DETAILED ACTION
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 10/23/2024 and 10/22/2025 were filed after the mailing date of the application. The submission is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-2, 4-11, 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Kreis et al. (US. Patent App. Pub. No. 2025/0182404, “Kreis”, hereinafter) in view of Yang et al. (“Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction”, retrieved from arXiv, Cornell University, https://arxiv.org/pdf/2309.13101, November 2023, “Yang”, hereinafter).
As per claim 1, as shown in Fig. 1 and 2, Kreis teaches a method of rendering video of a scene, comprising:
receiving a set of images of the scene, wherein each image comprises temporal data and spatial data relating to the scene (¶ [28], “For example, the content model can have dimensions [x, y, z, t] corresponding to three spatial dimensions and a temporal dimension”);
generating, based on the spatial data of each image, three-dimensional (3D) Gaussian splatting data (Fig. 2, step 204, ¶ [34]);
inputting the temporal data of each image and the 3D Gaussian splatting data to a neural network to generate spatial-temporal 3D Gaussian embeddings (¶ [15], “In some implementations, the content model includes a 3D Gaussian splatting representation corresponding to the plurality of spatial dimensions coupled with a multilayer perceptron (MLP) corresponding to the temporal dimension”).
Kreis does not expressly teach generating offset data based on the spatial-temporal 3D Gaussian embeddings; and rendering the video of the scene based on the 3D Gaussian splatting data and the offset data.
However, in a very similar method of rendering scene using 3D Gaussian splatting (see Abstract, page 1), Yang teaches the above feature, i.e., generating offset data based on the spatial-temporal 3D Gaussian embeddings (Fig. 2, page 3, “We use the position (detached) of 3D Gaussians γ(sg(x)) and time γ(t) with positional encoding as input to a deformation MLP network to obtain the offset (δx, δr, δs) of dynamic 3D Gaussians in canonical space” and at section 3.2. Deformable 3D Gaussians, page 5, “Given time t and center position x of 3D Gaussians as inputs, the deformation MLP produces offsets”); and rendering the video of the scene based on the 3D Gaussian splatting data and the offset data (page 4, left column, starting with “We use the position (detached) of 3D Gaussians γ(sg(x)) and time γ(t) with positional encoding as input to a deformation MLP network to obtain the offset (δx, δr, δs) of dynamic 3D Gaussians in canonical space…”. The rendering result is shown in Fig. 5, page 7).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the method as taught by Yang into the method as taught by Kreis as addressed above, the advantage of which is to achieve higher rendering quality and also real-time rendering speed (Abstract, page 1).
As per claim 2, as addressed above, the combined teachings of Kreis and Yang also include wherein rendering the video of the scene comprises: combining the 3D Gaussian splatting data with the offset data to generate spatial-temporal 3D Gaussian representations of the scene; and rendering the video of the scene by inputting the spatial-temporal 3D Gaussian representations to a rasterizer (Yang, Fig. 2, page 3, “…Following that, we use the fast differential Gaussian rasterization pipeline to perform joint optimization of the deformation field and the 3D Gaussians, as well as to adaptively control the density of the set of Gaussians”. See also Yang, left column on page 4). Thus, claim 2 would have been obvious over the combined references for the reason above.
As per claim 4, the combined Kreis-Yang does also teach generating the 3D Gaussian splatting data comprises inputting the spatial data of each image to a machine learning model trained to generate 3D Gaussian splatting data based on spatial data from one or more images (Kreis, ¶ [53], by processing spatial data using latent diffusion model (LDM)).
As per claim 5, the combined Kreis-Yang impliedly teaches each image further comprises viewpoint data of the scene (see Kreis, ¶ [54-55]); and generating the offset data is further based on the viewpoint data (Kreis, ¶ [54-55], producing a rendered view, and Yang, Fig. 3, page 6, rendering viewpoints using the offsets addressed in claim 1). Thus, claim 5 would have been obvious over the combined references for the reason above.
As per claim 6, the combined Kreis-Yang also impliedly teaches generating the offset data comprises inputting the viewpoint data to a neural network to generate one or more spherical harmonics offset parameters (Yang, page 4, top left column, “The view-dependent appearance of each 3D Gaussian is represented via spherical harmonics (SH)”). Thus, claim 6 would have been obvious over the combined references for the reason above.
As per claim 7, the combined Kreis-Yang also teaches the neural network is a multi-layer perceptron (Kreis, ¶ [28], e.g., a multilayer perceptron (MLP) neural network).
As per claim 8, the combined Kreis-Yang substantially teaches generating the offset data comprises inputting the spatial-temporal 3D Gaussian embeddings to one or more neural networks (see Yang, page 11, section B.1, “We learn the deformation field with an MLP network Fθ : (γ(x), γ(t)) → (δx, δr, δs) (which is offset as addressed above), which maps from each coordinate of 3D Gaussians and time to their corresponding deviations in position, rotation, and scaling”). Thus, claim 8 would have been obvious over the combined references for the reason above.
As per claim 9, as addressed in claims 7 and 8 above, the combined Kreis-Yang does teach at least one of the one or more neural networks is a multi-layer perceptron.
As per claim 10, as addressed in claims 7 and 8 above, the combined Kreis-Yang does also impliedly teach each of the one or more neural networks is a multi-layer perceptron (i.e., no other kind of neural network is specified).
