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 .
This action is in response to the Amendment filed on 5/11/2026.
Claims 1-20 are pending. Claims 1-20 have been amended.
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.
Claims 1-3, 7-13, 16-20 is rejected under 35 U.S.C. 103 as being unpatentable over Price et al. (US 9106812 B1, hereinafter Price) in view of Lupascu et al. (US 20230267652 A1, hereinafter Lupascu).
Regarding Claim 11, Price teaches a computer device (Price, Column 4, Lines 8-11, the shared computing resources 208), comprising at least one processor and a memory (Price, Column 4, Lines 8-11, the shared computing resources 208 comprise one or more processors 210 and one or more forms of computer-readable media 212 <read on memory>); wherein the memory stores machine-readable instructions executed by the at least one processor (Price, Column 4, Lines 20-26, the computer-readable media 212 may also contain an operating system for controlling software modules stored in the computer-readable media 212 and for controlling hardware associated with the shared computing resources 208. the automated storyboard producer 114 is as a network-based system discussed above), the at least one processor is configured to execute the machine-readable instructions stored in the memory (Price, Column 4, Lines 25-29, the automated storyboard producer 114 is as a network-based system discussed above, it is also possible for the automated storyboard producer 114 to be implemented as a local application running on one of the computing devices 204), and when the machine-readable instructions are executed by the at least one processor, the at least one processor executes a method of generating a comic image (Price, Column 3, Lines 3-5, the automated storyboard producer 114 may use a series of rules, logic, artificial intelligence, and machine learning to generate the storyboard 112 <read on comic image> from the screenplay 100), the method of generating the comic image comprises acquiring a novel to be used to generate a comic (Price, Column 4, Lines 34-46, the automated storyboard producer 114 may manage one or more projects 214 that each contains content which may be used to create storyboards for the respective projects...the project 214 may begin with a single screenplay 100 <read on novel>); determining keyword information corresponding to one or more comic storyboards that correspond to the novel in a plurality of comic image generation dimensions according to a content of the novel (Price, Column 14, Lines 49-58, the analysis may also include identifying a slugline 102 that includes a scene description...the character descriptions may include gender, physical characteristics, and facial expressions. For example, a description that Jack sees a tall, smiling woman provides information about one of the other characters in the scene with Jack. Furthermore, the scene description may describe a background image 118 and or a camera technique); [[ with respect to a first comic storyboard of the one or more comic storyboards, determining first model input information corresponding to the first comic storyboard according to a mapping relationship library between dimension keywords and model input information, and the keyword information of the one or more comic storyboards in the comic image generation dimensions, wherein the dimension keywords comprise a keyword that has been determined in any comic image generation dimension]]; and using an artificial intelligence model to generate a comic image corresponding to the first comic storyboard according to [[the model input information]] (Price, Column 3, Lines 3-5, the automated storyboard producer 114 may use a series of rules, logic, artificial intelligence, and machine learning to generate the storyboard 112 <read on comic image> from the screenplay 100).
Price does not explicitly disclose but Lupascu teaches
with respect to a first comic storyboard of the one or more comic storyboards, determining first model input information corresponding to the first comic storyboard according to a mapping relationship library between dimension keywords and model input information, and the keyword information of the one or more comic storyboards in the comic image generation dimensions, wherein the dimension keywords comprise a keyword that has been determined in any comic image generation dimension (Lupascu, Paragraph [0005], the disclosed systems receive style parameters as input (e.g., via a list of style images, a list of text prompts <read on keyword information>, or a combination of style images and text prompts); Paragraph (0067), the artistic content generation system 106 utilizes the multi-domain style encoder 308 to determine style encodings 320 <read on first model input information > from the style parameters <read on keyword information> associated with the style digital image 316 and the style text prompt 318; Paragraph (0069], the artistic content generation system 106 utilizes, as the multi-domain style encoder 308 <read on mapping relationship library>, an encoder that includes the cross-lingual-multimodal-embedding model and the image-embedding model...the artistic content generation system 106 utilizes, as the multi-domain style encoder 308 <read on mapping relationship library>, the Contrastive Language-Image Pre-training (CLIP) model); using an artificial intelligence model to generate a comic image corresponding to the comic storyboard according to the model input information (Lupascu, Paragraph [0072], the artistic content generation system 106 utilizes the decoder 304 of the artistic generative neural network <read on artificial intelligence model> to generate an initialized artistic digital image 324 <read on comic image> based on the learnable tensor 306 < read on first model input information >; Paragraph [0071], the artistic content generation system 106 utilizes the artistic image neural network 300 to generate an artistic digital image <read on comic image> by iteratively generating an intermediate artistic digital image from the learnable tensor 306 <read on model input information>, comparing the intermediate artistic digital image to the style parameters, and updating the learnable tensor 306 based on the comparison).
