v120 latency problem

Post Reply
calvinwang
Posts: 2
Joined: Tue Jun 19, 2018 8:23 am

v120 latency problem

Post by calvinwang » Tue Jun 19, 2018 8:39 am

I use a 300 fps global shutter camera to record motive rendering and the according physical moving.

After comparing, I found the motive rendering has about 15-25 frames latency, which means a rough 50ms-80ms latency.

I know motive rendering is 30fps (33ms), take this into account, the latency is still to high (17ms-47ms).

Any advice?

My environment:
motive: 2.0.2
optitrack: v120/120fps

only 1 rigid body in scene

steven.andrews
NaturalPoint Employee
NaturalPoint Employee
Posts: 469
Joined: Mon Jan 19, 2015 11:52 am

Re: v120 latency problem

Post by steven.andrews » Tue Jun 19, 2018 1:46 pm

Hi calvinwang,

17-42ms of latency sounds quite high, even for the USB cameras. The rendering, in Motive, is not given high priority the way data processing and streaming are. If you are recording your computer's monitor with a camera, there is usually also some latency introduced by the monitor itself. This is not a great way to measure the latency of your data.

To determine the actual latency of your data, you may want to look at the Software Latency reported in the bottom-right corner of Motive.
Capture.JPG
Capture.JPG (19.58 KiB) Viewed 475 times
This would give you a better idea of how long Motive is taking to process the data from the cameras.

Best,
Steven
--
Steven Andrews
OptiTrack | Senior Customer Support Engineer
help.naturalpoint.com
support@optitrack.com

calvinwang
Posts: 2
Joined: Tue Jun 19, 2018 8:23 am

Re: v120 latency problem

Post by calvinwang » Wed Jun 20, 2018 1:53 am

Thanks for your quick reply, steven.

I forgot to mention my device is v120-duo.

Today I did a precise test:
  1. using high speed camera (300fps) to shoot physical rigid body
    1. record images
    2. record image receiving timestamp (microsec)
  2. record motive rigid body position and according "CameraDataReceivedTimestamp"
Find the timestamp when the physical rigid body start to move, denote as camera_microsec.
Find the motive timestamp, when the rigid body position start to change, denote as motive_microsec.

Carefully get the high speed camera latency, denote as camera_latency. It's about 3.5 ~ 4 ms.

I found the motive's latency is:
motive_microsec - camera_microsec + camera_latency.

aver: 24.5 ms

It's unreasonable, cause motive's Software latency is about 0.4ms~0.6ms.

Maybe I missed something to config in motive?

steven.andrews
NaturalPoint Employee
NaturalPoint Employee
Posts: 469
Joined: Mon Jan 19, 2015 11:52 am

Re: v120 latency problem

Post by steven.andrews » Wed Jun 20, 2018 12:29 pm

Hi calvinwang,

I'm still a little unclear on exactly how you are measuring all of your values, such as Motive latency. You mention the Software Latency is less than 1ms, so I am not sure where the 24ms measurement is coming from.

A latency of 24ms is much higher than expected, and I cannot thing of any configuration that would immediately cause this result in Motive.

Could you please reach out to Support directly? When doing so, please share all of your recordings, screenshots, etc, and a detailed description and step-by-step for your process. This will help us to better understand how you are attempting to validate latency with your external camera.

Thanks,
Steven
--
Steven Andrews
OptiTrack | Senior Customer Support Engineer
help.naturalpoint.com
support@optitrack.com

Post Reply