I just finished watching an episode of one of my favorite shows (Modern Marvels) and it occurred to me that I had no idea how they managed some of their animated still effects. It's obvious that selected areas are placed on different layers and moved at different speeds, that can be done with many apps (such as After Effects). What's not so obvious is how they are able to get such a range of motion, seemingly far beyond the space originally occupied by the cutout area. And the sheer number of these used in a single episode means there must be an efficient workflow for this.

Does anyone have any knowledge of this technique? Are there indeed techniques for hiding the area originally cut out (I tend to doubt it's normal cloning or patch procedure). I'd especially like to see any websites that demonstrate the technique, or sites describing special software used.

I've never seen the show. It might be easier to disect what they did if you could post a couple of screen shots of exactly the images you had questions about.

That said, I'm pretty sure I know the type of effect you're talking about. The difficulty involved and the range of motion you would be able to get would be directly proportional to the control you had over the shoot and the subject matter, and what was in the actual image.

For instance, let's say you were shooting a statue against the background of a park and wanted to create that parallax effect. Instead of taking one photo, extracting the statue and having to clone in the cut out area, you take two photos. For the background photo you move behind the statue and shoot the background shot without the statue. Then you go back and take a photo of the statue and cut/mask it out of that photo. That way, you have the two elements you need to place on different layers, but without all the need to fill in the gap left by a cutout. Doing it like that would greatly increase the range of motion you could get out of the effect.

Again, it largely is dependent on the subject matter of the image. That technique would be a lot harder with something the size of a skyscraper or the Golden Gate Bridge. In cases like that, computer generated imagery could be used to produce the necessary sources. Or, you'd just have to be very very meticulous with the cloning.

Any of that make sense?

--Racc

It makes total sense. Except this technique is used with old photos, not newly-staged shots.

The only thing I can imagine is that they cut out the subject, then enlarge it a bit so it hides the hole it left, but not so much as to look odd. It can then be moved slightly in relation to its own background.

Well, in the case of old photos there is the possibility that photos from the same general angle were found that are before and after the object in question was constructed. Same results. Two photos one with the object and one without.

Or even possibly an old photo of the object combined with a newer, artificially aged photo. Or the old cutout and clone procedure.

One thing to look for that may help determine the method used is by the amount of motion and parallax in the effect. Just a slight movement or a small change in the camera angle would indicate that most likely the image was cut out and the gap filled by cloning, etc. If there is a large amount of motion or change in perspective, enough to clearly see what's behind the object and there are no obvious signs of cloning/pasting then that either suggests the two source method or very good cloning skills.

There are so many different ways to generate still image sources to get that parallax effect. My guess, is they use a combination of whatever works for the particular shot on a clip by clip basis.

--Racc

i watch modern marvels also. i believe they recently did one of stalin walking with that sort of stop-frame animation....you know, big gaps between where he was and now is but it makes it look like he's walked some distance.

i believe some of these types of groupings come from old black and white films. some of those films may be too degraded to actually run as a film any more so they take single decent frames and patch them together.

but, if it's not old film patched together, you can still do the effect and patch in the missing background where the person used to be. i've done this on several photos where i want to move someone in the frame. the method i normally use is clone painting. it's pure extrapolation based on the surrounding bits and pieces. i call it clone painting rather than just cloning because you use a slightly different technique to clone with, often a much bigger brush and often a continous clone rather than dabbing.

you'll normally see this type of patchwork where you dont need to bother with the parallax. the viewpoint of the camera remains the same and it's often people fairly close up. so parallax isnt much of an issue here.

in shots where they move a fair distance, more than what you'd thnk a normal photo would have in it, i can only guess that they've either got more than one image to work with or someone is pretty darn clever in building scenery in photoshop. but, i also know that some of those old black and whites were huge, especially some of the group shots of dignitaries and such. so, maybe they're taking advantage of that also.

but mostly, this is a lot guesswork. i'll have to watch the show again with all this in mind.