Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
Claims 25-27 are interpreted under 35 U.S.C. 112(f). Support is provided in the form of circuitry (Page 16, paragraph 76 of Specification).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1 is/are rejected under 35 U.S.C. 103 as being unpatentable over Gao et al. (CN 106776028) in view of Phan (U.S. PGPUB 20250111573).
With respect to claim 1, Gao et al. disclose an apparatus (paragraph 46, This implementation uses the system shown in Figure 2) comprising:
a memory configured to store a triangle mesh and a feature texture (paragraph 46, The host allocates memory/video memory space and loads the scene model into memory. Then, the CPU parses the scene model, breaking it down into triangular metadata, material data);
a graphics processing unit (GPU) coupled to the memory, the GPU configured to render an inferred three-dimensional (3D) scene based on the triangle mesh and the feature texture using a ray tracing (paragraph 46, The GPU stores the triangular metadata, material information, and light information in their respective global and constant memories. It also establishes a Kd-Tree with the bounding boxes of each scene model as leaf nodes and the entire 3D scene model as the root node. Finally, ray tracing is performed, paragraph 53, Step 4: Perform ray tracing on the GPU); and
a display unit coupled to the GPU, the display unit configured to display the inferred 3D scene (paragraph 54, The CPU reads the pixel information from the image buffer, displays it on the screen, and completes the rendering). However, Gao et al. do not expressly disclose a learned triangle mesh and a learned feature texture.
Phan, who also deals with rendering computer graphics, discloses using a learned triangle mesh and a learned feature texture (paragraph 24, a texture map machine learning model trained to generate 2D texture maps; and mesh machine learning model trained to generate 3D meshes).
Gao et al. and Phan are in the same field of endeavor, namely computer graphics.
Before the effective filing date of the claimed invention, it would have been obvious to apply the method of using a learned triangle mesh and a learned feature texture, as taught by Phan, to the Gao et al. system, because realistic facial models (including texture maps and meshes) and animations may be rapidly generated and optimized for a virtual character for use in electronic games (paragraph 5 of Phan), thus generate realistic images.
Claim(s) 2-6 is/are rejected under 35 U.S.C. 103 as being unpatentable over Gao et al. (CN 106776028) in view of Phan (U.S. PGPUB 20250111573) and further in view of Ikeda et al. (U.S. PGPUB 20240104844).
With respect to claim 2, Gao et al. as modified by Phan disclose the apparatus of claim 1. However, Gao et al. as modified by Phan do not expressly disclose the GPU comprises a shader processor configured to process the learned triangle mesh and the learned feature texture.
Ikeda et al., who also deal with rendering computer graphics, disclose a method wherein the GPU comprises a shader processor configured to process the learned triangle mesh and the learned feature texture (paragraph 35, The acceleration structure traversal stage may be implemented in software (e.g., as a shader program executing on the SIMD units 138), in hardware, or as a combination of hardware and software, paragraph 36, for triangles that are intersected by the ray, the ray tracing pipeline 300 triggers execution of an any hit shader 306 and/or an intersection shader 307 if those shaders are specified by the material of the intersected triangle).
Gao et al., Phan, and Ikeda et al. are in the same field of endeavor, namely computer graphics.
Before the effective filing date of the claimed invention, it would have been obvious to apply the method wherein the GPU comprises a shader processor configured to process the learned triangle mesh and the learned feature texture, as taught by Ikeda et al., to the Gao et al. as modified by Phan system, because much of the work involved in ray tracing is performed by programmable shader programs, executed on the SIMD units 138 in the compute units 132 (paragraph 33 of Ikeda et al.), thus distributing the work of ray tracing to multiple computing units.
With respect to claim 3, Gao et al. as modified by Phan and Ikeda et al. disclose the apparatus of claim 2, wherein the GPU further comprises a ray traversal unit configured to perform the ray tracing (Ikeda et al.: paragraph 34, traversal through the ray tracing pipeline 300 is performed partially or fully by the scheduler 136, either autonomously or under control of the processor 102, or partially or fully by a shader program (such as a BVH traversal shader program) executing on one or more of the SIMD units 138).
With respect to claim 4, Gao et al. as modified by Phan and Ikeda et al. disclose the apparatus of claim 3, wherein the ray tracing includes a determination of primary visibility (Gao et al.: paragraph 69, Step 4.2: Determine whether the current ray intersects with the bounding box of the Kd-Tree: If it intersects, perform an intersection test between the scene model inside the bounding box and the current ray, and proceed to step 4.3; otherwise, if the current ray does not intersect with the scene model inside the bounding box of the Kd-Tree, fill the image buffer area with the scene model as background information).
