1 person found this helpful
Stabilize first, at the original speed and frame rate. If you stabilize after the speed change, you are literally throwing away temporal data, which could reduce the ability of the stabilization to work successfully.
my idea to retime first it's because my footage has soft pan movements that at normal speed are not detected and are very visible when playing back faster, so I was thinkin that maybe if the analysis is done at a faster time the pans are more visible and better stabilized.. I have some aerial footage, everything very stable on tilt and roll, but pan moves a bit..
Tell you what: run a test. Find a piece of footage that you think might be a problem, and do it both ways. Then you'll know which looks best.
First rule, preserve your data.. Stabilizing using warp stabilizer re-interprets all pixels. Changing frame rate by interpretation does nothing to the data, it just changes the playback speed. Re-timing changes the pixels in every frame. In this case you have to decide if the temporal data, the difference between each frame is more important than the image in each frame. If you have more temporal data you are going to be able to generate better interpolated images for each frame so I would stabilize first unless there is something going on in the footage that causes stabilization problems. I would just stabilize from the needed in point to the needed outpoint and not stabilize the entire shot. You'll also get better results. Then I would retime and blend the better images to the new frames for the retimed version.