Geometry tracking

• US 8,666,119 B1
• Filed: 11/29/2011
• Issued: 03/04/2014
• Est. Priority Date: 11/29/2011
• Status: Active Grant

• Current Assignee: Lucasfilm Limited (The Walt Disney Company)
• Original Assignee: Lucasfilm Limited (The Walt Disney Company)
• Inventors: Mallet, Ronald, Bhat, Kiran S., Wooley, Kevin
• Primary Examiner(s): Koziol, Stephen R
• Assistant Examiner(s): Azima, Shaghayegh
• Application Number:
  US13/306,572
• Time in Patent Office: 826 Days
• Field of Search: 382/103, 382/107, 382/156, 345/419, 345/427
• US Class Current: 382/107
• CPC Class Codes: G06T 19/00 Manipulating 3D models or images for computer graphics
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/251 involving models

Abstract

A first frame includes a first 2D image of a motion capture subject in the real 3D space. A 3D model is deformed into a first deformation and placing the deformed 3D model in a virtual 3D space in a first location so that the deformed 3D model corresponds to the first 2D image. A second frame includes a second 2D image of the motion capture subject in the real 3D space. 2D changes in the second 2D image relative to the first 2D image are tracked. The 2D changes are projected on the virtual 3D space as 3D constraints. Based on the second 2D image, the 3D model is again deformed into a second deformation and placing the 3D model in the virtual 3D space in a second location in conformance with the 3D constraints.

Claims

1. A computer implemented method of motion capture, the method comprising:

receiving a first frame captured from a real 3D space, the first frame including a first 2D image of a motion capture subject in the real 3D space;

deforming a 3D model into a first deformation and placing the deformed 3D model in a virtual 3D space in a first location so that the deformed 3D model corresponds to the first 2D image, the 3D model having at least one deformation constraint that restricts how the 3D model can be deformed;

receiving a second frame captured from the real 3D space, the second frame including a second 2D image of the motion capture subject in the real 3D space;

tracking 2D changes in the second 2D image relative to the first 2D image;

projecting the 2D changes on the virtual 3D space as 3D constraints; and

based on the second 2D image, again deforming the 3D model into a second deformation and placing the 3D model in the virtual 3D space in a second location in conformance with the 3D constraints.

2. The method of claim 1, wherein the 3D constraints apply to at least a location and a deformation in which the deviation of the 3D model is at a minimum.

3. The method of claim 1, wherein again deforming the 3D model and placing the 3D model in the virtual 3D space is performed in a single use of a non-linear solver.

4. The method of claim 1, “

wherein deforming a 3D model and placing the deformed 3D model in a virtual 3D space so that the deformed 3D model corresponds to the first 2D image is based on user input.”

5. The method of claim 1, the method further comprising, after again deforming the 3D model and placing the 3D model in the virtual 3D space in conformance with the 3D constraints, refining the tracking of 2D changes in the second 2D image relative to the first 2D image using texture tracking constraints based on the second location.

6. The method of claim 5, wherein the texture tracking constrains are further based on the second deformation.

7. The method of claim 5, the method further comprising determining, for the second frame, a refined location and a refined deformation of the 3D model in the virtual 3D space based on the refined locations of the texture in the second 2D image.

8. The method of claim 1, wherein again deforming the 3D model comprises solving for control values for one of an animation rig that defines non-linear relationships between at least some of the control values and rig deformation.

9. The method of claim 8, wherein the animation rig comprises at least one of a facial rig, a body rig, or a blend shape.

10. The method of claim 1, the method further comprising refining the tracking of 2D location changes of textures by identifying subsections of textures and tracking 2D location changes of the subsections in the first 2D image to the second 2D image.

11. The method of claim 1, the method further comprising:

accessing one or more reference frames, each reference frame showing the motion capture subject; and

tracking 2D changes in the second 2D image relative to the first 2D image includes using a reference frame as a key frame specifying texture locations in a future frame of the motion capture video.

12. The method of claim 1, wherein tracking 2D changes in the second 2D image relative to the first 2D image includes:

identifying the location of an occlusion object that indicates an occlusion of textures in the second 2D image; and

ignoring textures in the second 2D image that overlap a projection of the occlusion object in the second 2D image.

13. The method of claim 1, the method further comprising:

reading a plurality of first camera parameters for the first frame, the first camera parameters corresponding to the physical state of a real camera when the real camera recorded the first frame; and

determining a plurality of second camera parameters for the second frame, the second camera parameters corresponding to the physical state of the real camera when the real camera recorded the second frame.

14. The method of claim 1, the method further comprising:

determining, after determining a new location and a new deformation for the 3D model, a freeform deformation to be applied to the 3D model; and

adding the freeform deformation to the new deformation.

15. A motion capture system, the system comprising:

a geometry tracker configured to;

receive a first frame captured from a real 3D space, the first frame including a first 2D image of a motion capture subject in the real 3D space;

deform a 3D model into a first deformation and placing the deformed 3D model in a virtual 3D space in a first location so that the deformed 3D model corresponds to the first 2D image, the 3D model having at least one deformation constraint that restricts how the 3D model can be deformed;

receive a second frame captured from the real 3D space, the second frame including a second 2D image of the motion capture subject in the real 3D space;

project, after tracking, 2D changes on the virtual 3D space as 3D constraints; and

based on the second 2D image, again deform the 3D model into a second deformation and place the 3D model in the virtual 3D space in a second location in conformance with the 3D constraints

a texture tracker configured to;

track, after receiving the second frame, the 2D changes in the second 2D image relative to the first 2D image.

16. The system of claim 15, wherein again deforming the 3D model and placing the 3D model in the virtual 3D space is performed in a single use of a non-linear solver.

17. The system of claim 15, wherein the texture tracker is further configured to:

identifying the location of an occlusion object that indicates an occlusion of textures in the second 2D image; and

ignoring textures in the second 2D image that overlap a projection of the occlusion object in the second 2D image.

18. The system of claim 15, the geometry tracker further configured to:

determine, after determining a new location and a new deformation for the 3D model, a freeform deformation to be applied to the 3D model; and

add the freeform deformation to the new deformation.

19. A non-transitory computer readable medium storing instructions that, when executed by one or more processing devices, cause the one or more processing devices to perform operations including:

receiving a first flame captured from a real 3D space, the first flame including a first 2D image of a motion capture subject in the real 3D space;

deforming a 3D model into a first deformation and placing the deformed 3D model in a virtual 3D space in a first location so that the deformed 3D model corresponds

to the first 2D image, the 3D model having at least one deformation constraint that restricts how the 3D model can be deformed;

receiving a second flame captured from the real 3D space, the second flame including a second 2D image of the motion capture subject in the real 3D space;

tracking 2D changes in the second 2D image relative to the first 2D image;

projecting the 2D changes on the virtual 3D space as 3D constraints; and

based on the second 2D image, again deforming the 3D model into a second deformation and placing the 3D model in the virtual 3D space in a second location in conformance with the 3D constraints.

20. The medium of claim 19, wherein the operations further include:

determining, after determining a new location and a new deformation for the 3D model, a freeform deformation to be applied to the 3D model; and

adding the freeform deformation to the new deformation.