With respect to claim 5, Gao et al. as modified by Phan and Ikeda et al. disclose the apparatus of claim 3, wherein the ray tracing includes a bounding volume hierarchy (BVH) technique (Ikeda et al.: paragraph 40, For efficiency, the ray tracing test uses a representation of space referred to as a bounding volume hierarchy. This BVH is the “acceleration structure” referred to elsewhere herein). It would have been obvious for the ray tracing to include a bounding volume hierarchy because the BVH data structure allows the number of ray-triangle intersections (which are complex and thus expensive in terms of processing resources) to be reduced as compared with a scenario in which no such data structure were used and therefore all triangles in a scene would have to be tested against the ray (paragraph 41 of Ikeda et al.).
With respect to claim 6, Gao et al. as modified by Phan and Ikeda et al. disclose the apparatus of claim 3, wherein the shader processor is further configured to infer the inferred three-dimensional (3D) scene to output view-dependent colors (Ikeda et al.: paragraph 38, If the ray hits an object, that pixel is colored based on the closest hit shader 310. If the ray does not hit an object, the pixel is colored based on the miss shader 312. Multiple rays may be cast per pixel, with the final color of the pixel being determined by some combination of the colors determined for each of the rays of the pixel).
Claim(s) 7-8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Gao et al. (CN 106776028) in view of Phan (U.S. PGPUB 20250111573), Ikeda et al. (U.S. PGPUB 20240104844), and further in view of Cole et al. (U.S. PGPUB 20190147642).
With respect to claim 7, Gao et al. as modified by Phan and Ikeda et al. disclose the apparatus of claim 6. However, Gao et al. as modified by Phan and Ikeda et al. do not expressly disclose the shader processor is further configured to synthesize the inferred three-dimensional (3D) scene using a reduced neural network with the ray tracing.
Cole et al., who also deal with rendering an image, disclose a method wherein the shader processor is further configured to synthesize the inferred three-dimensional (3D) scene using a reduced neural network with the ray tracing (paragraph 35, the 3D rendering engine 230 may utilize wireframe rendering, polygon-based rendering, scanline rendering, ray tracing, radiosity, or any other 3D rendering technique. In some implementations, the 3D rendering engine 230 may be a neural network, such that the 3D rendering engine 230 receives the estimated 3D shape and texture from the 3D estimator neural network 220 as input to the neural network, and generates multiple views of the estimated 3D shape and texture by processing this input through the layers of the neural network).
Gao et al., Phan, Ikeda et al., and Cole et al. are in the same field of endeavor, namely computer graphics.
Before the effective filing date of the claimed invention, it would have been obvious to apply the method wherein the shader processor is further configured to synthesize the inferred three-dimensional (3D) scene using a reduced neural network with the ray tracing, as taught by Cole et al., to the Gao et al. as modified by Phan and Ikeda et al. system, because use of a differentiable renderer may provide advantages in some implementations, by generating multiple renderings of a single 3D shape and texture with greater efficiency and simplicity (paragraph 34 of Cole et al.).
With respect to claim 8, Gao et al. as modified by Phan, Ikeda et al., and Cole et al. disclose the apparatus of claim 7, wherein the shader processor is further configured to backpropagate a plurality of two-dimensional (2D) images to the reduced neural network, an initial feature field neural network and an initial opacity field neural network to generate an updated learned triangle mesh and an updated learned feature texture and an updated reduced neural network (Cole et al.: paragraph 35, the 3D rendering engine 230 receives the estimated 3D shape and texture from the 3D estimator neural network 220 as input to the neural network, and generates multiple views of the estimated 3D shape and texture by processing this input through the layers of the neural network. Any other technique may be employed at the 3D rendering engine 230, so long as the 3D renderings output by the 3D rendering engine 230 can be back-propagated to the object recognition engine 210 for processing).
Claim(s) 9-10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Gao et al. (CN 106776028) in view of Phan (U.S. PGPUB 20250111573), Ikeda et al. (U.S. PGPUB 20240104844), Cole et al. (U.S. PGPUB 20190147642), and further in view of Sardari et al. (U.S. PGPUB 20200312042).
With respect to claim 9, Gao et al. as modified by Phan, Ikeda et al., and Cole et al. disclose the apparatus of claim 8. However, Gao et al. as modified by Phan, Ikeda et al., and Cole et al. do not expressly disclose the shader processor is further configured to infer the updated reduced neural network using a forward propagation and with the ray tracing.
