Accurate frame timing

I’m attempting to acquire images very quickly (>1kHz) for microrheology measurements. By setting an ROI with fewer than 64 rows I can achieve ~1600 fps from the camera (OptiMOS CMOS) which is great, however I run into a problem when it comes to analysing this. To be confident of my measurements, I need to be confident of the frame timing.

The acquisition metadata has several fields that should be of use, such as “PVCAM-TimeStamp”, however these are only ms precision. There is also “Camera-1-ActualInterval-ms” which appears to be 10ns precision, but is the same for every frame within an acquisition. I would expect some small variation in this time, but perhaps I’m unjustified here.

Simply put: Can I trust this ActualInterval field? How else can I accurately get frame times?

I also have a problem with the size of the metadata files - a typical acquisition is >10k images, leading to metadata.txt file sizes of >1GB. Is it possible to save a reduced version of the metadata, or do I need to post-process it to reduce it?

Hi @Will_Hardiman,

Welcome to this forum!

Generally, when running at such high fame rates, cameras will use so called internal trigger mode. This implies that a clock inside the camera drives exposure and readout. This means that the interval between images is highly reliable, and consists of the exposure time, plus some dead-time for reading out the image and/or to clear signal on the sensor. Once images are being send to the computer, timing information becomes less and less reliable. Some cameras provide hardware time stamps, others (like the PVCAM adapter) insert it when the adapter code receives the image (but as you say that only provides ms accuracy in this case). I forgot where the “ActualInterval-ms” is inserted, and would need to research before trusting that number.

In general, I would subtract take the PVCAM timestamp of the first image from the last, and divide by the number of images. Make sure that no images are lost (probably best to look at frame number inserted by the PVCAM adapter code). When the camera is internally triggered, variation in the interval between frames should be as regular as the camera can do (you would need to check the spec sheet and/or contact the vendor if this is important).