Walking twice through the same low light environment, this time the Classic Car Night at the Bob's Big Boy in Toluca Lake, California, I shot some scenes of the cars lined up for review.
One clip had me holding the camera in a simple grip, the way most people might shoot. Not too strident, not too careless.
The other clip had the same iPhone 4S on the Steadicam Smoothee, trying to approximate the same camera move. Again, not too strident, not too careless. One-handed for both shots.
In the first shot, the images are side by side. In the second, the camera-only shot is inset within the Smoothee clip. Even reduced this way, you can easily see how much less stable the camera-only image is.
In low light, the iPhone 4S raises its ISO to around 800 and slows the frame rate down to 24 fps. Presumably to gather a little more light.
Interestingly, iMovie does everything it can to combine footage of different frame rates without your ever seeing an obvious visual error in playback.
It places a shot of any frame rate into the edit with its original *duration* intact, no matter what the specs were for the project for its actual playback frame rate.
The final edit is 29.97 fps and the car shots were 24 fps and I can't see obvious general timing flaws in this sort of footage, even though the 24 fps is playing back in a 30 fps file.
At this hour, I don't know if the iPhone simply gives up on the idea of performing in-camera stabilization in low light, but it rather looks like that here. Without coordinated footage showing what zero in-camera stabilization would have looked like, it's just a guess.
You can see a test of that idea here: vimeo.com/30657394
The Smoothee clip in both shots is much more watchable. All the clips were shot with a single hand supporting and steering the camera. You can use the Smoothee two-handed for greater control.
You may view the original 1080p30 edit by downloading it, first.