As per claim 11, the combined Kreis-Yang further teaches:
generating the 3D Gaussian splatting data comprises generating one or more of: one or more 3D Gaussian position parameters (Kreis, ¶ [55]); one or more 3D Gaussian scale parameters; one or more 3D Gaussian rotation parameters; and one or more 3D Gaussian opacity parameters (Kreis, ¶ [55]); and
generating the offset data comprises inputting one or more of:
the one or more 3D Gaussian position parameters to a neural network to generate one or more position offset parameters (Yang, Fig. 2, page 3, “We use the position (detached) of 3D Gaussians γ(sg(x)) and time γ(t) with positional encoding as input to a deformation MLP network to obtain the offset (δx, δr, δs) of dynamic 3D Gaussians in canonical space”);
the one or more 3D Gaussian scale parameters to a neural network to generate one or more scale offset parameters;
the one or more 3D Gaussian rotation parameters to a neural network to generate one or more rotation offset parameters; and
the one or more 3D Gaussian opacity parameters to a neural network to generate one or more opacity offset parameters.
Thus, claim 11 would have been obvious over the combined references for the reason above.
As per claim 17, as addressed in claim 8, the combined Kreis-Yang does also teach wherein generating the offset data comprises inputting the spatial-temporal 3D Gaussian embeddings to a single neural network to generate the offset data (Kreis does teach the system can use single neural network as described in ¶ [56]).
Claim 18, which is similar in scope to claim 1 as addressed above, is thus rejected under the same rationale.
Claim 19, which is similar in scope to claim 1 as addressed above, is thus rejected under the same rationale.
Claims 3, 12-16 are rejected under 35 U.S.C. 103 as being unpatentable over Kreis et al. (US. Patent App. Pub. No. 2025/0182404) in view of Yang et al. (“Deformable 3D Gaussians for High-Fidelity Monocular Dynamic Scene Reconstruction”) further in view of Yu et al. (US. Patent App. Pub. No. 2025/0315923, “Yu”).
As per claim 3, the combined Kreis-Yang does not explicitly teach wherein generating the 3D Gaussian splatting data comprises generating the 3D Gaussian splatting data using 3D point cloud reconstruction.
However, in a very similar method of processing video using 3D Gaussian splatting (see Abstract), Yu teaches the above feature, i.e., generating the 3D Gaussian splatting data comprises generating the 3D Gaussian splatting data using 3D point cloud reconstruction (¶ [23], “For example, for each video clip, the 3D point clouds for foreground moving objects, derived from Equation 1, serve as initialization for optimizing the first set of 3DGS for each clip. This process may yield a set of 3D Gaussian”).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the method as taught by Yu to the combined method of Kreis and Yang as addressed above, the advantage of which is to optimize Gaussian splatting for each clip (¶ [23]).
As per claim 12, although not explicitly taught by the combined Kreis-Yang, Yu does teach generating the 3D Gaussian splatting data comprises:
identifying, within the spatial data of each image:
foreground spatial data relating a foreground of the scene; and background spatial data relating a background of the scene (Yu, ¶ [16], “For each video clip, two sets of 3D gaussian splatters are used to represent the foreground and background 3D points, respectively”. See also ¶ [18]);
generating, based on the background spatial data, background 3D Gaussian splatting data (Yu, ¶ [23]); and
generating, based on the foreground spatial data, foreground 3D Gaussian splatting data (Yu, ¶ [24]). Thus, claim 12would have been obvious over the combined references for the reason above.
As per claim 13, as addressed in claims 1 and 12 above referring to Yu, ¶ [23-24], the combined Kreis-Yang-Yu substantially teach generating the spatial-temporal 3D Gaussian embeddings comprises:
generating background spatial-temporal 3D Gaussian embeddings based on the temporal data of each image and the background 3D Gaussian splatting data; and
generating foreground spatial-temporal 3D Gaussian embeddings based on the temporal data of each image and the foreground 3D Gaussian splatting data.
Thus, claim 13 would have been obvious over the combined references for the reason above.
As per claim 14, as also addressed above in claims 1 and 12 above, the combined Kreis-Yang-Yu does impliedly teach generating the offset data comprises:
generating background offset data based on the background spatial-temporal 3D Gaussian embeddings; and generating foreground offset data based on the foreground spatial-temporal 3D Gaussian embeddings (i.e., utilizing the offset generation based on the spatial-temporal 3D Gaussian embeddings addressed in claim 1 taught by Kreis-Yang, and applying to the background and foreground respectively of the image taught by Yu). Thus, claim 14 would have been obvious over the combined references for the reason above.
As per claim 15, as addressed in claims 1, 12-14 above, the combined Kreis-Yang-Yu does impliedly teach rendering the video of the scene comprises:
combining the background 3D Gaussian splatting data with the background offset data to generate spatial-temporal 3D Gaussian representations of the background of the scene;
combining the foreground 3D Gaussian splatting data with the foreground offset data to generate spatial-temporal 3D Gaussian representations of the foreground of the scene; and
rendering the video of the scene by inputting the spatial-temporal 3D Gaussian representations of the background and the foreground of the scene to the rasterizer.
Thus, claim 15 would have been obvious over the combined references for the reason above.
As per claim 16, for at least the above reasons, the combined Kreis-Yang-Yu does also impliedly teach generating the background offset data comprises inputting the background spatial-temporal 3D Gaussian embeddings to a single neural network to generate the background offset data (see claims 1, 8, and 12); and generating the foreground offset data comprises inputting the foreground spatial-temporal 3D Gaussian embeddings to a single neural network to generate the foreground offset data (see claims 1, 8, and 12. Kreis does teach the system can use single neural network as described in ¶ [56]). Thus, claim 16 would have been obvious over the combined references for the reason above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Hau H. Nguyen whose telephone number is: 571-272-7787. The examiner can normally be reached on MON-FRI from 8:30-5:30.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Tammy Goddard, can be reached on (571) 272-7773.
The fax number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
/HAU H NGUYEN/Primary Examiner, Art Unit 2611