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 .
Response to Arguments
The objections to the Drawings is withdrawn in view of the amendment filed on 8/21/2026.
The objections to the specification is withdrawn in view of the amendment filed on 8/21/2026.
The objections to claim 12 is withdrawn in view of the amendment filed on 8/21/2026.
Applicant’s arguments in view of the claim amendments filed on 8/21/2026, with respect to the 35 U.S.C. §112 rejections have been fully considered and are persuasive. The 35 U.S.C. §112 rejections have been withdrawn.
Applicant’s arguments in light of the amendment filed 8/21/2026, with respect to the 35 U.S.C. §103 rejections have been fully considered and are persuasive. Therefore, the 35 U.S.C. §103 rejections have been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Block-NeRF: Scalable Large Scene Neural View Synthesis to Tancik et al. (previously cited), Photographic Visualization of Weather Forecasts with Generative Adversarial Networks to Sigg et al., BERT for Joint Intent Classification and Slot Filling to Chen et al. (previously cited), and Instruct-NeRF2NeRF: Editing 3D Scenes with Instructions to Haque et al. (previously cited).
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 9, 17 and 22 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 9, 17 and 22 each recite the limitation "predicting future values ... based on current values for the one or more conditions at the location". Independent claims 1, 15 and 20, which each of the aforementioned claims depend on respectively, already introduce “current values”, so the reintroduction in the dependent claims makes it unclear whether the same or different current values are meant. To expedite prosecution, Examiner interpret “current values” in the dependent claims to refer back to ‘the current values’ from the independent claims.
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.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 2, 4, 9-11, 13, 15-17 and 20-22 are rejected under 35 USC 103 as being unpatentable over Block-NeRF: Scalable Large Scene Neural View Synthesis to Tancik et al. (hereinafter Tancik) in view of Photographic Visualization of Weather Forecasts with Generative Adversarial Networks to Sigg et al. (hereinafter Sigg).
Per claim 1, Tancik discloses A computer platform for generating a video (Tancik: p. 1, Abstract…Tancik describes a computer-implemented neural rendering system that reconstructs a city-scale environment and renders it, which constitutes the computer platform limitation under BRI, "We present Block-NeRF, a variant of Neural Radiance Fields that can represent large-scale environments"; p. 1, Abstract…the rendered output is published as video, which constitutes generating a video under BRI, "Video results can be found on the project website waymo.com/research/block-nerf"), comprising:
one or more memories configured to store instructions (Tancik: p. 1, FIG. 1…each Block-NeRF must fit into memory, and rendering a trained Block-NeRF is a computation executed from that memory, so Tancik's system necessarily includes memory holding the model and the instructions that render it, which constitutes the one or more memories configured to store instructions limitation under BRI, "Block-NeRF is a method that enables large-scale scene reconstruction by representing the environment using multiple compact NeRFs that each fit into memory"); and
one or more processors configured to execute the instructions to perform operations (Tancik: p. 9, §6…Tancik measures rendering as a computation taking up to seconds per image, and a computation of that kind necessarily runs on a processor executing the rendering instructions, which constitutes the one or more processors configured to execute the instructions limitation under BRI, "In many applications, real-time rendering is key, but NeRFs are computationally expensive to render (up to multiple seconds per image)"; p. 2, §1…Tancik names the compute constraint directly, "Reconstructing such large-scale environments introduces additional challenges, including the presence of transient objects (cars and pedestrians), limitations in model capacity, along with memory and compute constraints"), the operations comprising:
receiving a query from a user relating to a location (Tancik: p. 2, FIG. 2…Tancik renders a requested target view in the reconstructed neighborhood, and the requested target view identifies the place in the environment to be rendered, which constitutes the query relating to a location under BRI, "To render a target view in the scene, the visibility maps are computed for all of the NeRFs within a given radius"; p. 5, §4.3.1…the renderer selects the blocks to render from the given target viewpoint, confirming that the target viewpoint is received as an input to the system, "For efficiency, we utilize two filtering mechanisms to only render relevant blocks for the given target viewpoint");
