Oil painting stroke simulation using neural network

1. A computer-implemented method to simulate a painting brush stroke, the method comprising:

inferring, by a trained neural network having (i) a first input configured to receive a bristle trajectory map that represents a new painting brush stroke and (ii) a second input configured to receive a first height map of existing paint on a canvas, a second height map of existing paint on the canvas after the new painting brush stroke is applied to the canvas, wherein the first height map of existing paint on the canvas is indicative of, for at least a pixel location, a thickness of the existing paint for the pixel location that extends above the canvas surface in the z-direction before the new painting brush stroke is applied to the canvas; and

generating, by a render module, a rendering of the new painting brush stroke based on the second height map of existing paint on the canvas after the new painting brush stroke is applied to the canvas and a color map.

2. The method of claim 1 , wherein the trained neural network is trained using a plurality of training data sets, each training data set of the plurality of training data sets comprising a height map, a bristle trajectory map, and a ground truth.

3. The method of claim 2 , wherein, for each training data set, the bristle trajectory map and the ground truth are generated by a simulator in response to being provided the height map as an input.

4. The method of claim 3 , wherein the simulator utilizes fluid simulation to generate the ground truth.

5. The method of claim 1 , wherein the bristle trajectory map is a bristle trajectory map segment that represents a new painting brush stroke segment, and further wherein the second height map is a second height map segment of existing paint on the canvas after the new painting brush stroke segment is applied to the canvas.

6. The method of claim 5 , wherein the bristle trajectory map segment is delineated by a bounding box, or the second height map segment is delineated by a bounding box, or both the bristle trajectory map segment and the second height map segment are delineated by a respective bounding box.

7. The method of claim 1 , wherein the new painting brush stroke is a new oil painting brush stroke.

8. The method of claim 1 , wherein the trained neural network is one of a plurality of trained neural networks, each trained neural network of the plurality of trained neural networks having a respective first input configured to receive a respective bristle trajectory map segment that represents a new painting brush stroke segment and a respective second input configured to receive a first height map of existing paint on a canvas, and the inferring is carried out by the plurality of trained neural networks such that each trained neural network is configured to infer a respective second height map segment of existing paint on the canvas after the new painting brush stroke segment is applied to the canvas, the method further comprising:

generating, by a height map combiner module, a combined second height map based on the respective second height map segment inferred by each trained neural network of the plurality of trained neural networks,

wherein generating, by the render module, a rendering of the new painting brush stroke is based on the combined second height map.

9. The method of claim 8 , wherein a length of each bristle trajectory map segment is based on a threshold bristle trajectory map segment length.

10. A computer program product including one or more non- transitory machine readable mediums encoded with instruction that when executed by one or more processors cause a process to be carried out to render painting brush strokes, the process comprising:

receiving a new painting brush stroke;

generating a plurality of bristle trajectory map segments, each bristle trajectory map segment representing a corresponding section of the new painting brush stroke;

receiving, by each of a plurality of trained neural networks, (i) a corresponding bristle trajectory map segment of the plurality of bristle trajectory map segments and (ii) a first height map of existing paint on a canvas prior to application of the new painting brush stroke;

inferring, by each of the plurality of trained neural networks, a corresponding second height map segment of existing paint on the canvas subsequent to application of the new painting brush stroke, such that a plurality of second height map segments are generated by the plurality of trained neural networks;

generating a combined second height map based on the plurality of second height map segments, wherein the combined second height map is indicative of, for at least a pixel location, a thickness of the paint for the pixel location in a direction that is substantially perpendicular to a plane of the canvas, the thickness being subsequent to application of the new painting brush stroke; and

generating a rendering of the new painting brush stroke based on the combined second height map.

11. The computer program product of claim 10 , wherein the process further comprises:

training at least one of the plurality of trained neural networks using a plurality of training data sets, each training data set of the plurality of training data sets comprising a training height map, a training bristle trajectory map, and a ground truth.

12. The computer program product of claim 11 , wherein the process further comprises:

for each training data set, generating, using a simulator, the training bristle trajectory map and the ground truth, in response to being provided the training height map as an input.

13. The computer program product of claim 12 , wherein to generate the ground truth, the process further comprises:

utilizing fluid simulation to generate the ground truth.

14. The computer program product of claim 10 , wherein a particular one of the bristle trajectory map segments is delineated by a first bounding box, or a particular one of the second height map segments is delineated by a second bounding box, or both the particular one of the bristle trajectory map segments and the particular one of the second height map segments are delineated by respective first and second bounding boxes.

15. The computer program product of claim 10 , wherein the new painting brush stroke is a new oil painting brush stroke.