Lupascu and Price are analogous since both of them are dealing with automated generation of visual content by analyzing textual narrative content and utilizing computing systems to generate images based on extracted parameters from the narrative. Price provides an automated storyboard producer that analyzes a screenplay to extract multi-dimensional storyboard content (character descriptions, scene descriptions, background, camera technique) and assembles storyboard frames using AI/ML and libraries of visual assets. Lupascu provided a way of mapping text-based style parameters (text prompts) to neural network model input encodings using a multi-domain style encoder that projects keywords from different domains (text and images) into a common encoding space, and then using an artificial intelligence generative neural network to generate artistic visual content based on these mapped encodings. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the multi-domain style encoder mapping technique taught by Lupascu into the modified invention of Price such that when generating storyboard images from a screenplay (novel), the system would use a mapping relationship library (multi-domain encoder) to convert the extracted keyword information from various dimensions (character, scene, props, etc.) into precise model input information (style encodings) for the artificial intelligence model. The motivation is to improve the accuracy and consistency of automatically generated visual content by utilizing a more sophisticated mapping mechanism that can handle multiple types of textual keyword information from different dimensions and convert them into optimized neural network inputs, thereby enabling the Al model to generate comic images that more accurately reflect the fine-grained details specified in the keyword information across the plurality of comic image generation dimensions.
Regarding Claim 12, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein the keyword information comprises at least one keyword (Lupascu, Paragraph [0005], the disclosed systems receive style parameters as input (e.g., via a list of style images, a list of text prompts <read on keyword information, keyword>); the mapping relationship library is used for storing different mapping relationships between the dimension keywords and the model input information (Lupascu, Paragraph [0067], the artistic content generation system 106 utilizes the multi-domain style encoder 308 to determine style encodings 320 <read on model input information> from the style parameters <read on dimension keywords> associated with the style digital image 316 and the style text prompt 318; Paragraph [0069], the artistic content generation system 106 utilizes, as the multi-domain style encoder 308 <read on mapping relationship library>, an encoder that includes the cross-lingual-multimodal-embedding model and the image-embedding model...the artistic content generation system 106 utilizes, as the multi-domain style encoder 308, the Contrastive Language-Image Pre-training (CLIP) model); the determining first model input information corresponding to the comic storyboard according to a mapping relationship library between dimension keywords and model input information, and the keyword information of the comic storyboards in the comic image generation dimensions, comprises: when a mapping relationship that matches the keyword comprised in the keyword information is searched from the mapping relationships included in the mapping relationship library, taking model input information indicated by the mapping relationship as the model input information (Lupascu, Paragraph [0067], the artistic content generation system 106 utilizes the multi-domain style encoder 308 to determine style encodings 320 <read on model input information> from the style parameters <read on keyword information> associated with the style digital image 316 and the style text prompt 318; Paragraph [0077] , based on the comparison, the artistic content generation system 106 updates the learnable tensor 306 <read on selecting model input information> and generates an artistic digital image 212 utilizing the artistic image neural network 300).
As explained in rejection of claim 11, the obviousness for combining the multi-domain style encoder mapping technique of Lupascu into the automated storyboard generation system of Price is provided above.