in response to receiving the query, generating conditioning parameters based at least in part on the query, wherein the conditioning parameters provide values for one or more conditions associated with a scene to be rendered at the location (Tancik: p. 6, §4.3.3…Tancik sets the appearance latent codes at render time, after training, and matches them across the blocks selected for the requested view, which constitutes generating conditioning parameters in response to the query under BRI, "The appearance of our learned models can be controlled by an appearance latent code after the Block-NeRF has been trained"; p. 6, §4.3.3…the codes are fitted to a target appearance for the blocks that will be rendered, "Given a target appearance in one of the Block-NeRFs, we aim to match its appearance in the remaining blocks"; p. 2, §1…for the requested view Tancik generates the appearance embeddings that condition each rendered block, which constitutes generating conditioning parameters in response to and based on the query under BRI, "To allow for more seamless compositing, we propose an appearance matching technique which brings different Block-NeRFs into visual alignment by optimizing their appearance embeddings"; p. 5, FIG. 4…the appearance codes carry values for the lighting and weather conditions of the rendered scene, which constitutes the values for one or more conditions associated with a scene limitation under BRI, "The appearance codes allow the model to represent different lighting and weather conditions"; p. 7, §5.2…the numeric exposure input is a further condition value that is set at inference time, "Finally, the exposure input marginally improves the reconstruction, but more importantly provides us with the ability to change the exposure during inference");
generating, using a generative machine-learned model, the video, wherein the video depicts the scene at the location and with the values for the one or more conditions… (Tancik: p. 3, FIG. 3…the Block-NeRF MLPs synthesize the color of each rendered point from the exposure level and appearance embedding, so the output frames depict the scene with the supplied condition values, which constitutes generating the video using a generative machine-learned model under BRI, "The network also outputs a feature vector that is concatenated with viewing direction d, the exposure level, and an appearance embedding. These are fed into a second MLP fc that outputs the color for the point"; p. 8, §5.4…Tancik renders flythrough video of the location, "For our flythrough video results, we opt for 2D inverse distance weighting (IDW) as it produces temporally consistent results"); …
providing the video for presentation to the user (Tancik: p. 1, Abstract…Tancik makes the rendered video available for viewing, which constitutes providing the video for presentation to the user under BRI, "Video results can be found on the project website waymo.com/research/block-nerf").
Tancik does not expressly disclose, but Sigg does teach:
…and wherein generating the conditioning parameters comprises retrieving current values for the one or more conditions at the location, and wherein the current values comprise values detected based on real-time data (Sigg: p. 4, §1.3…Sigg's generator is conditioned on the current image I0 of the site and the NWP model's initial state w0, both obtained for the present time, which constitutes retrieving current values for the one or more conditions under BRI, "Instead of synthesizing images from a random input z, G3(I0, z|w0, wt) transforms the current image I0 into Ît, based on the initial state w0 and the forecast wt of the NWP model"; p. 10…the weather-descriptor values are taken at the camera's location and, in operation, are the current forecast values, which constitutes current values for the conditions at the location under BRI, "The weather descriptor w consists of the time of day, day of year and a subset of the hourly forecast fields provided by the COSMO-1 model (see Schättler et al. [2021] and Table 1), evaluated at the location of the camera"; p. 10…in operation the generator is fed the current forecast rather than archival analysis data, "For an operational use of the trained generator, where the analysis of the future is of course not available, one would use the current forecast for lead time t instead, without having to modify the visualization method in any other way"; p. 1, §1…the current image comes from cameras on observation sites, and such images show the present weather at the site, "Outdoor webcam images visualize past and present weather conditions, and are consulted by meteorologists and the general public alike, e.g. in aviation weather nowcasting or the planning of a leisure outdoor activity"; p. 10…the cameras register changes in weather at a 10-minute resolution, so the current image is a value detected from live observation, which constitutes values detected based on real-time data under BRI, "The camera location can thus be atypical for the NWP model grid point, and changes in weather conditions can be visible at the 10 min resolution of the camera before or after they occur in the model parameters").
Tancik and Sigg are analogous art because they are from the same field of endeavor, specifically machine-learned synthesis of photorealistic imagery of a real-world location under controllable environmental conditions such as weather and illumination. They address the same problem of making a learned image generator depict a real place under specified weather and lighting conditions.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to supply the condition inputs of Tancik's weather- and exposure-conditioned Block-NeRF renderer from current condition values retrieved for the rendered location from real-time observations, as Sigg does for its weather-conditioned generator, so that the video of the queried location depicts the location as it currently is.
