With oneVPL, you can access media functionality on a much wider range of accelerators from integrated graphics abstractions.
For media processing across CPUs, GPUs, and other accelerators, Intel has released a new version of its oneVPL library under the oneAPI Video Processing Library open source project.
Intel’s solution for video encoding, decoding, and processing on Windows, Linux, and GPU and CPU-based platforms is the oneVPL library. Targeting CPUs, Intel Media SDK for older iGPU generations, native oneVPL targets for Tigerlake and newer, and being open to supporting other hardware targets in the future are all possible with the oneVPL dispatcher implementation.
It now implements the oneVPL API 2.8 specification, introduces a new infer sample, supports the XDG Shell protocol on the Weston 10 compositor, and contains a number of bug corrections with the oneVPL 2023.1 release.
A hint is available to turn off external video frame caching for GPU copies, new experimental APIs, support for the 16-bit floating point 4:4:4 ABGR format, a Camera Processing API for camera RAW data, and other changes are included in the oneVPL 2.8 API definition.
