1 person found this helpful
Most of that was ever done in Nuke or Fusion due to their much better 3D spaces and ability for scripting. The Matrix stuff started out as a custom Nuke TCL script for instance. Of course you can do it in AE on some level, it's just going to be much more of a chore. The basic drill is to arrange everything as a virtual panorama and then use distortion effects, masking and other stuff to make it seamless and then use a render cam that matches the imaginary move, so it only sees one "frame" of the pano at a time In the programs I mentioned you can also use additional imported 3D models, curve and bend the layers, use custom maps and so on to spice things up. A lot of the stuff was shot against greenscreen also, so only the actors were photographed as sort of a spritesheet sequence to be inserted into computer generated backgrounds. Most of these surfer/ skateboard clips use much simpler setups and rely on painting out stuff and masking to isolate foregrounds and backgrounds and then use morphing techniques, time-remapping and artificial 3D moves to create the illusion. They are not necessarily genuine bullet time shots, but often patched together from existing footage with clever framing and perhaps that extra GoPro that was mounted somewhere. Anyway, there's no definitive answer to this, since this technique depends a lot on actually planning and setting up the shots in teh first place. If you don't have many cameras, it's more likely you will "hack it", if you have a full synchronized cam rig with 20 cams you would do it differently. And then of course there's always the chance you could be using a motion control camera or slide your camera across the rails during shooting to create parts of the illusion. Really a million ways to skin a cat...
thanks for the response. Sounds like a complicated procedure (probably beyond my limited AE abilities), thought there might be a plugin since the film to help replicate the effect.
I did think about a fast moving rail, but for the purpose I would like to use it for, it would scare the animal (thinking birds feeding with my rig at the side as one example [but could work in other examples]).
It was this video that really got me thinking...
Ive contacted the person but I have not heard anything.
The result from 12 gopro looks too good to be true (dont think the car park example was 12 gopro's though!)
having a better think about it now. I may be looking at this wrong.
So if camera 1 was the main camera (video). The subject comes up and then it does its 3d wrap. The next cameras would then shoot images, not video, and then these images compiled together. The images would be higher quality making a better effect. I suppose the biggest problem then would be to make sure the cameras are close enough together to avoid a jagged turn.
Does that sound like a better solution? Come to think about it, thats how Matrix initially did the effect.
You can fake it with a couple of angles, but to achieve the best quality you need to use as many cameras as you can afford. There are a bunch of fair tutorials on the web showing how to FAKE the matrix effect. They will give you some ideas but none of them that I have seen are really great.
You can shoot video or stills. It doesn't matter that much. If you shoot stills the procedure is to resize those stills to the appropriate size for a movie, import them into AE, set the duration to 1 frame, sequence the layers, and then, pre-compose that layer and use it as a movie in your main comp to cut between the two parts of the between the move. You can use Time Remapping to change the duration of the "matrix" move.
If you shoot video you have more options for picking the time for the move but the procedure is the same. Line up all the videos in time, cut out one frame, sequence the layers, pre-compose or render a movie and go from there.