3
$\begingroup$

hypothetical question here. I've been playing a lot with various animation methods over the past few days here, and have had a lot of successful results with clean full body motion capture (here)

So I have four playstation eyes and 2 kinects at my disposal. Is there a way to do a multi camera, depth based facial track, a multi video track, or just depth inside of blender's existing tools, or is there a scriptable workaround?

I realize there are some problems inherent in a multi camera setup, such as:

1: assuring the cameras are properly aligned within the software to the real world setup, and I'm aware Blender probably doesn't solve that for you. I'm okay with creating a precise measured setup though and replicating the camera orientations in blender.

2: removal of head movement within the frame of video, but retaining gimbal. I feel like this is something that should be able to be tracked out though, if you define a root somewhere on the head.

3 standard distortion/fov inconsitencies between cameras.

4: controller input? I know with Ipisoft it can only do 2 kinects or 8 pseyes (on a standard motherboard without external controllers, which can only bring your number to 3 without custom hardware. I'm unsure about the compatibility of running both kinects and cameras/pseyes simultaneously.

I've seen a few methods for deriving facial based capture in Blender, both through faceshift export, or Blazraidrs method here: (here) These are both interesting methods. I prefer Blazraidr's, but I see problems with both. First off, I'm not adverse to using marker based tracking, and I feel like faceshift loses out on some qualities because it doesn't take advantage of that on top of the depth sensor.

anyway, I suppose it's an open ended discussion. Curious to see what people have to say, if you read all of this thanks. TL;DR: how does multi camera facial capture in teh open source softwures

$\endgroup$
  • $\begingroup$ Maybe this link is helpfull. cs.rit.edu/~iaw5991/report.pdf Please keep me updated if you use it succesfull. $\endgroup$ – user3468 May 13 '14 at 11:49
  • $\begingroup$ Inlined another link to the current GSoC project below... $\endgroup$ – Samoth Apr 25 '16 at 20:46
1
$\begingroup$

It's not possible yet as Blender has no Multicam Support. There is a GSoC project currently and it seems to be possible later this year.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.