Open Source Summit North America 2025

In digital imaging, video devices often face limitations that impact output quality and functionality. Post-processors, such as m2m geometric distortion correction, m2m scalers, or m2m JPEG encoders, can be employed to enhance the output of an Image Signal Processor (ISP). Managing multiple post-processors individually can be complex and inefficient. To address this, the concept of a composite device could be used. This abstraction consolidates various functionalities, including a video node and multiple post-processors, under a single abstraction that acts like a composite video device. The composite video device provides an effective abstraction layer for the libcamera pipeline handler. It determines which post-processor to utilize and when, without requiring manual intervention for each operation. Without such an abstraction, designing a pipeline handler would be significantly more complex. In this talk, we go over the design of this abstraction.