The suggestion/motivation for doing so would have been provided by Tancik itself, which teaches that its conditioned renderer can change the depicted weather, lighting and time of day, "Beyond basic view synthesis, scene conditioned NeRFs are also capable of changing environmental lighting conditions such as camera exposure, weather, or time of day, which can be used to further augment simulation scenarios" (Tancik: p. 1, §1), and by Sigg, which teaches that imagery showing the present weather conditions at a place is information the public actually consults for decisions such as planning outdoor activity, "Outdoor webcam images visualize past and present weather conditions, and are consulted by meteorologists and the general public alike, e.g. in aviation weather nowcasting or the planning of a leisure outdoor activity" (Sigg: p. 1, §1). Sigg further teaches delivering such generated location imagery to users through apps, "Meteorological services could use them to communicate localized forecasts over their own webcam feeds, smartphone apps and other distribution channels" (Sigg: p. 2, §1). Furthermore, this is the use of a known technique (conditioning a learned image generator on current, observation-derived weather values for a location, per Sigg) to improve a similar device (Tancik's condition-controllable renderer) in the same way, yielding predictable results, the rationale of MPEP 2143(I)(C), with a reasonable expectation of success because Tancik's model accepts condition inputs at inference and its appearance codes can be fitted to a target appearance after training, "The appearance of our learned models can be controlled by an appearance latent code after the Block-NeRF has been trained" (Tancik: p. 6, §4.3.3), and "Finally, the exposure input marginally improves the reconstruction, but more importantly provides us with the ability to change the exposure during inference" (Tancik: p. 7, §5.2).
Per claim 2, Tancik combined with Sigg discloses claim 1. Tancik further teaches the generative machine-learned model comprises a neural radiance field (NeRF) (Tancik: p. 1, Abstract…Block-NeRF is itself a neural radiance field variant, which constitutes the NeRF limitation under BRI, "We present Block-NeRF, a variant of Neural Radiance Fields that can represent large-scale environments"; p. 4, §3…Tancik builds its model on NeRF and mip-NeRF, "We build upon NeRF [42] and its extension mip-NeRF [3]").
Per claim 4, Tancik combined with Sigg discloses claim 1. Tancik further teaches generating, using the generative machine-learned model, the video comprises conditioning the generative machine-learned model with the conditioning parameters (Tancik: p. 3, FIG. 3…the exposure level and appearance embedding are concatenated into the color network's input, which constitutes conditioning the generative machine-learned model with the conditioning parameters under BRI, "The network also outputs a feature vector that is concatenated with viewing direction d, the exposure level, and an appearance embedding. These are fed into a second MLP fc that outputs the color for the point").
Per claim 9, Tancik combined with Sigg discloses claim 1. Sigg further teaches generating the conditioning parameters further comprises predicting future values for the one or more conditions based on current values for the one or more conditions at the location… (Sigg: p. 4, §1.3…Sigg's generator takes the current image and the NWP model's initial state for the present time and produces the image for a future lead time t, so the future depiction is computed from present condition values, which constitutes predicting future values for the one or more conditions based on current values under BRI, "Instead of synthesizing images from a random input z, G3(I0, z|w0, wt) transforms the current image I0 into Ît, based on the initial state w0 and the forecast wt of the NWP model"; p. 1, Abstract…the generator is conditioned on both the analysis and the forecasting state of the numerical weather prediction model, "The generator network, conditioned on the analysis and the forecasting state of the numerical weather prediction (NWP) model, transforms the present camera image into the future"). The rationale to combine Sigg with Tancik is the same as the parent claim.
Per claim 10, Tancik combined with Sigg discloses claim 1. Tancik further teaches generating, using the generative machine-learned model, the video comprises: generating a series of camera poses based at least in part on the query; and rendering, respectively from the series of camera poses, a series of images of the scene at the location and with the values for the one or more conditions (Tancik: p. 8, §5.4…a flythrough video is rendered frame by frame from successive camera viewpoints along a path through the requested area, which constitutes generating a series of camera poses and rendering a respective series of images under BRI, "For our flythrough video results, we opt for 2D inverse distance weighting (IDW) as it produces temporally consistent results"; p. 6, §5.1…each rendered image is produced by casting rays from a given camera pose, "Using this information, we calculate the corresponding camera ray origins and directions in a common coordinate system, also accounting for the rolling shutter of the cameras"; p. 3, FIG. 3…each rendered point takes the exposure level and appearance embedding as inputs, so each image carries the condition values, "The network also outputs a feature vector that is concatenated with viewing direction d, the exposure level, and an appearance embedding. These are fed into a second MLP fc that outputs the color for the point").
