Conan O’Brien-style Lip-Overs

[David Bogie]

Matching skin tone is best done with makeup by a makeup artist who understand video. Trying to do it in post is absurd.
Using AE, you’re completely free to move, rotate, scale, and edit the lips clips but trying to motion track the footage is silly. Much easier to do the whole effect if you have the model’s head steadied somehow. Gaffer tape?

bogiesan

[Bill Russell]

This ridiculous little diddy is like the Conan mouth trick:

Hillary Clinton responding to YouTube absurdity

It’s fairly basic AE process:

1. Filmed mouth “performance” fullish frame. (Diffuse lighting from an open window helped matched face lighting in the footage, sort of, good enough.)

2. Soft edge mask shape around mouth — or mouth and nose in my case. Sized and position the mouth layer to fit over the face.

3. Use color correction tools (levels or curves separate color controls, and HSL) to match black and white levels, mid-range, and skin color.

4. If your “celebrity” is a photo, then you are done. If moving footage, then do two more things: a) Nest the “mouth” comp (pre-compose). b) In the master comp, use motion-tracking to “lock” to the face.

🙂 Good luck and enjoy!

“THE LOST SKELETON OF CADAVRA” –

And more…

[First Last]

Isn’t Conan’s done live with a still frame?

I think they mask out the celebrities mouth and composite it over a close up of someone talking.

You can do it in post the same way.

Mask out mouth of celebrity still.
Shoot close up of someone’s mouth from a similar angle, who’s standing still, saying what you want.
Composite celebrity still over stand in footage.
Resize and position stand in footage of mouth to match celebrity.
Color correct stand in footage to match celebrity still.

Production Associate
Bend, Oregon, United States

[Bill Russell]

P.S. As far as step 3, if your photo is from the web, you might want to enlarge it in photoshop first. PShops’s Bicubic interpolation (choose “smoother”) does a pretty good job of smoothing over jaggies in enlargement. There’s also a jpeg artifacts filter in newer versions. I would save the color matching steps for after effects.

Generally, my approach would be to color correct the photo first until it looks good and strong on video. Then I would color correct the mouth to match. The mouth footage was purple and dark, with a much lower black level for the face shadlows; but lots of tinkering and I got it to match Hillary’s skin pretty precisely. However, of course, for this kind of thing one need not do seemless matching of course, part of what makes it amusing. As a silly YouTube video commentary, even less so 🙂 Cheereo!

“THE LOST SKELETON OF CADAVRA” –

And more…

[Bill Russell]

> Isn’t Conan’s done live with a still frame?

You’re right. No doubt — Conan banters with the performer and it is all fairly ad-lib. They have the photo fed from a computer probably, or have a camera trained on it (old school) or in a TBC frame grab. The other “mouth” feed is from a camera trained on the performing mouth. They use a production switcher to position, size, garbage matte and key the mouth over the photo — very quick and dirty if you’ve used a switcher.

Like you said, the two “feeds” correspond to two layers in After Effects. Or even in Premiere or FCP, would work just as well for photo and stationary mouth – so many options! 🙂

“THE LOST SKELETON OF CADAVRA” –

And more…