Re: What is the fastest possible time it takes to, using code, access a camera and then get a depth frame?

KChow4 · ‎03-04-2019

Hi,

I was testing with the D415 camera to find out how long it took, via code, to access the camera device and capture a single frame.

When testing using the code snippet below, I noticed that it actually took a second or so to do this.

Do RealSense cameras normally take this long to run code like this and is there a faster way to access the camera and take a frame using code (aside from dropping the number of frames that it waits for when stabilizing)?

Thanks in advance!

// Declare RealSense pipeline, encapsulating the actual device and sensors
		rs2::pipeline c_pipe;
		//Set the configurations
		rs2::config c_cfg;
		c_cfg.enable_device(serial_no);
		//Changes in the resolution of the stream 
		c_cfg.enable_stream(RS2_STREAM_COLOR, 848, 480, RS2_FORMAT_BGR8, 30);
	
		// Start streaming with the defined configurations
		c_pipe.start(c_cfg);
		//Declaration of frameset
		rs2::frameset data;
		//Stablize frames
		for (int i = 0; i < 30; i++) {
			data = c_pipe.wait_for_frames();
		}
 
		//Get frame
		rs2::frame colour = data.get_color_frame();

MartyG · ‎03-04-2019

I went over your code and could not see any obvious problems.

The SDK documentation's section on latency may give some useful insight.

https://github.com/IntelRealSense/librealsense/wiki/Frame-Buffering-Management-in-RealSense-SDK-2.0#latency-vs.-performance

KChow4 · ‎03-04-2019

Thanks, I will check it out!