Video shake is hard to fix. Video from vehicle-mounted or hand- held cameras has a high level of irregular movement from frame to frame.
What makes Ikena different?
Competitive stabilization algorithms use rudimentary techniques to normalize the displacement between frames, using information from only a few frames of video, and relying on simplified motion models (such as translation).
Ikena’s stabilization algorithms leverage MotionDSP’s accurate multi-frame motion estimation (that is required for our super-resolution reconstruction). Because of the accuracy and robustness of our motion fileds, Ikena uses a buffer of 31 video frames (51 in Ikena ISR) to vastly reduce shake and produce the best stabilization correcton.
(Click the video above to see before/after comparison of original and stabilized video)
(Training: how to use stabilization in Ikena)