Please Tell Me How This Guy Did This!

Well he most likely started with motion tracking, had another compositiong and just applied the motion tracking data to it, and then just animated all the movements in the other comp. And for the shadow, he used, the filter, drop shdow.
 
3D layers in Photoshop with drop shadow + trackers. Simple but tedious in his case cause there are a lot of letters! He may have streamlined his work with some scripts that mathematically chained letters together
 
Thats actually quite easy to do. Photoshop generates some elements as well as clean plates for the video. Add motion tracking. Put the letters and number as 3D layers in After Effects, and move forward along the z axis. With the 3D layer, a virtual camera, and virtual lights in AE, you have the option of casting shadows, as well as some simple texture attributs like specularity.

He could have used something like Sound Keys to create their movements too.
 
SomewhereinLA said:
I would say it was done using a 3d software, much faster and more efficient than using after fx.
Not quite.. this would be quite easy in AE. It could be done in a 3D app as well, but AE's motion tracker and 3D space are more than adequate to do this.

BTW, it is a pretty cool piece. I like how it takes a limitation (shaky prosumer camera) and turns it into a nice piece. I think the final shot (the FLOAT text) could have been something different.. maybe a better resolve as to WHY these letters and symbols decided to take the day off.
 
I would say it's definetly done in AE (or other compositing software), not a 3D app (like Maya, etc).

Author created a clean-plate, motion tracked , used 3D layers for the letters, and probably set-up lights in AE, to create somewhat controllable shadows. There are many different ways to create all the effects, but that would be my guess as the best possible solution. He might have actually created artificial camera-shake inside of AE to cover-up some of his tracking. I have motion tracked some stuff, and it is amazing that if you don't get a 100% accurate track how putting a little artificial camera movement in the shot will cover it up. That's one of the beauties of HD. If you are working with HD content, and outputting to SD, you have a ton of pixels that are getting cropped off that you can use to simulate camera movements.
 
SomewhereinLA said:
I would say it was done using a 3d software, much faster and more efficient than using after fx.


In which universe and with which software? Even modest renders in a 3D app with even the basest necessary antialiasing are sometimes 10-20 seconds per frame. That's not even going into the fact that those 3D elements must be put back into the footage with a compositor. Yeah, 3D apps let you put an image or video file as a backdrop, but rendering all from the software with no way to adjust the individual elements from the background (to say nothing of any hope of Multipass setup) would be about the worst route you could take.

Lastly, none of the objects moving around had any depth to them at all. Perfectly flat planes. That is a pretty good indicator that something like AE was used, and not a 3D package. If I had the Z axis at my disposal come element creation time, you'd bet I would use it.
 
Back
Top