Mixing live footage into animated video, also possibly extruding 2d keyed footage into z space


#1

I’ve combined C4d and footage before in Premiere / After Effects, but current vid is mostly animation, and I have just a few spots where I’d like to drop in greenscreened foreground character into c4d scene with animated background.

Since I’m finding C4d much quicker to work in, and most of it is animated, wondering if I can just drop the footage into c4d instead of compositing elsewhere.

I can keep both the physical greenscreen camera and the c4d virtual camera still to no perspective issues, but not clear if I would need to create a null object or plane of some kind in order to place the 2d footage within the scene.

Is there a good way to do this? Good reason NOT to do this even if it is possible? Does this necessitate instead doing the compositing elsewhere?

Also interested in perhaps doing another scene where foreground object is extruded. So, just like text can be readable from front, but extruded into z space, thinking of greenscreening person so they are visible from front, but extruded back with just black sides to match extruded text.

Anyone done this? Do edges get too messy, jump around since they are constantly changing when person moves? Better way to achieve this? Thinking of flipping them around so every time it flips, 2d character is wearing different clothes.


#2

It’s not really a good idea generally speaking.

You can achieve a quick and dirty “slap comp” for previz purposes, but C4D lacks the tools to make a good composite, starting with a proper keyer.

What you can do, however is import a pre-keyed footage and render it inside C4D, on a 2D plane, facing the camera. The type of projection will give you different effects.

But in general the image pipeline isn’t really made for interactive image editing, layering and post effects. It’s made to create assets that you then finish in another software, like photoshop, After Effects, Fusion, Nuke…

The one advantage to have the keyed footage inside the 3D scene is that you get real reflections, refractions and shadows in the final image. If your scene has many of those, then why not. But the drawback is that you have to render everything again each time you want to make changes.


#3

C4D have motion tracking and matchmoving

This is an intro to Cineversity tuto by Noseman but much more tutos are around - i think this one is only for premium members :
https://www.youtube.com/watch?v=MrdlZl1F1Ss but shows what you can get.

Edit you can also do scene reconstruction, so 2D into 3D

https://www.youtube.com/watch?v=ciZcYPLtYq4


#4

I’ve done what you are talking about kanicki. Keyed the Greencreened footage in AE and imported into c4d…for TV commercial.

The commercial had a game show motif. I added each of four show contestants as materials and applied to planes w/alpha channel. Worked great. Alpha edges were fine.

Each of the characters were of course moving…just added as image sequences.

Commercial aired during super bowl (regionally) so I think it turned out well.

In total the 30-second commercial had about 8 After Effects videos brought in…some as scoreboards and such.


#5

Were edges extruded back for the contestants (making them like cookie cutouts of themselves), or just dropped in to look normal?

Not trying to do the keying in C4d. I’m assuming it should be pre-keyed.

I’ve seen a couple tuts before on matching up a virtual camera to a real one. I think they were for virtual 3d cam in AE, but concept should be similar. It was all about tracking points in 3d space, etc. and the greenscreen itself wrapped across two walls and the floor with tracking points embedded.

I don’t have that kind of setup. It will be a homemade green cloth draped on my living room wall, so nothing to track to.

Trying to keep it simple. Character is revealed from behind an animated object, he is sitting on another c4d object. He then gets up and walks off screen. The entire time he is on camera, both the physical and virtual cameras do not move.

It’s not that I mind so much compositing elsewhere, but I’m foreseeing an issue of needing to see the footage in c4d in order to dial in the virtual camera perspective since the physical one is already baked and needs to be matched.

So… sounds like I should be ok doing a greenscreen where the screen drapes over a box (matching approx height of seat object in c4d), then pre-key, and import footage to apply as material on a plane… just in front of the seat on plane where feet lower legs hang off the seat before he stands and walks off to side?

Oh, and just out of curiosity, IceCaveMan, is that commercial up on YouTube anywhere or otherwise publicly viewable? I’d like to check it out.


#6

I did very detailed story-boarding of every single shot before we were in for the studio shoot. So the camera angles (which were sometimes extreme) were all correctly composed in camera (pre-baked) before I brought the footage into c4d.

Simple planes worked fine. They were of course square to/facing the camera.

I am dubious that you can fake extra dimensionality by wrapping the footage around extruded geo. It would likely just distort the footage of the peeps.

If the videos weren’t shot with the desired angles in mind… you will likely need to design your staging and camera angles to be compatible w/the existing footage. You can perhaps get away with 6-10 degree of cheat on the camera angles. By ‘cheat’ I mean deviation from a camera-plane square-angle.

The piece was created in 2009. It’s the first one on this portfolio page:
http://icecavetechnology.com/portfolio.html

I was in Alaska back then…and had only been using c4d for 14 months.


#7

Wasn’t trying to create anything realistic with extrusion. Going for exactly what you get with extruded text. The text is 2d on front face with different color extrusion into z space. So… picture a 2 sticker of a person stuck to the front of a person shaped playdough extrusion in the shape of that person.

I can ballpark the camera angle, but sure I’ll need to adjust it in c4d once it’s dropped in. Much less sophisticated than your shot in that there’s no camera movement while foreground character is onscreen (aside from possible z-space to move camera closer to character from same angle.

Looking now at also doing some background sky footage and other items. Is it best to project on plane object? Background object? Foreground object? Any particular reason one works best? Looking at them now, but I’ve only ever used 3d objects.