Sardari et al., who also deal with rendering an image, disclose a method wherein the shader processor is further configured to infer the updated reduced neural network using a forward propagation and with the ray tracing (paragraph 73, The three dimensional reconstruction module 114 can perform ray tracing to generate shadows based on the light sources for the three dimensional representations of the objects and/or the features, paragraph 74, The discriminator 126 can be trained to determine whether the image looks like a real image or a fake image. For example, the discriminator 126 may forward propagate the two dimensional view through layers of a neural network trained to determine whether the image looks real and/or fake and output a binary determination of real and/or fake, or a probability that the image is real).
Gao et al., Phan, Ikeda et al., Cole et al., and Sardari et al. are in the same field of endeavor, namely computer graphics.
Before the effective filing date of the claimed invention, it would have been obvious to apply the method wherein the shader processor is further configured to infer the updated reduced neural network using a forward propagation and with the ray tracing, as taught by Sardari et al., to the Gao et al. as modified by Phan, Ikeda et al., and Cole et al. system, because the stored objects and/or features that was determined to appear “real” by the discriminator 126 and/or exceeded a certain threshold of comparison from the original image can be reused for similar objects and/or features in different locations (paragraph 78 of Sardari et al.), thus ensuring realistic looking images.
With respect to claim 10, Gao et al. as modified by Phan, Ikeda et al., Cole et al., and Sardari et al. disclose the apparatus of claim 9, wherein the shader processor is further configured to synthesize the inferred three-dimensional (3D) scene using the updated reduced neural network with the updated learned triangle mesh and the updated learned feature texture and with the ray tracing (Sardari et al.: paragraph 128, At block 316, the three dimensional reconstruction system can render a two dimensional view of the three dimensional scene, such as using standard 3D rendering techniques known in the art given the arranged 3D scene and custom object information described above).
Claim(s) 11-14, 22-25, 27-28, and 30 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hedman et al. (WO 2024025782) in view of Gao et al. (CN 106776028).
With respect to claim 11, Hedman et al. disclose a method comprising:
using an initial mesh and an initial feature texture generated by an initial feature field neural network and an initial opacity field neural network with a plurality of two-dimensional (2D) images (paragraph 38, The computing system can process a plurality of positions associated with a learnable mesh model 204 with a feature field model 208 to generate feature data 212 and with an opacity field model 210 to generate opacity data 214, paragraph 45, the computing system can bake the feature data 212 and the opacity data 214 respectively output by the feature field model 208 and the opacity field model 210 into one or more texture images (not shown), for all positions (e.g., polygons) of the mesh model 204, the feature field model 208 and the opacity field model 210 can be evaluated and their respective outputs can be stored in the one or more texture images for use during later rendering); and
synthesizing an initial three-dimensional (3D) scene using an initial reduced neural network, the initial mesh and the initial feature texture (paragraph 48, a polygonal mesh whose texture maps store features and opacity. At rendering time, given a camera pose, some example implementations adopt a two-stage deferred rendering process, paragraph 49, Rendering Stage 1, paragraph 50, Rendering Stage 2). However, Hedman et al. do not expressly disclose using ray tracing.
Gao et al., who also deal with rendering an image, disclose synthesizing an initial 3D scene using the initial mesh and initial feature texture using ray tracing (paragraph 46, The GPU stores the triangular metadata, material information, and light information in their respective global and constant memories. It also establishes a Kd-Tree with the bounding boxes of each scene model as leaf nodes and the entire 3D scene model as the root node. Finally, ray tracing is performed, paragraph 53, Step 4: Perform ray tracing on the GPU).
Hedman et al. and Gao et al. are in the same field of endeavor, namely computer graphics.
Before the effective filing date of the claimed invention, it would have been obvious to apply the method of synthesizing an initial 3D scene using the initial mesh and initial feature texture using ray tracing, as taught by Gao et al., to the Hedman et al. system, because ray tracing is used to generate realistic 3D virtual scenes (paragraph 4 of Gao et al.), thus improving visual quality over other rendering methods.
With respect to claim 12, Hedman et al. as modified by Gao et al. disclose the method of claim 11, wherein the initial reduced neural network is a multilayer perceptron (MLP) neural network (Hedman et al.: paragraph 50, Rendering Stage 2 - some example implementations convert these features into a color image via a (neural) deferred renderer running in a fragment shader, e.g., a small MLP, which receives a feature vector and view direction and outputs a pixel color).
