Forum

Using Cameras from C4D via Collada

So would like to have a reliable workflow for importing from Cinema4D to VVVV multiple cameras, both static and animated.

Let’s put aside for now if this is even a good idea…

I found results pretty frustrating at first but I think I am almost there. Maybe we can collect some tips and then find a home for them on the wiki somewhere.

I’ve attached a zip with:
-c4d source file
-collada export
-Obj export
-vvvv patch that should be useful for testing purposes

CamDebug.zip (132.0 kB)

C4D notes so far:

  • Set near and far clipping planes for all cameras Eg 0.1 - 10,000
  • Cam FOV is changeable per cam, but not animated as far as I can tell
  • Any spline keyframes need to be baked (also cam using Xpresso etc)
  • Export collada 1.4 with at least one non camera object in scene
  • Not camera related, but normals don’t seem to be exporting right

Please chip in with additions, corrections, heckling etc

I see a slight difference between baked\unbaked cameras (adding a mainloop set to 30 fps doesn’t help). Is it this way because unbaked relies more than baked (which is more “driven”, in a certain way) on system resources? Or simply baked interpolation is not fine as it should be?

Well the baked one is the correct one, insofar as it is accurate to the c4d timeline. This is because the unbaked one does not export it’s keyframes as splines correctly- hence the reason for baking it.

I slowed the movement down in the patch, just so you could see the difference more clearly. If you want a 100% match to the c4d you should probably set the lfo driving the camera to match the collada length.

Would like to know if there is still wrong distortion going on- you mentioned something about shearing before?

You noticed this looking at the animation in both programs or you used another method - like digging into collada file for example?

Also, to answer to distortion question, and here I come to my very limits inside vvvv, in the movie there’s a change in camera Focal Lenght: though above you stated “Cam FOV is changeable per cam, but not animated as far as I can tell”, I’d like to dig into this topic to understand if it’s possible to build a module allowing us to get same animation.

Ciao

P.s. Sorry for quality, but the point is the camera animation, not the movie itself

moovie_redux.mp4_.zip (519.6 kB)

I came to this conclusion just by looking. But if you want ‘proof’ here is the debug patch with the exact same camera and path in all three cases- linear keys, spline keys & spline keys baked. You can very clearly see the first two are the same.

Collada is xml based so if you want to parse it in vvvv xpath node is your friend.

I don’t think you will be able fix the fov animation from thee vvvv side though- it’s simply not there in the file c4d spits out. I included an animated fov c4d export in the zip- you can open it with wordpad, or even better ‘xml marker’ (this is a great program to look through a patch structure too btw).

So it’s probably more a case of doing something on the c4d side. Mr woods has been doing amazing things with python I hear.

I would probably just adjust the fov in vvvv. If you want the kind of effect in your video check out the vertigo nodes.

CamDebug_KeyframeDemo.zip (16.0 kB)