About inputTimeStamp parameter

Gaming and Visualization Technologies Video Processing & Optical Flow
1

Hi,

I am new to video development. I develop an application based on CUDA-8.0 and NVEncode-5.0 to stitch frames from multiple cameras. In nvp_nvenc_encode(), the inputTimeStamp would be set to 0. Does this mean that the encoder would calculate output timestamp by FPS only ? Or, maybe, I should ask what the definition and usage of inputTimeStamp is instead.

Is there any document explaining this ?

Any advice and suggestions will be greatly appreciated.

2

Hi bohan,

There are two variables NV_ENC_PIC_PARAMS::inputTimeStamp and NV_ENC_LOCK_BITSTREAM::outputTimeStamp. The driver internally copies NV_ENC_PIC_PARAMS::inputTimeStamp to NV_ENC_LOCK_BITSTREAM::outputTimeStamp. This can be used to keep track of which bitstream corresponds to which frame fed to the NVENCODEAPI for encoding.

Please do explain what problem you are facing or trying to solve, that will help us advise better.

Thanks,
Ryan Park

3

Hi, Ryan

Thank you for your reply.

We stitch frames from multiple cameras to generate a final single frame and we need to add corresponding timestamp for this frame on upper container. I misunderstood that I could use these parameters to generate the timestamp we need, but I realize that these two parameters are not for this purpose, now. The encoder is not responsible for this.

Thanks for your explanation. :)