With respect to claim 13, Hedman et al. as modified by Gao et al. disclose the method of claim 11, wherein the initial mesh is a set of three-dimensional (3D) spatial samples which represents a geometric object (Hedman et al.: paragraph 61, Without loss of generality, some example implementations operate with respect to the polygonal mesh used in Synthetic 360° scenes).
With respect to claim 14, Hedman et al. as modified by Gao et al. disclose the method of claim 11, wherein the ray tracing includes a determination of primary visibility (Gao et al.: paragraph 69, Step 4.2: Determine whether the current ray intersects with the bounding box of the Kd-Tree: If it intersects, perform an intersection test between the scene model inside the bounding box and the current ray, and proceed to step 4.3; otherwise, if the current ray does not intersect with the scene model inside the bounding box of the Kd-Tree, fill the image buffer area with the scene model as background information).
With respect to claim 22, Hedman et al. as modified by Gao et al. disclose the method of claim 11, wherein the initial reduced neural network has a lower dimensionality than the initial feature field neural network and the initial opacity field neural network (Hedman et al.: paragraph 38, feature field model 208 and the opacity field model 210 can be neural radiance field models (e.g., implemented using relatively larger MLPs), paragraph 39, The computing system can process the feature data 212 and the camera pose 202 with a neural fragment shader (e.g., implemented using a relatively smaller MLP) to generate color data 220).
With respect to claim 23, Hedman et al. as modified by Gao et al. disclose the method of claim 21, further comprising establishing the initial feature field neural network and the initial opacity field neural network (Hedman et al.: paragraph 38, The computing system can process a plurality of positions associated with a learnable mesh model 204 with a feature field model 208 to generate feature data 212 and with an opacity field model 210 to generate opacity data 214, feature field model and opacity field model established before use).
With respect to claim 24, Hedman et al. as modified by Gao et al. disclose the method of claim 23, further comprising ingesting the plurality of two-dimensional (2D) images for machine learning (ML) training (Hedman et al.: paragraph 38, Specifically, a computing system can obtain a training image 206 that depicts a scene from a camera pose 202, paragraph 46, As another example, in some implementations, after the one or more training iterations, any portions (e.g., polygons) of the mesh model that are not-visible in any of the plurality of training images of the scene can be pruned).
With respect to claim 25, Hedman et al. as modified by Gao et al. disclose an apparatus (Hedman et al.: paragraph 79, Figure 3A depicts a block diagram of an example computing system 100 according to example embodiments of the present disclosure) comprising: means for executing (Hedman et al.: paragraph 81) the method of claim 11; see rationale for rejection of claim 11.
With respect to claim 27, Hedman et al. as modified by Gao et al. disclose the apparatus of claim 26, further comprising: means for executing the method of claims 23-24; see rationale for rejection of claims 23-24.
With respect to claim 28, Hedman et al. as modified by Gao et al. disclose a non-transitory computer-readable medium storing computer executable code, operable on a device comprising at least one processor and at least one memory coupled to the at least one processor (Hedman et al.: paragraph 81, The memory 114 can include one or more non-transitory computer-readable storage media), wherein the at least one processor is configured to implement a three-dimensional (3D) scene synthesis using a ray tracing, the computer executable code comprising the method of claim 11; see rationale for rejection of claim 11.
With respect to claim 30, Hedman et al. as modified by Gao et al. disclose the non-transitory computer-readable medium of claim 29, further comprising: instructions for causing the computer to execute the method of claims 23-24; see rationale for rejection of claims 23-24.
Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hedman et al. (WO 2024025782) in view of Gao et al. (CN 106776028) and further in view of Ikeda et al. (U.S. PGPUB 20240104844).
With respect to claim 15, Hedman et al. as modified by Gao et al. and Ikeda et al. disclose the method of claim 11, wherein the ray tracing includes a bounding volume hierarchy (BVH) technique (Ikeda et al.: paragraph 40, For efficiency, the ray tracing test uses a representation of space referred to as a bounding volume hierarchy. This BVH is the “acceleration structure” referred to elsewhere herein). It would have been obvious for the ray tracing to include a bounding volume hierarchy because the BVH data structure allows the number of ray-triangle intersections (which are complex and thus expensive in terms of processing resources) to be reduced as compared with a scenario in which no such data structure were used and therefore all triangles in a scene would have to be tested against the ray (paragraph 41 of Ikeda et al.).
