Distributed image compression framework for digital cinema

High quality images and their compression impose heavy computational burden on digital cinema production. While many optimizations have been proposed, they require additional hardware support and are far away from accomplishing real time or quasi real time digital cinema image compression. This paper proposed a cost-saving distributed image compression framework, which can be easily integrated with render farms frequently used in digital cinema post-production. It can be seen from the system model and task assignment algorithms that this framework accommodates important image compression features including GOP and frame-skip, and it also consider the multi-core capability of render farms. Mathematical performance analysis and practical experiment results are both performed to show the efficiency of this framework. Quasi real time image compression speed 12.5 fps can be achieved in the test render farm. And real time image compression may be possible by improving the capability of the shared repository.