TDA4VMXEVM: Is it necessary to run a "camera-based application" using pipelining ?

Hello,

Details -

We have already developed an integrated application (around 22 nodes). And tested it with file i/o on the hardware. It is working as expected for 10 frames of file input. We wanted to display LDC outputs on Display. And this was working fine. This application is not a pipelining based application.

Now, in order to make this application to take the inputs from camera instead of file i/o, hence we have added a "capture node"(4 camera inputs) with appropriate "sensor parameters". and added colour convert node, and merged the outputs of Colour convert node to our already working graph as mentioned above.

We don't want to use pipelining initially. And want to have the application working with camera frame.

After making these modifications, this application, ran for one frame and got stuck at "vxWaitGraph".

We are unable to identify the reason behind the above issue.

For your reference we are uploading the code, usecase diagram and the execution log . And these details will be shared through the TI representative.

We would appreciate quick response on this query.

Thank you,

Ameya Wadekar.

Hello Ameya,

It is better to work in pipeline mode when capture node is part of the pipeline. Otherwise lots of frames could get dropped.. 

Also how many graph parameters do you have? 

Rgds,

Brijesh