Claim(s) 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hedman et al. (WO 2024025782) in view of Gao et al. (CN 106776028) and further in view of Sardari et al. (U.S. PGPUB 20200312042).
With respect to claim 16, Hedman et al. as modified by Gao et al. disclose the method of claim 11, further comprising using a forward propagation for synthesizing the initial three-dimensional (3D) scene (Sardari et al.: paragraph 73, The three dimensional reconstruction module 114 can perform ray tracing to generate shadows based on the light sources for the three dimensional representations of the objects and/or the features, paragraph 74, The discriminator 126 can be trained to determine whether the image looks like a real image or a fake image. For example, the discriminator 126 may forward propagate the two dimensional view through layers of a neural network trained to determine whether the image looks real and/or fake and output a binary determination of real and/or fake, or a probability that the image is real). it would have been obvious to apply the method of using a forward propagation for synthesizing the initial three-dimensional (3D) scene, because the stored objects and/or features that was determined to appear “real” by the discriminator 126 and/or exceeded a certain threshold of comparison from the original image can be reused for similar objects and/or features in different locations (paragraph 78 of Sardari et al.), thus ensuring realistic looking images.
Claim(s) 17-19, 21, 26, and 29 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hedman et al. (WO 2024025782) in view of Gao et al. (CN 106776028) and further in view of Cole et al. (U.S. PGPUB 20190147642).
With respect to claim 17, Hedman et al. as modified by Gao et al. and Cole et al. disclose the method of claim 11, further comprising backpropagating the initial 3D scene to the initial reduced neural network, the initial feature field neural network and the initial opacity field neural network to create a trained reduced neural network using a forward propagation and with the ray tracing (Cole et al.: paragraph 35, the 3D rendering engine 230 receives the estimated 3D shape and texture from the 3D estimator neural network 220 as input to the neural network, and generates multiple views of the estimated 3D shape and texture by processing this input through the layers of the neural network. Any other technique may be employed at the 3D rendering engine 230, so long as the 3D renderings output by the 3D rendering engine 230 can be back-propagated to the object recognition engine 210 for processing). It would have been obvious to apply the method of backpropagating the initial 3D scene to the initial reduced neural network, the initial feature field neural network and the initial opacity field neural network to create a trained reduced neural network using a forward propagation and with the ray tracing, because use of a differentiable renderer may provide advantages in some implementations, by generating multiple renderings of a single 3D shape and texture with greater efficiency and simplicity (paragraph 34 of Cole et al.).
With respect to claim 18, Hedman et al. as modified by Gao et al. and Cole et al. disclose the method of claim 17, further comprising synthesizing an inferred three-dimensional (3D) scene using an updated reduced neural network with an updated learned mesh and an updated learned feature texture and with the ray tracing (Cole et al.: paragraph 35, the 3D rendering engine 230 may utilize wireframe rendering, polygon-based rendering, scanline rendering, ray tracing, radiosity, or any other 3D rendering technique. In some implementations, the 3D rendering engine 230 may be a neural network, such that the 3D rendering engine 230 receives the estimated 3D shape and texture from the 3D estimator neural network 220 as input to the neural network, and generates multiple views of the estimated 3D shape and texture by processing this input through the layers of the neural network).
With respect to claim 19, Hedman et al. as modified by Gao et al. and Cole et al. disclose the method of claim 18, wherein the ray tracing includes a determination of primary visibility (Gao et al.: paragraph 69, Step 4.2: Determine whether the current ray intersects with the bounding box of the Kd-Tree: If it intersects, perform an intersection test between the scene model inside the bounding box and the current ray, and proceed to step 4.3; otherwise, if the current ray does not intersect with the scene model inside the bounding box of the Kd-Tree, fill the image buffer area with the scene model as background information).
With respect to claim 21, Hedman et al. as modified by Gao et al. and Cole et al. disclose the method of claim 18, further comprising outputting one or more view-dependent 3D scenes from an updated mesh and an updated feature texture (Cole et al.: paragraph 50, The 3D estimator neural network 220 may use the image feature vector as input and output an estimated 3D shape and texture corresponding to the object depicted in the synthetic image).
With respect to claim 26, Hedman et al. as modified by Gao et al. and Cole et al. disclose the apparatus of claim 25, further comprising: means executing the method of claims 17-18; see rationale for rejection of claims 17-18.
With respect to claim 29, Hedman et al. as modified by Gao et al. and Cole et al. disclose the non-transitory computer-readable medium of claim 28, further comprising: instructions for causing the computer execute the method of claims 17-18; see rationale for rejection of claims 17-18.
