GitHub.
