Movit 1.6 Released For GPU-Based Video Filter Library
Movit, the "Modern Video Toolkit", that aims to provide high-quality, high-performance GPU-based video filters is out with a new release.
The Movit 1.6 release today by Steinar Gunderson adds support for video effects that work as compute shaders. Gunderson's hopes for using more compute shaders in Movit didn't work out quite as planned with fragment shaders generally being faster for this use-case. But this release does have compute shaders for deinterlacing as one area where it's faster.
With the Movit 1.6 release it is now dependent upon a C++11 compiler and support migrated as well away from SDL1 to strictly SDL2. As part of his evaluation into compute shader performance, Movit now has basic benchmarking support through integration with Google's micro-benchmarking framework. I will be exploring this when time allows to see if it's sufficient for setting up a test profile for being able to deliver Movit benchmarks with the Phoronix Test Suite.
More details on the Movit 1.6 GPU video filter library via this blog post. Those wishing to learn more about the project itself can do so via this detailed README.
The Movit 1.6 release today by Steinar Gunderson adds support for video effects that work as compute shaders. Gunderson's hopes for using more compute shaders in Movit didn't work out quite as planned with fragment shaders generally being faster for this use-case. But this release does have compute shaders for deinterlacing as one area where it's faster.
With the Movit 1.6 release it is now dependent upon a C++11 compiler and support migrated as well away from SDL1 to strictly SDL2. As part of his evaluation into compute shader performance, Movit now has basic benchmarking support through integration with Google's micro-benchmarking framework. I will be exploring this when time allows to see if it's sufficient for setting up a test profile for being able to deliver Movit benchmarks with the Phoronix Test Suite.
More details on the Movit 1.6 GPU video filter library via this blog post. Those wishing to learn more about the project itself can do so via this detailed README.
1 Comment