Approximating mesh deformations for character rigs

Patent metadata

Field	Value
Publication number	US-10565792-B2
Application number	US-201815910898-A
Country	US
Kind code	B2
Filing date	Mar 2, 2018
Priority date	Sep 7, 2017
Publication date	Feb 18, 2020
Grant date	Feb 18, 2020

Abstract

Systems, methods, and computer-readable medium for approximating mesh deformations for character rigs are disclosed. An embodiment includes applying a first deformation function to one or more mesh elements to determine an intermediate position based on a transform to a first structural element, wherein the one or more mesh elements are assigned to the first structural element, generating an offset based on a second deformation function for the one or more mesh elements using a deformation function approximation model, wherein the offset is a positional offset of the one or more mesh elements from the intermediate position to a target position corresponding to the transform applied to the first structural element, and generating a combined mesh deformation for the one or more mesh elements by combining the intermediate position and the offset.

First claim

What is claimed is:

1. A method for generating approximated mesh deformations of a model comprising one or more mesh elements and a first structural element of a plurality of structural elements, the method comprising:
applying a first deformation function to the one or more mesh elements to determine an intermediate position of the one or more mesh elements based on a transform applied to the first structural element, wherein the one or more mesh elements are assigned to the first structural element;
generating an offset, based on a second deformation function, for the one or more mesh elements using a deformation function approximation model, wherein the offset is a positional offset of the one or more mesh elements from the intermediate position to a target position corresponding to the transform applied to the first structural element; and
generating a combined mesh deformation for the one or more mesh elements by combining the intermediate position of the first deformation function and the offset of the second deformation function.

2. The method of claim 1 , wherein:
each of the plurality of structural elements is assigned with at least one mesh element; and
a corresponding deformation function approximation model is provided for each of the plurality of structural elements.

3. The method of claim 1 , wherein the plurality of structural elements correspond to bone representations of a skeleton of the model and the one or more mesh elements correspond to vertices of a mesh of the model.

4. The method of claim 1 , wherein the deformation function approximation model corresponds to a neural network using a learned weight value for the second deformation function.

5. The method of claim 4 , wherein the deformation function approximation model is based on a learned weight value that is determined by training the neural network on a set of training data comprising deformation positions of the one or more mesh elements defined with respect to the first structural element resulting from various transforms applied to the first structural element.

6. The method of claim 4 , wherein an input to the neural network includes a selected subgroup of the plurality of structural elements, wherein a transform applied to each of the subgroup of structural elements affects a position of the one or more mesh elements.

7. A machine-readable non-transitory medium having stored thereon machine-executable instructions for generating approximated mesh deformations of a model comprising one or more mesh elements and a first structural element of a plurality of structural elements, wherein the instructions comprise:
applying a first deformation function to the one or more mesh elements to determine an intermediate position of the one or more mesh elements based on a transform to the first structural element, wherein the one or more mesh elements are assigned to the first structural element;
generating an offset, based on a second deformation function, for the one or more mesh elements using a deformation function approximation model, wherein the offset is a positional offset of the one or more mesh elements from the intermediate position to a target position corresponding to the transform applied to the first structural element; and
generating a combined mesh deformation for the one or more mesh elements by combining the intermediate position of the first deformation function and the offset of the second deformation function.

8. The machine-readable non-transitory medium of claim 7 , wherein:
each of the plurality of structural elements is assigned with at least one mesh element; and
a corresponding deformation function approximation model is provided for each of the plurality of structural elements.

9. The machine-readable non-transitory medium of claim 7 , wherein the plurality of structural elements correspond to bone representations of a skeleton of the model and the one or more mesh elements correspond to a vertices of a mesh of the model.

10. The machine-readable non-transitory medium of claim 7 , wherein the deformation function approximation model corresponds to a neural network using a learned weight value for the second deformation function.

11. The machine-readable non-transitory medium of claim 10 , wherein the learned weight value is determined by training the neural network on a set of training data comprising deformation positions of the one or more mesh elements defined with respect to the first structural element resulting from various transforms applied to the first structural element.

12. The machine-readable non-transitory medium of claim 10 , wherein an input to the neural network includes a selected subgroup of the plurality of structural elements, wherein a transform applied to each of the subgroup of structural elements affects a position of the one or more mesh elements.

13. A terminal for generating approximated mesh deformations of a model comprising one or more mesh elements and a first structural element of a plurality of structural elements, the terminal comprising:
a display configured to display information; and
at least one controller configured to:
apply a first deformation function to the one or more mesh elements to determine an intermediate position of the one or more mesh elements based on a transform to the first structural element, wherein the one or more mesh elements are assigned to the first structural element;
generate an offset, based on a second deformation function, for the one or more mesh elements using a deformation function approximation model, wherein the offset is a positional offset of the one or more mesh elements from the intermediate position to a target position corresponding to the transform applied to the first structural element;
generate a combined mesh deformation for the one or more mesh elements by combining the intermediate position of the first deformation function and the offset of the second deformation function; and
cause the display to display an approximated mesh deformation of the model corresponding to the generated combined mesh deformation.

14. The terminal of claim 13 , wherein:
each of the plurality of structural elements is assigned with at least one mesh element; and
a corresponding deformation function approximation model is provided for each of the plurality of structural elements.

15. The terminal of claim 13 , wherein the plurality of structural elements correspond to bone representations of a skeleton of the model and the one or more mesh elements correspond to a vertices of a mesh of the model.

16. The terminal of claim 13 , wherein the deformation function approximation model corresponds to a neural network using a learned weight value for the second deformation function.

17. The terminal of claim 16 , wherein the learned weight value is determined by training the neural network on a set of training data comprising deformation positions of the one or more mesh elements defined with respect to the first structural element resulting from various transforms applied to the first structural element.

18. The terminal of claim 16 , wherein an input to the neural network includes a selected subgroup of the plurality of structural elements, wherein a transform applied to each of the subgroup of structural elements affects a position of the one or more mesh elements.

19. A method for generating approximated mesh deformations of a model comprising a plurality of structural elements and a plurality of mesh elements, the method comprising:
associating one or more mesh elemen

Assignees

• Dreamworks Animation Llc

Inventors

• Bailey Stephen
• Dilorenzo Paul
• Otte David
• O'Brien James

Classifications

• G06T7/75

  involving models · CPC title

• G06T17/20

  Finite element generation, e.g. wire-frame surface description, {tesselation} · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T17/205 Primary

  Re-meshing · CPC title

• G06T2207/30196

  Human being; Person · CPC title