This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

[FAQ] PROCESSOR-SDK-J721E: How to capture different types/formats of camera from the single CSIRX port ?

Part Number: PROCESSOR-SDK-J721E

This article explains how to capture different types/formats of cameras from same CSIRX port in OpenVX.

  • The attached example demonstrates how multiple capture node can opened for the same CSIRX port and how each opened node can be configured for different image type/format


    1. The example creates a independent separate graph for each capture node.
    2. The graph contains only one node ie capture node. The capture node is configured to capture from the same instance id ie same CSIRX node.
    3. Each graph is created in a separate task, so that multiple capture node can run in parallel.
    4. The API app_init_capture is used to configure different parameter for each capture node. On EVM, this is tested by using same image format, RAW12, but this can be chagned on custom platform.
    5. Capture output is declared a graph parameter, to track the number of frames captured.
    6. Please note that deserializer output must be enabled only after all graphs are verified, so sync mechanism using semaphore is used to wait for all graphs to be verified, then enable deserializer output and then start enqueue/dequeue operations in each graph. 

    In order to run this example, copy app_capture folder in vision_apps\apps\basic_demos folder and rebuild vision apps. It will create vx_app_capture.out example, which can be run on EVM.

    Also there is a small change required in the PDK CSIRX driver to run this example. Apply attached patch on ti-processor-sdk-rtos-j721e-evm-07_01_00_11\pdk_jacinto_07_01_00_45 folder.