Tall versus wide versus rotating camera
Optimizing Frame Rate
The camera pipelines the exposure of frame N (current) with the readout of frame N-1 (previous). As a result the maximum frame rate is determined longer of there two operations.
The maximum frame rate due to exposure is approximately 1/(exposure_time + 9us). Please note that this is an approximation and can change in future software releases.
The maximum frame rate due to resolution is approximately 1/(8us + (.3us + horizontal * 1.3ns) * vertical) Once again, note that this is an approximation and can change in future software releases.
From the above formula, we see that there are fixed per-frame and per-line overheads, on top of the time to actually readout the pixels. As a result, for the same total pixels, wide resolution settings like 1024 x 256 will have faster frame rates than tall ones like 256 x 1024. We planned for this by putting a second tripod mount on the side. If you are taking a video where the scene is tall, like dropping an object, and you want the maximum frame rate, it's best to rotate the camera, take the video and then un-rotate the video in your post editor.