I can't get frame 250fps by using S250e camera

[Ushikubo Tomohiro]

Hi.

I'm using S250e camera, and I try to get frames and check frame rate by using FrameSynchronization sample programs.

I set the frame rate option to 250fps, and segment mode. I check the frame rate, the camera don't move in 250fps. The frame rate indicate about 60fps.

Another computer indicate 250fps.



Is camera depends on the hardware?
Are there any equipment before using camera system?


I'm using Camera SDK 1.0.1 beta1, and I show the machine spec.
----------------------
Windows 7 Professional
Dell Vostro 3700
Core i7 CPU Q740 @ 1.73GHz 1.73GHz
4.00GB RAM
32bitOS
Realtek PCIe Family Controller


Thanks.

[beckdo]

When you call camera->SetFrameRate(250) on an S250e, the camera will always run at 250FPS regardless of the computer's performance. If the computer can't keep up, frame will be dropped without being processed.

Also of note, when running a s250e in MJPEG video mode, the camera will max at 125 FPS.

[Ushikubo Tomohiro]

I tested several dell computers. And all computer don't capture camera data by 250fps. I think this phenomena depends on dell computer's chip set.

If you can get frames in 250fps by using dell computer, would you tell me the model and components?

[Ushikubo Tomohiro]

I inform you of information on the addition.

I can get 250fps in dell computer which can't get 250fps before.


In FrameSynchronization sample programs, I comment out FrameSynchronization.cpp source to

//Sleep(2);

in while loop.


Why do you insert the sleep command?

Thanks.

[beckdo]

I'm using a Dell T3500 here. Something you should do is simplify things a bit. Connect a single camera and attempt to receive frames from it. It will default to 125FPS. You should always set the frame rate with camera->SetFrameRate(250); in order to set the camera to output at that rate.

The Sleep() command is important. It's important that you yield the thread to other threads and processes. Without the sleep or other mechanism in place you'll easily poll the cameras one million times for every frame that comes in, which is a massive waste of computing resources. Additionally it will drive that processing core to 100% and starve the other threads in the process, including those involved with processing data coming from the camera.

It first glance it's easy to overlook this, but you always need some mechanism to yield in arrangements like this.