Per claim 11, Tancik combined with Sigg discloses claim 1. Tancik further teaches the computer platform comprises a database configured to store a plurality of generative machine-learned models respectively associated with a plurality of different locations; and generating, using the generative machine-learned model, the video comprises retrieving, from among the plurality of generative machine-learned models, the generative machine-learned model associated with the location (Tancik: p. 2, FIG. 2…each Block-NeRF is a separately trained model tied to its own origin coordinate, which constitutes a plurality of generative machine-learned models respectively associated with a plurality of different locations under BRI, "The scene is split into multiple Block-NeRFs that are each trained on data within some radius (dotted orange line) of a specific Block-NeRF origin coordinate (orange dot)"; p. 2, §1…the independently stored blocks can be updated or added region by region, which constitutes a stored collection of per-location models, "Modeling these Block-NeRFs independently allows for maximum flexibility, scales up to arbitrarily large environments and provides the ability to update or introduce new regions in a piecewise manner without retraining the entire environment as demonstrated in Figure 1"; p. 2, §1…only the blocks associated with the target view's location are retrieved and rendered, which constitutes retrieving the generative machine-learned model associated with the location under BRI, "To compute a target view, only a subset of the Block-NeRFs are rendered and then composited based on their geographic location compared to the camera").
Per claim 13, Tancik combined with Sigg discloses claim 1. Tancik further teaches the generative machine-learned model has been trained on a training dataset comprising a plurality of reference images of the location, and the training dataset comprises values for the one or more conditions for at least some of the plurality of reference images (Tancik: p. 7, §5.1…each Block-NeRF is trained on a large set of captured images of its neighborhood, which constitutes a training dataset comprising a plurality of reference images of the location under BRI, "The overall dataset is composed of 13.4 h of driving time sourced from 1,330 different data collection runs, with a total of 2,818,745 training images"; p. 6, §5.1…each captured training image carries its exposure value, which constitutes values for the one or more conditions for at least some of the plurality of reference images under BRI, "Each camera captures images at 10 Hz and stores a scalar exposure value").
Claims 15 and 16 are substantially similar in scope and spirit as claims 1 and 2, respectively. Therefore, the rejections of claims 1 and 2 are applied accordingly.
Per claim 17, Tancik combined with Sigg discloses claim 15. Sigg further teaches generating the conditioning parameters further comprises: predicting future values for the one or more conditions based on current values for the one or more conditions at the location… (Sigg: p. 4, §1.3…the generator produces the future-lead-time depiction from the current image and the model's initial state for the present time, which constitutes predicting future values based on current values under BRI, "Instead of synthesizing images from a random input z, G3(I0, z|w0, wt) transforms the current image I0 into Ît, based on the initial state w0 and the forecast wt of the NWP model"). The rationale to combine Sigg with Tancik is the same as the parent claim.
Claim 20 is substantially similar in scope and spirit as claim 1. Therefore, the rejection of claim 1 is applied accordingly. Sigg further teaches A non-transitory computer readable medium storing instructions which, when executed by a processor, cause the processor to perform operations for generating a video (Sigg: p. 16…Sigg's generator exists as program code held in a repository, and code kept in a repository is necessarily recorded on a persistent storage medium from which a processor loads and executes it, which constitutes the non-transitory computer readable medium storing instructions limitation under BRI, "A Tensorflow implementation of the generator and discriminator networks, as well as the code to reproduce the experiments presented in Sec. 4, is available in the companion repository"; p. 6, §2.1…the implementation runs on GPU hardware whose memory holds the network, "After a Conv-BN-ReLU preprocessing block, there are five encoder stages E1, E2, . . . , E5 that progressively halve the input height and width and double the depth, except for the last stage, where the depth is not doubled anymore to conserve GPU memory"). The rationale to combine Sigg with Tancik is the same as for claim 1.
