Three dimensional modeling and rendering of head hair
A method for three dimensional modeling and rendering of head hair, the method may include obtaining a three dimensional (3D) point cloud of head hair points of a person; determining head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; grouping the hear hair points into multiple strand groups, wherein the grouping is based on distances between the head hair points and the head hair points 3D directions; wherein each strand group is associated with strand spatial information; and generating, by a machine learning process, a representation of the head hair based on strand spatial information of the multiple strand groups and on strand shape information.
1. A method for three dimensional modeling and rendering of head hair, the method comprises:
obtaining a three dimensional (3D) point cloud of head hair points of a person;
determining head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
grouping the head hair points into multiple strand groups, wherein the grouping is based on distances between the head hair points and the head hair points 3D directions; wherein each strand group is associated with strand spatial information; and
generating, by a machine learning process, a representation of the head hair based on strand spatial information of the multiple strand groups and on strand shape information.
2. The method according to claim 1 wherein the representation is renderable from any point of view.
3. The method according to claim 1 wherein the applying of the direction finding process comprises applying a filter out of a Gabor filter and a Sobel filter.
4. The method according to claim 1 wherein the head hair points of the 3D point cloud are associated with color and transparency.
5. The method according to claim 1 comprising receiving point of view and distance information and generating the representation as being viewed by a virtual camera having the point of view and located at the distance from the head hair.
6. The method according to claim 1 wherein the point of view and the distance are arbitrary.
7. The method according to claim 1 wherein the point of view and the distance are selected out of an arbitrary range of points of view and distances.
8. The method according to claim 1 wherein the multiple strand groups provide a volume-less represent of the strands.
9. The method according to claim 8 wherein the strand shape information provides volumetric strand information.
10. The method according to claim 1 wherein the machine learning process is trained by being fed with (a) images of head hair of persons that are acquired from different combinations of point of view and distance, and (b) models of the head hair of the person.
11. A non-transitory computer readable medium for three dimensional modeling and rendering of head hair, the non-transitory computer readable medium that stores instructions for:
obtaining a three dimensional (3D) point cloud of head hair points of a person;
determining head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
grouping the head hair points into multiple strand groups, wherein the grouping is based on distances between the head hair points and the head hair points 3D directions; wherein each strand group is associated with strand spatial information; and
generating, by a machine learning process, a representation of the head hair based on strand spatial information of the multiple strand groups and on strand shape information.
12. The non-transitory computer readable medium according to claim 11 wherein the representation is renderable from any point of view.
13. The method according to claim 12 wherein the generating is also responsive to a requested point of view.
14. The non-transitory computer readable medium according to claim 11 wherein the applying of the direction finding process comprises applying a filter out of a Gabor filter and a Sobel filter.
15. The non-transitory computer readable medium according to claim 11 wherein the head hair points of the 3D point cloud are associated with color and transparency.
16. The non-transitory computer readable medium according to claim 11 that stores instructions for receiving point of view and distance information and generating the representation as being viewed by a virtual camera having the point of view and located at the distance from the head hair.
17. The non-transitory computer readable medium according to claim 11 wherein the point of view and the distance are arbitrary.
18. The non-transitory computer readable medium according to claim 11 wherein the point of view and the distance are selected out of an arbitrary range of points of view and distances.
19. The non-transitory computer readable medium according to claim 11 wherein the multiple strand groups provide a volume-less represent of the strands.
20. The non-transitory computer readable medium according to claim 19 wherein the strand shape information provides volumetric strand information.
21. The non-transitory computer readable medium according to claim 11 wherein the machine learning process is trained by being fed with (a) images of head hair of persons that are acquired from different combinations of point of view and distance, and (b) models of the head hair of the person.
22. A method for three dimensional modeling and rendering of head hair, the method comprises:
obtaining a three dimensional (3D) point cloud of head hair points of a person;
determining head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
grouping the head hair points into multiple strand groups, wherein the grouping is based on distances between the head hair points and the head hair points 3D directions; wherein each strand group is associated with strand spatial information;
generating, by a machine learning process, head hair metadata regarding head hair based on strand spatial information of the multiple strand groups and on strand shape information; wherein the head hair metadata is renderable from any point of view.
23. A non-transitory computer readable medium for three dimensional modeling and rendering of head hair, the non-transitory computer readable medium that stores instructions for:
obtaining a three dimensional (3D) point cloud of head hair points of a person;
determining head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
grouping the head hair points into multiple strand groups, wherein the grouping is based on distances between the head hair points and the head hair points 3D directions; wherein each strand group is associated with strand spatial information; and
generating, by a machine learning process, head hair metadata regarding head hair based on strand spatial information of the multiple strand groups and on strand shape information; and wherein the head hair metadata is renderable from any point of view.
24. A method for three dimensional modeling and rendering of head hair, the method comprises:
obtaining a machine learning process that is trained to generate a representation of a head hair from any point of view;
determining by the machine learning process, head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
receiving by the machine learning process, (a) strand spatial information of multiple strand groups that are formed from head hair points, and (b) strand shape information;
generating, by the machine learning process, a representation of the head hair based on the strand spatial information of the multiple strand groups and on the strand shape information.
25. A non-transitory computer readable medium for three dimensional modeling and rendering of head hair, the non-transitory computer readable medium that stores instructions for:
obtaining a machine learning process that is trained to generate a representation of a head hair from any point of view;
determining by the machine learning process, head hair points related 3D directions indicative of 3D directions of candidate strands that pass through the head hair points; wherein the determining of the head points directions comprises: creating multiple two dimensional (2D) projections of the 3D point cloud; applying a direction finding process on each of the multiple 2D projections of the 3D point cloud to provide multiple head hair points 2D direction related estimations; and determining the head hair points 3D directions based on the multiple head hair points 2D direction estimations;
receiving by the machine learning process, (a) strand spatial information of multiple strand groups that are formed from head hair points, and (b) strand shape information;
generating, by the machine learning process, a representation of the head hair based on the strand spatial information of the multiple strand groups and on the strand shape information.
26. The non-transitory computer readable medium according to claim 25 wherein the generating is also responsive to a requested point of view.