Granted Patent B2

US 12,573,130 · App. 18/236,338 · Granted Mar 10, 2026

Inventors: Chelhwon Kim (Palo Alto, CA); Nicolas Dahlquist (Redwood City, CA)

Assignee: LEIA SPV LLC

G06T15/08G06T7/70G06T15/04G06T15/503

Abstract

Claims (55)

1 . A computer-implemented method of providing a three-dimensional (3D) model, the method comprising:

generating a volumetric density function from a first image set, the first image set corresponding to different viewing angles of an object;

generating a second image set from the volumetric density function and from a predefined color function;

blending the first image set with the second image set to generate a third image set;

generating a 3D surface model from the third image set; and

generating a texture map for the 3D surface model from the first image set,

wherein the 3D surface model and texture map are configured to be rendered for display;

wherein blending comprises:

assigning a blending weight indicating whether a pixel is part of a textureless region of the first image set; and

blending the first image set with the second image according to the blending weight; and

wherein the textureless region comprises pixels within a threshold pixel value variance.

2 . The method of claim 1 , wherein the volumetric density function generates a set of volumetric density values corresponding to an input camera pose.

3 . The method of claim 2 , wherein the volumetric density function is a neural network model that comprises a neural radiance field model.

4 . The method of claim 1 , further comprising:

identifying a plurality of camera poses of the first image set; and

generating the volumetric density function from the camera poses.

5 . The method of claim 1 , wherein generating the 3D surface model of the third image set comprises:

identifying a plurality of camera poses of the third image set;

identifying a plurality of 3D points within the third image set using the plurality of camera poses of the third image set; and

reconstructing a surface of the object according to the 3D points.

6 . The method of claim 1 , wherein rendering the 3D surface model and texture map for display comprises contemporaneously rendering a set of views of the object as a multiview image.

7 . A three-dimensional (3D) model generation system comprising:

a processor; and

a memory that stores a plurality of instructions, which, when executed, cause the processor to:

generate a volumetric density model from a first image set, the first image set corresponding to different viewing angles of an object;

generate a second image set from volumetric density model, wherein a pseudo-random texture is applied to generate the second image set;

blend the first image set with the second image set to generate a third image set;

generate a 3D surface model from the third image set; and

generate a texture map for the 3D surface model from the first image set,

wherein a computing system is configured to render the 3D surface model and the texture map for display;

wherein blending comprises:

assigning a blending weight indicating whether a pixel is part of a textureless region of the first image set; and

blending the first image set with the second image according to the blending weight; and

wherein the textureless region comprises pixels within a threshold pixel value variance.

8 . The system of claim 7 , wherein the volumetric density model comprises a function configured to determine a set of volumetric density values corresponding to an input camera pose.

9 . The system of claim 7 , wherein the volumetric density model comprises a neural radiance field model.

10 . The system of claim 7 , wherein the memory that stores the plurality of instructions, which, when executed, further cause the processor to:

identify a plurality of camera poses of the first image set; and

generate the volumetric density model from the first image set according to the camera poses.

11 . The system of claim 7 , wherein the memory that stores the plurality of instructions, which, when executed, further cause the processor to:

identify a plurality of camera poses of the third image set;

identify a plurality of 3D points within the third image set using the plurality of camera poses of the third image set; and

reconstruct a surface of the object according to the 3D points to generate the 3D surface model.

12 . The system of claim 7 , wherein the 3D surface model and texture map are configured to be rendered for display by contemporaneously rendering a set of views of the object as a multiview image.

13 . A non-transitory, computer-readable storage medium storing executable instructions that, when executed by a processor of a computing system, performs operations to generate a three-dimensional (3D) model of a first image set, the operations comprising:

generating a neural radiance field (NeRF) model from the first image set, the first image set corresponding to different viewing angles of an object;

generating a second image set from the NeRF model and a predefined color function;

blending the first image set with the second image set to generate a third image set;

generating a 3D surface model from the third image set; and

generating a texture map for the 3D surface model from the first image set, wherein the computing system is configured to render the 3D surface model and texture map for display:

wherein blending comprises:

assigning a blending weight indicating whether a pixel is part of a textureless region of the first image set; and

blending the first image set with the second image according to the blending weight; and

wherein the textureless region comprises pixels within a threshold pixel value variance.

14 . The non-transitory, computer-readable storage medium of claim 13 , wherein the 3D surface model and texture map are configured to be rendered for display by contemporaneously rendering a set of views of the object as a multiview image.

Assignments (3)

Continuity (2)

References Cited (60)