The Joy of Stats

[Jeremy Allen]

My best guess is plan, plan, plan.. and then plan some more!

I’m sure it started with a few storyboards, and before shooting, they may have had the data overlay on the monitor for reference so they could direct him on where to make the gestures. I would also take a hint from them as far as editing. This would be really difficult to achieve in one take unless the actor was just perfect. So you can see they use the tight shots to break up the performance and probably cut around bad motions or tongue slips.

The tight shots present another small challenge in keeping the data looking relatively the same size and position vs. the wide shots. Not a huge deal, just make sure you create the data large enough in the first place to zoom in on..

Hope this helps some.

———————————————
8core MacPro, 3.0 GHZ, 10GB RAM, OSX 10.5.8

C4D 11.5
AE CS5
FCP 7.0.3

[Cassius Marques]

make cuts, have a good actor and a better director. I believe everything was done in post, no realtime guide reference for the actor, just directives from a well planned storyboard.

[Jon Bagge]

I agree with how this was probably done. But just to mention, Hans Rosling has done this speach many times before, including on Ted where the graphics were in realtime (although on a regular projector). So he obviously knew exactly how the graphics worked and how to do the performance.
You’ll likely need a lot of practice to get a regular performer to do this well, and you’ll need to know how the graphics will look before you shoot.

————–
http://www.jonbagge.net
Jon Bagge – Editor – London, UK
Avid – FCP – After Effects

[Craig Wales]

Thanks everyone for your collective insight. I agree that planning is essential for something like this. I’m really interested in how one could project the completed animation in front of the actor so he could point to the animated data in real time. The width of the projected image (perhaps rear-projected screen) would need to precisely match the dimensions of the alpha-channeled overlay in post. And of course, shoot plenty of CU shots for cutaways. Does that sound feasible to you all?
–Craig

[Bret Williams]

I suppose you could have the live feed half dissolved over the graphic, and fed back to him over a big teleprompter screen. So he could look directly into the camera whilst making sure he’s hitting the spots. Or just for rehearsal. I would assume the shooter/director is doing something like that at least so they can tell him point higher, etc.

He really didn’t have that many points to hit. And it appeared to me that they were using floor marks. He basically had to point to his lower right, middle, and upper left while standing in the correct spots. Except in the beginning when he pointed to the lower right, lower middle, and lower left. I would assume through some dissolve techniques for the director to double check and giving him a couple places to point to for the main points (could be a 2×4 with an x on it for all that matters) then they would have it. I can guarantee they did it the simplest way possible. No rear projections, and probably no teleprompter with a graphic on it. I’m voting for the c-stand with a piece of tape or even a index card with a number or word.

[Paul Roper]

I think some of the evidence of the inevitable render-test-redo cycle is in the name of the After Effects CS4 comp on the screen at about 15 secs. into the YouTube piece – it’s called “Final comp resized 3 THIS IS THE ONE”. And it’s about time he checked his 135 unread emails!

It is impressive how spot-on the co-ordination between hand gestures and animation is – I wouldn’t be surprised if there was probably also a bit of “if we take down China’s wealth in 1850 by a few million, nobody will notice and it’ll hit his hand exactly!”