Regarding Claim 13, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein the mapping relationship is searched as following steps: taking a mapping relationship corresponding to a dimension keyword that is consistent with the keyword in the mapping relationship library as the mapping relationship (Lupascu, Paragraph [0067], the artistic content generation system 106 utilizes the multi-domain style encoder 308 to determine style encodings 320 from the style parameters associated with the style digital image 316 and the style text prompt 318, the artistic content generation system 106 utilizes the neural network image encoder 310 to project the style digital image 316 into the multi-domain encoding space and utilizes the neural network text encoder 312 to project the style text prompt 318 into the multi-domain encoding space <read on taking a mapping relationship corresponding to a dimension keyword that is consistent with the keyword>); or determining the mapping relationship according to correlation degrees between keyword semantics of the dimension keywords of the mapping relationships in the mapping relationship library and a semantic of the keyword (Lupascu, Paragraph [0069], the artistic content generation system 106 utilizes, as the multi-domain style encoder 308, the Contrastive Language-Image Pre-training (CLIP) model described by Alec Radford et al., Learnable Transferable Visual Models from Natural Language Supervision, ICML, 2021...the artistic content generation system 106 utilizes the multi-domain style encoder 308 to project the style digital image 316 and the style text prompt 318 into a multi-domain encoding space <read on determining the mapping relationship according to correlation degrees between keyword semantics and a semantic of the keyword>).
As explained in rejection of claim 11, the obviousness for combining the multi-domain style encoder mapping technique of Lupascu into the automated storyboard generation system of Price is provided above.
Regarding Claim 16, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein the determining keyword information corresponding to one or more comic storyboards that correspond to the novel in a plurality of comic image generation dimensions according to a content of the novel, comprises: determining occurrence numbers of each storyboard scene corresponding to the novel according to the content of the novel (Price, Column 9, Lines 47-54, a user 202 1 that provided a screenplay 100 <read on novel> may have used the automated storyboard producer 114 multiple times in the past with other screenplays 100 having similar scene descriptions <read on storyboard scene> e.g., hillside, hill, knoll, mound, etc. and the previous selections by the user 202 1 may indicate which hillside image from the background image library 220 he or she picks most often <read on occurrence numbers of each storyboard scene>; Column 9, Lines 40-43, past behavior of users 202 interacting with the shared computing resources 208 may be analyzed to identify the most frequently selected <read on occurrence numbers> background image for a hillside); and determining the keyword information separately corresponding to the one or more comic storyboards that corresponds to the novel in the plurality of the comic image generation dimensions according to the occurrence numbers, wherein an information amount of the keyword information is positively correlated with the occurrence number (Price, Column 7, Lines 63-66, the text analytics module 304 may include a set of linguistic, statistical, and machine learning techniques that model and structure the information content of textual content from the screenplay 100; Column 8, Line 1-4, Specific textual analysis techniques may include lexical analysis to study word frequency distributions <read on information amount positively correlated with occurrence number>, pattern recognition, tagging/annotation, information extraction, data mining techniques; Column 14, Lines 10-20, the descriptions of scenes in the sluglines 102, the action 104, and the dialogue 106 provide a basis from which the screenplay 100 is used to create a storyboard 112...the analysis may leverage the formatting and style conventions of scripts to enhance automatic recognition of script elements and use text analytics to assign meanings to words in the script <read on determining keyword information in multiple dimensions according to frequency/occurrence>).
Regarding Claim 17, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein after generating the comic images corresponding to the one or more comic storyboards, the method further comprises: determining storyboard frames corresponding to the one or more comic storyboards according to the keyword information corresponding to each of the one or more comic storyboards (Price, Column 3, Lines 52-60, FIG. 1 description, one frame of the storyboard 112 may include character images 116 for each of the characters mentioned in the corresponding scene from the screenplay 100...the background 118 may be selected based on analysis of the slugline 102, the action 104, and or the dialogue 106 from the screenplay 100 <read on determining storyboard frames according to keyword information>); filling the comic images corresponding to each of the one or more comic storyboards in the storyboard frames to obtain storyboard images corresponding to each of the one or more comic storyboards (Price, Column 3, Lines 60-67, the character images 116 may be placed in front of a background 118...the finished draft of the storyboard 112 may include multiple frames <read on filling the comic images in storyboard frames to obtain storyboard images>); and according to a number of episodes to which each of the one or more comic storyboards belongs in a strip comic and a storyboard order of each of the one or more comic storyboards in the number of the episodes, typesetting the storyboard images corresponding to the comic storyboards to obtain a strip comic corresponding to the one or more novel (Price, Column 6, Lines 1-9, the project 214 may begin with a single screenplay 100...the finished draft of the storyboard 112 may include multiple frames...the scene on the hill is preceded by an earlier scene 120 and followed by a later scene 122 <read on ordering storyboard images in sequence>; Column 2, Lines 24-30, storyboards are graphic organizers that show a series of illustrations or images displayed in sequence for the purpose of pre-visualizing some type of video media <read on typesetting storyboard images to obtain a strip comic>).
