The Fundamentals of Rotoscoping in DaVinci Resolve

Thanks for this insight Dan! As a freelance colorist & one-man band, it has surprised me how often I end up doing rotoscope work. In the low-medium budget independent world, the lines between VFX & Color are constantly blurred. I completely agree on approaching the situation in Resolve first, as it is generally less complicated and more efficient.

In a short film I recently graded, I was tasked with changing the color of trees to yellow, which is generally a fairly easy qualifier/hue adjustment. Unfortunately, the talent was also wearing green and the key would constantly interfere with his costume. To handle this issue, I actually ended up doing a roundtrip alpha matte workflow from After Effects. I would export the clip from Resolve, rotoscope it in AE, then export the matte back into Resolve. It was the only way to create a clean separation of the costume and trees.

I know this is a common workflow for post-houses, but having some ability in VFX can be a powerful asset for freelance colorists who are often asked to perform tasks like blurring out logos in shots with a lot of foreground movement, changing the color of a particular object that masks & qualifiers don’t handle smoothly enough, etc.

Again, thanks for getting into the nitty-gritty on this insight. These non-glamorous techniques are so important.

Thanks, good insight. I just wonder do you ever use the tracking of Resolve to speed up your Roto? Or got Mocha and generate a mask from there?

I think the most valuable thing Dan has here is the tip of using multiple shapes to create a complex shape. He did a tutorial on this a few years ago, and that was of tremendous help to me in separating foreground actors from backgrounds. Multiple additive shapes are far easier to deal with (at least for me) than a single really complex shape, particularly in roto.

I’m super late to this party, but it seems that – rather than using auto keyframing and the wiggle – it would be easier to right click and manually add each static key frame whenever the frame changes (thereby avoiding dynamic/auto keyframing all together).

That said, I really enjoyed watching the learning process unfold. And the multipe-shapes workflow is hard to beat: simplifis the entire process by breaking it up into bits!