the kid stays in the picture

[Mark Briody]

I do this effect by creating a Photoshop psd file. The main layer is the photograph and the layer above it is the person cut out.

By bringing in both layers to AE and setting up a parent/child relationship you can pan/scan/scale the image and it will appear to be just a photo. But any time I want that look I just move or scale the top layer only (the person cutout layer). This gives the illusion that the cutout is popping out of the photo.

Mark

[Jack Hilkewich]

I do this slightly differently. I create the layers in photoshop. In AE I turn them into 3d layers and then position them so that there’s depth between the different layers then create a camera and just move my camera around. no need to scale or move several layers. You can also use DOF to get some nice looks.

The most time consuming part of the whole process, I find, is the work in photoshop. Cutting out the layers and then seamlessly cloning in the holes can be tedious. I like to divide up the photogaph into at least 3 layers usually more. A fore, mid and back ground layer at least. This will give you some nice depth. Also when cloning the BG always make it bigger(clone the edges out so that the picture is maybe 110% bigger) than it actually is that way you’ll have some extra room to do bigger moves and the camera won’t shoot off the edge.

[Chris Smith]

Also, instead of moving the cam, some find it easier to parent the layers to a null and just animate the null around leaving the came still.

Also check the archives for ‘kid picture’ back when the movie was out there was a ton of discussion on it.

Chris Smith
https://www.sugarfilmproduction.com

[Jack Hilkewich]

you know, that sounds like a good way too! I come from a 3d animator BG so the depth thing and controling cameras come easy but this may be even better. I hardly ever use nulls, i should really start to learn the many ways to use them.

[Chris Smith]

I use nulls all the time in 3D, but in AE where I feel that moving the cam is far less easy to deal with than a true 3D environment, I lean heavily on nulls. Especially to seperate the classes of movement. Like position keying on one null, rotation keying on another, vibration/handheld effects on another, etc. Then just chain them together with the cam at the end of the chain. So if you need to change something or especially offset one from the other it’s really straight forward.

Like if I have motion tracker data, I paste it to a null and parent my video to it. So the video tracks with it, but if I want to offset the movement, it’s as easy as grabbing the video layer in the viewport and moving it, yet it still tracks along because of parenting.

So in the end , all I ever keyframe on the cam layer is cam specific things like focal length, focus distance, etc.

Chris Smith
https://www.sugarfilmproduction.com

[Jack Hilkewich]

Hey Chris,

Thanks for the tips, do you have a preset rig that you use and import into all your projects or do you make it everytime?

I will definitely try this out.

[Chris Smith]

Not for AE.

In C4D, I have all kinds of preset stuff. I have Xpresso/nulls that move in sin/cos with noise functions that you parent an object or cam to to get shaking or cyclical motion. It’s amazing how much movement in life is sin and noise.

For example I loaded a guy’s character rig into C4D and just assigned sin waves with some noise functions on some of the IK controls and made the character dance with incredibly lifelike motion! Looked like mocap almost.

But it’s usually so quick to create a null and select the parent pulldown, I find it just as easy as having a rig.

But in C4D, ppl have made all kinds of cool rigs. One is called “Steadycam” (yeah misspelled per the Steadicam brand) that does amazing cam motion with little effort. There are also rigs that simulate dollies and jibs.

Chris Smith
https://www.sugarfilmproduction.com

[Jack Hilkewich]

Yeah, I meant AE. I use Maya and have a few rigs that I use as well.