Regarding Claim 18, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein the one or more comic storyboards corresponding to the novel is determined as following steps: splitting the novel according to information semantics of text information corresponding to the one or more novel, and obtaining novel segments corresponding to the novel {Price, Column 15, Lines 1-10, a screenplay analysis module 302 may analyze a screenplay 100 to extract information both from the content of the text and the layout and formatting of the text...the text analytics module 304 may include a set of linguistic, statistical, and machine learning techniques that model and structure the information content of textual content from the screenplay 100); and determining the comic storyboards corresponding to the one or more novel according to segmented texts of the novel segments and the comic image generation dimensions (Price, Column 3, Lines 64-67, a storyboard or animatic may be created automatically from a script based on automatic analysis of the script; Column 15, Lines 10-18, the screenplay analysis module 302 identifies sections of dialogue 106, sections of the action 104, and relative locations of the sections of dialogue and the sections of action on the pages of the screenplay 100).
Regarding Claim 19, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination further teaches wherein the comic image generation dimensions are determined as following steps: determining a novel genre of the novel (Price, Column 10, Lines 38-44, The preproduction module 318 may receive an indication of a genre for the screenplay 100 <read on novel genre of the novel> that is used by the automated storyboard producer 114 to select elements of the storyboard 112 (e.g., a soundtrack); Column 6, Lines 41-44, the songs available in the music library 222 may be grouped by song genre e.g., rock n roll, blues, etc., corresponding movie genre <read on novel genre> e.g., Western, film noir, etc.); and determining the plurality of the comic image generation dimensions corresponding to the novel from a plurality of preset image generation dimensions according to the novel genre (Price, Column 7, Lines 4-11, the camera technique library 226 may include multiple templates for a given type of camera technique...the templates may be organized by director, genre, or other category <read on determining plurality of dimensions according to novel genre>. For example, there may be a template that includes camera techniques <read on comic image generation dimensions> which are suggestive of a television soap opera or a movie directed by Woody Allen <read on selecting dimensions based on genre>; Column 10, Lines 38-44,, a genre for the screenplay 100 that is used by the automated storyboard producer 114 to select elements of the storyboard 112 <read on determining the plurality of dimensions from preset dimensions according to genre>).
Regarding Claim 1, it recites limitations similar in scope to the limitations of Claim 11 but as a method and the combination of Price and Lupascu teaches all the limitations as of Claim 11. Therefore is rejected under the same rationale.
Regarding Claim 2, it recites limitations similar in scope to the limitations of Claim 12 and therefore is rejected under the same rationale.
Regarding Claim 3, it recites limitations similar in scope to the limitations of Claim 13 and therefore is rejected under the same rationale.
Regarding Claim 7, it recites limitations similar in scope to the limitations of Claim 16 and therefore is rejected under the same rationale.
Regarding Claim 8, it recites limitations similar in scope to the limitations of Claim 17 and therefore is rejected under the same rationale.
Regarding Claim 9, it recites limitations similar in scope to the limitations of Claim 18 and therefore is rejected under the same rationale.
Regarding Claim 10, it recites limitations similar in scope to the limitations of Claim 19 and therefore is rejected under the same rationale.
