I am using the temporal calibration algorithm in the ImFusion Suite and I was wondering if there was any guidance on how to use the optical flow and rigid motion cost functions. For the optical flow, I assumed the calibration was being done by aligning optical tracking data with estimated motion from speckle tracking but I found the temporal calibration to vary from 377 ms to -56 ms and -95ms when I used three difference sweeps of my forearm. For each sweep, I only moved the transducer in the imaging plane (i.e. I only rocked side to side or pressed it into/pulled it away from my forearm.
Are there any guidelines on what I should be imaging for either of these cost function settings?
This algorithm was very experimental, partly misleading, and was thus removed in the most recent release. The rigid motion cost function was not usable at all, and the first two cost functions relied on a bone segmentation ML model, which only makes sense in some very niche applications. The only useful cost function is the optical flow one.
Record a sweep in a water bath, only visualizing the bottom of the container, and having the rest of the image pitch-black. Do multiple manual rapid up&down motions while recording. If the temporal calibration is correct, all these bottom reflection lines should nicely line up an form a plane in 3D space
You can try to run the algorithm on such a sweep, but honestly, the fastest way to do it is to open the Tracking Properties and to manually adjust the temporal offset until the variance of the reflection lines in space is visually minimal.