Claims 21 and 22 are substantially similar in scope and spirit as claims 2 and 9, respectively. Therefore, the rejections of claims 2 and 9 are applied accordingly.
Claims 5-8 and 19 are rejected under 35 USC 103 as being unpatentable over Tancik in view of Sigg, as applied in the rejection of claims 1 and 15 above, and further in view of BERT for Joint Intent Classification and Slot Filling to Chen et al. (hereinafter Chen).
Per claim 5, Tancik combined with Sigg discloses claim 1. Tancik combined with Sigg does not expressly disclose, but Chen does teach generating the conditioning parameters further comprises extracting the values for the one or more conditions from the query (Chen: §1…Chen's slot filling extracts the parameter values stated in a user query as labeled slots, which, applied to the values for the conditions stated in the location query, constitutes extracting the values from the query under BRI, "Intent classification focuses on predicting the intent of the query, while slot filling extracts semantic concepts. Table 1 shows an example of intent classification and slot filling for user query “Find me a movie by Steven Spielberg”").
Per claim 6, Tancik combined with Sigg discloses claim 1. Tancik combined with Sigg does not expressly disclose, but Chen does teach generating the conditioning parameters further comprises inferring the values for the one or more conditions from the query (Chen: §4.4…Chen's model assigns slot values that are not explicitly marked in the query by drawing on knowledge learned in pre-training, which constitutes inferring the values from the query under BRI, "However, joint BERT correctly predicts the slot labels and intent because “mother joan of the angels” is a movie entry in Wikipedia. The BERT model was pre-trained partly on Wikipedia and possibly learned this information for this rare phrase").
Per claim 7, Tancik combined with Sigg and Chen discloses claim 6. Chen further teaches inferring the values for the one or more conditions comprises providing the query to a sequence processing model, wherein the sequence processing model is configured to output the values for the one or more conditions in response to the query (Chen: §1…Chen's model takes the query's word sequence as input and outputs a sequence of slot labels, one per word, which constitutes a sequence processing model configured to output the values in response to the query under BRI, "Intent classification is a classification problem that predicts the intent label yi and slot filling is a sequence labeling task that tags the input word sequence x = (x1, x2, · · · , xT) with the slot label sequence ys = (ys1, ys2, · · · , ysT)"; Abstract…the model that does this is BERT, "In this work, we propose a joint intent classification and slot filling model based on BERT").
Per claim 8, Tancik combined with Sigg discloses claim 1. Tancik combined with Sigg does not expressly disclose, but Chen does teach implementing one or more large language models to determine a plurality of variables based on the query, wherein the plurality of variables comprise an underlying structure of a language of the query (Chen: Abstract…BERT is a large pre-trained language model, which constitutes the one or more large language models limitation under BRI, "Recently a new language representation model, BERT (Bidirectional Encoder Representations from Transformers), facilitates pre-training deep bidirectional representations on large-scale unlabeled corpora, and has created state-of-the-art models for a wide variety of natural language processing tasks after simple fine-tuning"; §4.1…BERT is pre-trained on billions of words of text, "BERT is pre-trained on BooksCorpus (800M words) (Zhu et al., 2015) and English Wikipedia (2,500M words)"; §1…from each query the model determines an intent plus a slot label for every word, which together describe how the query's language is organized and constitute a plurality of variables comprising an underlying structure of a language of the query under BRI, "Intent classification is a classification problem that predicts the intent label yi and slot filling is a sequence labeling task that tags the input word sequence x = (x1, x2, · · · , xT) with the slot label sequence ys = (ys1, ys2, · · · , ysT)"; Abstract…Chen measures the result as the accuracy of the whole semantic frame of the sentence, confirming that the variables capture the sentence's structure, "Experimental results demonstrate that our proposed model achieves significant improvement on intent classification accuracy, slot filling F1, and sentence-level semantic frame accuracy on several public benchmark datasets, compared to the attention-based recurrent neural network models and slot-gated models").
Claim 19 is substantially similar in scope and spirit as claim 6. Therefore, the rejection of claim 6 is applied accordingly.