Claim(s) 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hedman et al. (WO 2024025782) in view of Gao et al. (CN 106776028), Cole et al. (U.S. PGPUB 20190147642), and further in view of Ikeda et al. (U.S. PGPUB 20240104844).
With respect to claim 20, Hedman et al. as modified by Gao et al., Cole et al., and Ikeda et al. disclose the method of claim 18, wherein the ray tracing includes a bounding volume hierarchy (BVH) technique (Ikeda et al.: paragraph 40, For efficiency, the ray tracing test uses a representation of space referred to as a bounding volume hierarchy. This BVH is the “acceleration structure” referred to elsewhere herein). It would have been obvious for the ray tracing to include a bounding volume hierarchy because the BVH data structure allows the number of ray-triangle intersections (which are complex and thus expensive in terms of processing resources) to be reduced as compared with a scenario in which no such data structure were used and therefore all triangles in a scene would have to be tested against the ray (paragraph 41 of Ikeda et al.).
Response to Arguments
Applicant's arguments filed February 4, 2026 have been fully considered but they are not persuasive. Applicant argues that Gao does not disclose or suggest a “learned triangle mesh” or “learned feature texture” in that Gao’s triangle meshes and material data are pre-defined (top of page 8 of remarks). In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986). In this case, Gao teaches a triangle mesh and feature texture. Phan teaches use of a “learned triangle mesh” and “learned feature texture.” It is the combination of Gao with Phan that teaches the claimed limitation, as a whole.
Applicant argues that Phan does not teach or suggest the “learned feature texture” in that the specification defines “learned feature texture” as a texture that includes “per pixel neural features” that are “ingested by a neural network…” (middle of page 8). In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “learned feature texture” as a texture that includes “per pixel neural features” that are “ingested by a neural network…”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). As known to one skilled in the art, “learned feature texture” is interpreted as a texture obtained through a machine learning algorithm. Phan discloses such a texture (paragraph 24, a texture map machine learning model trained to generate 2D texture maps).
Applicant argues that neither Gao nor Phan discloses using ray tracing to render a 3D scene based on a “learned triangle mesh” and a “learned feature texture” in that Phan does not use ray tracing while Gao’s ray tracing operates on non-learned data (bottom of page 8). Again, Applicant appears to be arguing against the references individually. As previously explained, the rejection of claim 1 is based on the combination of references.
Applicant argues that Hedman does not disclose or suggest ray tracing and that Hedman’s rendering pipeline does not sample rays or sort polygons in depth order (top of page 9). However, the cited paragraph 23 of Hedman lists an “example implementation” and is not relied on in the Office Action. Furthermore, Gao teaches ray tracing; it is the combination of Hedman and Gao that teaches the limitation as a whole.
Applicant argues that Hedman identifies volumetric rendering as too slow, uses rasterization instead of ray tracing, and concludes Hedman teaches away from using ray tracing. However, the cited paragraph 4 is directed towards “Traditional NeRF implementations,” not the Hedman algorithm.
Applicant argues that Gao does not disclose or suggest using ray tracing with neural network-generated meshes and feature textures (middle of page 9). Again, Applicant appears to be arguing against the references individually when the rejection is based on Hedman’s teachings in combination with the ray tracing of Gao.
Applicant argues that Gao teaches traditional ray tracing and does not disclose or suggest using ray tracing with neural network-generated meshes and feature textures (middle of page 9). Again, such an argument is improper when the rejection is based on the combination of references.
Applicant argues that the Office Action improperly combines Gao’s ray tracing as a drop-in replacement for Hedman’s rasterization (bottom of page 9). In response to applicant's argument that Hedman’s system is designed around rasterization and that none of the cited art disclose or suggests using this specific three-network architecture with ray tracing, the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981). In this case, the combined teachings would have been suggested because ray tracing is used to generate realistic 3D virtual scenes (paragraph 4 of Gao et al.), thus improving visual quality over other rendering methods. Furthermore, claim 11 recites synthesizing a scene with an initial reduced neural network, the initial mesh, and the initial feature texture. It is unclear how said “three-network” architecture is applied with ray tracing.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANDREW GUS YANG whose telephone number is (571)272-5514. The examiner can normally be reached M-F 9 AM - 5:30 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kent Chang can be reached at (571)272-7667. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/ANDREW G YANG/Primary Examiner, Art Unit 2614
2/19/26