What is your footage interpretation? Paint operations happen on the layer, not the comp and odd footage interpretation might have exactly this result, since in relation to the comp timebase you naturalyl can't paint on a fractional frame...
I checked that but Interpret footage is set to use FPS from file - 30 fps
What kind of camera records at 30fps?
Sure, there are cameras that do indeed record at 30, but they usually cost as much as a new Volkswagen. Did you use one of those?
Something smells a little fishy here: did you get this footage from a different source than your own camera?
it's an encoded MP4, an animation
I for one don't trust mp4 encoding to work well in AE. I would transcode the clip to a lossless codec, and I recommend Quicktime's PNG codec for the job.
Anybody else care to chime in on this particular bias against mp4?
If someone is expecting you to edit something or work on it in AE, they should give you a proper intermediate codec. MP4 is compressed and, therefore, a terrible choice. DNxHD or Cineform are both great codecs to use if file size is an issue. They're not technically lossless, but they're practically lossless and, unless you're keying (and you wouldn't be keying an animation), they work great.
I've been unable to reproduce this issue. If you make the file available and tell us exactly what your interpretation settings and composition settings are (prefereably with screenshots), I'll try to determine what's going on.
Regarding the recommendations from others for source material:
They are correct that you really shouldn't be using highly compressed video files as source material, especially for a compositing application like After Effects. So much information has already been thrown away by the compression schemes used for these files that you start off at a srious disadvantage for compositing work.