As it pertains to Claims 5-8 and 19, Tancik, Sigg and Chen are analogous art because Chen is reasonably pertinent to the particular problem of converting a user's request into the structured parameter values consumed by a downstream machine-learned model, which is the problem Tancik and Sigg's condition-driven generators face when their condition inputs are to come from a user query; Tancik and Sigg are from the same field of endeavor as set forth for claims 1 and 15.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to place Chen's BERT intent-and-slot model in front of the Tancik and Sigg renderer so that the location, the condition values and the query structure are extracted or inferred from a natural-language query.
The suggestion/motivation for doing so would have been provided by Chen itself, which teaches that natural-language understanding is what makes user-facing request systems work, "Natural language understanding (NLU) is critical to the performance of goal-oriented spoken dialogue systems" (Chen: §1), and that a BERT-based model is a state-of-the-art way to do it, "Recently a new language representation model, BERT (Bidirectional Encoder Representations from Transformers), facilitates pre-training deep bidirectional representations on large-scale unlabeled corpora, and has created state-of-the-art models for a wide variety of natural language processing tasks after simple fine-tuning" (Chen: Abstract). Furthermore, this is the combination of known elements according to known methods (a language-understanding front end and a condition-driven generative back end) yielding the predictable result of renderings requested in natural language, the rationale of MPEP 2143(I)(A).
Claim 12 is rejected under 35 USC 103 as being unpatentable over Tancik in view of Sigg, as applied in the rejection of claim 1 above, and further in view of Instruct-NeRF2NeRF: Editing 3D Scenes with Instructions to Haque et al. (hereinafter Haque).
Per claim 12, Tancik combined with Sigg discloses claim 1. Tancik combined with Sigg does not expressly disclose, but Haque does teach the query comprises a text query that specifies one or more objects to be included in the scene and wherein the video depicts the one or more objects included in the scene (Haque: p. 1, FIG. 1…Haque edits a NeRF scene from a text instruction such as adding a cowboy hat, which constitutes a text query that specifies one or more objects to be included in the scene under BRI, "We propose Instruct-NeRF2NeRF, a method for consistent 3D editing of a NeRF scene using text-based instructions"; p. 1, Abstract…the scene is re-optimized so that renderings respect the instruction, "Given a NeRF of a scene and the collection of images used to reconstruct it, our method uses an image-conditioned diffusion model (InstructPix2Pix) to iteratively edit the input images while optimizing the underlying scene, resulting in an optimized 3D scene that respects the edit instruction"; p. 1, Abstract…the edited scene is rendered as video, which constitutes the video depicts the one or more objects included in the scene limitation under BRI, "Result videos can be found on the project website: https://instruct-nerf2nerf.github.io").
Tancik, Sigg and Haque are analogous art because they are from the same field of endeavor, specifically machine-learned synthesis of photorealistic imagery of real-world scenes, and Haque applies it to the same neural radiance field scene representation Tancik uses.
Before the effective filing date of the claimed invention, it would have been obvious to a PHOSITA to accept a text query naming objects to add to the scene and to apply Haque's text-instructed NeRF editing to the Tancik and Sigg renderer, so that the rendered video of the location depicts the named objects.
The suggestion/motivation for doing so would have been provided by Haque itself, which teaches that text-instructed editing makes NeRF scenes easy for ordinary users to change, "Our approach makes 3D scene editing accessible and intuitive for everyday users" (Haque: p. 2, §1). Furthermore, this is the application of a known technique to a known device ready for improvement to yield predictable results, the rationale of MPEP 2143(I)(D).
Allowable Subject Matter
Claim 14 is objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
The following is the statement of reasons for the indication of allowable subject matter: The prior art disclosed by the applicant and cited by the Examiner fail to teach or suggest, alone or in combination, all the limitations of the independent claim 1, further including the particular notable limitations of receiving a further query from the user relating to the video; in response to receiving the further query, generating further conditioning parameters based at least in part on the further query, wherein the further conditioning parameters provide values for one or more further conditions associated with the scene to be rendered at the location; generating, using the generative machine-learned model, an adjusted video, wherein the adjusted video depicts the scene at the location and with the values for the one or more further conditions; and providing the adjusted video for presentation to the user.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 ALAN CHEN whose telephone number is (571)272-4143. The examiner can normally be reached M-F 10-7.
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, Kamran Afshar can be reached at (571) 272-7796. 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.
/ALAN CHEN/Primary Examiner, Art Unit 2125
