[Ryan Hill]

Well, there’s a lot going on in that ad. Your earlier post asked for a “step by step process” which is also probably why nobody went to the trouble. There’s going to be a lot of steps.

The foreground was green-screened or the like. A close look at the paint splashes shows that they are still images that fade in or wipe in, rather than being full motion splashes. The Ipod wires might need to be animated, because something that thin will be hard to see in the footage. The background has had the levels heavily adjusted to produce the colours it does, with a couple layers of ink textures put on top of it, maybe a ramp.

The good news is the movement of the layers don’t quite match up, so you don’t need to track anything exactly.

[Ryan Hill]

Easiest way is the pick-whip.
https://jjgifford.com/expressions/basics/pickwhip.html

[Ryan Hill]

Glue down a ping-pong ball, shoot it, then slather it in vaseline.

Apply difference matte between the two versions. I can’t say for sure how well that will work, but at least if you have to do it by hand you’ll know what you want it to look like.

[Ryan Hill]

If you want to create a new layer for the background, you’ll want to “create a new solid.” You’ll get to pick a colour. If you want to change it later, it’s under the layer menu, I think as “layer properties” or “solid properties.”

[Ryan Hill]

You might be able to match more precisely with levels. Grab a photo you want to imitate, and write down the RGB values of the lightest area, darkest area, and the mid-tone. Apply a levels effect to your layer and adjust the output values and the mid-range input level to match your recorded values.

[Ryan Hill]

So you’re saying just do it by hand?

If I’m doing it by hand, I’m going with time displacement. Maybe it won’t be as bad as I thought, at first.

[Ryan Hill]

What font are you using?

[Ryan Hill]

You could try pre-composing the layers.

The thing is, transformations are applied after the masks are, so it will transform the masks. If you pre-compose the transformed version, you can change that order.

[Ryan Hill]

“As long as you don’t make the anchor move back to the left faster than you are moving the position to the right”

That might be the key point I’m sticking on. How do I determine the wiggle’s top speed?

[Ryan Hill]

change to mod 3 instead of mod 2, and add a greater than

First layer:
(time%3>1)*100

Second layer:
(time%3>0)*100

Also, if you want to reduce the frequency of the change, you can do something like this:
((time/2)%2>1)*100
to divide the frequency in half.