Conforming a grid/ring to a real life wrist, neck, or torso?

[PlatinumFishy]

How would I go about conforming a model to a surface of the body? I am thinking of marking dots in a five row grid around the actor then using Nuke to track the points along with the scene. Using the points I hope to draw a curve from each of the rows then connect the curves to create shapes. But what about when the tacking points go to where the camera cant see? Wouldn't they just hover in the last place they were? To fix this would I just kill them when their velocity is zero? Any tips for achieving this effect would be appreciated.

[anim]

if it's a rigid part of body, where all the tracking points are not moving relative to each other you can do object track from them, it doesn't matter if you see all the points all the time, important part is that you see enough of them for tracking software to give you a solve, never used Nuke for object track, but Syntheyes or 3DEqualizer do a good job

then model simple geo to represent the body part for shadow casting and  parent it as well as your your ring/necklace/...to tracked object

 

if your points are on deforming body part and therefore moving independently from each other, you will need to solve it as motion capture and therefore will most probably need at least 2 synced cameras to reconstruct the individual point movements

[PlatinumFishy]

6 minutes ago, anim said:

if your points are on deforming body part and therefore moving independently from each other, you will need to solve it as motion capture and therefore will most probably need at least 2 synced cameras to reconstruct the individual point movements

My intention is to use it for a character's neck specifically where they will be talking and thus the geometry would be deforming. Do you have any recommended programs for motion capturing from two (or more) camera sources?

[anim]

I've used Syntheyes and 3DEqualizer in past for such tracks and they work well, the easiest is if both (or more) cameras are static or with locked relative position to each other (e.g part of the same rig), otherwise you'd need to solve for camera positions first. Only points visible from at least 2 cameras would be possible to solve since in motion capture each point moves independently

but there are other options as well

you can still object track rigid parts separately Head, Torso, ... then use them as bones for your body, therefore getting smooth skinning for neck etc.

or if you have already body rig and the body movement in the plate is pretty fluid, you may want to try rotomation, where you manually match the body with a few keyframes, it's usually good for contact geometry