Tracking Experiment

Azman, Nadia and I are working on a VFX short, so there's definitely a lot of tracking work ahead for us.

I was trying to think of a practical method of tracking everything that :

• didn't rely on auto-tracking [too many points, accuracy is suspect + too hard to make precise corrections]
• doesn't require a whole lot of measuring work for survey data [too tiring and prone to errors if measurements don't match up]
• doesn't depend too much on different perpendicular planes to define spatial dimensions [limits the locations/settings that we plan to shoot in]
• uses tracking points from whatever's already in the scene [no need for manual tracking markers, minimizing painting and post-work to cover them up]
  

After a night of experimenting, a little trigonometry, and a little help from Nadia, I think I've managed to come up with a low-tech solution that might work. Here's our first test. A bit shaky [the measurements were done very carelessly], but I think we proved the technique works.