Regarding Claim 20, it recites limitations similar in scope to the limitations of claim 11 and the combination of Price and Lupascu teaches all the limitations as of Claim 11. And Price discloses these features can be implemented on a computer readable storage medium (Price, Column 4, Line 13-20, The computer-readable media 212 may include… any other non-transitory computer-readable medium which can be used to store information and which can be accessed by a processor; Column 14, Line 9-14, In the context of software, the blocks represent computer-executable instructions stored on one or more computer-readable storage media that, when executed by one or more processors, perform the recited operations).
Claim 4, 5, 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Price et al. (US 9106812 81, hereinafter Price) in view of Lupascu et al. (US 20230267652 A1, hereinafter Lupascu) as applied to Claim 1, 11 above respectively and further in view of Hibbert et al. (US 20110102424 A1, hereinafter Hibbert).
Regarding Claim 14, the combination of Price and Lupascu teaches the invention in Claim 11.
The combination does not explicitly disclose but Hibbert teaches wherein the determining model input information corresponding to the comic storyboard according to a mapping relationship library between dimension keywords and model input information, and the keyword information of the one or more comic storyboards in the comic image generation dimensions, comprises: when the mapping relationship library does not store a mapping relationship that matches a keyword comprised in the keyword information, determining pieces of candidate input information corresponding to the keyword (Hibbert, Paragraph [0096], the system allows multiple 30 sets to be saved in a library corresponding to each storyboard project and accessible by the storyboard artist...the library provides a real-time visual preview of the set in question (utilising a default camera position) so the correct set can be easily chosen; Paragraph [0098], the system allows multiple 3D objects to be saved in a library corresponding to each storyboard project...the 3D object library provides a real-time visual preview of the model or prop in question...so the storyboard artist can easily choose the correct model or prop for any particular panel <read on when no exact match, choosing from multiple candidate options>); inputting each candidate input information into the artificial intelligence model separately to obtain images corresponding to each candidate input information (Hibbert, Paragraph [0064), step 13 the 3D image data is rendered from the viewpoint of the virtual camera to form a 20 background image; Paragraph [0009], retrieving three-dimensional image data defining at least one three-dimensional object, rendering the three-dimensional image data from a predefined viewpoint to generate two-dimensional background image data <read on obtaining images from each candidate>); and determining the first model input information from the candidate input information according to matching degrees between the images and the keyword information
PNG
media_image1.png
112
9
media_image1.png
Greyscale
{Hibbert, Paragraph [0096], the library provides a real-time visual preview of the set in question utilising a default camera position so the correct set can be easily chosen <read on determining model input based on matching>; Paragraph [0098], the storyboard artist can easily choose the correct model or prop for any particular panel).
Hibbert and Price are analogous since both of them are dealing with automated generation of visual content for storytelling purposes by analyzing textual or parametric input and utilizing computing systems to generate images based on extracted or specified parameters. Price provided a way of automatically analyzing screenplays to extract storyboard content and generating storyboard images using libraries of visual assets accessed based on extracted content. Hibbert provided a way of storing multiple candidate 30 sets and objects in libraries, rendering each candidate as a preview image from a specified viewpoint, and enabling selection of the correct visual element by comparing the rendered preview images to determine which candidate best matches the storyboard requirements. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the library-based candidate selection with preview rendering technique taught by Hibbert into the modified invention combining Price such that when the mapping relationship library does not store an exact match for keyword information, the system would determine multiple pieces of candidate input information from the library, input each candidate separately into the artificial intelligence model to render corresponding images, and determine the model input information by comparing the matching degrees between the rendered images and the keyword information requirements which yielding predictable results of enabling selection among multiple candidate visual options when an exact match is not immediately available.
Regarding Claim 4, it recites limitations similar in scope to the limitations of Claim 14 and therefore is rejected under the same rationale.
Regarding Claim 5, the combination of Price, Lupascu, and Hibbert teaches the invention in Claim 4.
The combination further teaches wherein after the determining the model input information from the candidate input information, the method further comprises: establishing a mapping relationship between the first model input information and the
keyword (Hibbert, Paragraph [0170], the system contains a user-managed look-up table that lists all the low-polygon 3D sets and 3D objects (proxies), used in any particular storyboard project. This lookup table links the low-polygon proxies (used in the system) to the location and filename of the corresponding final high-polygon model <read on establishing a mapping relationship between model input information and keyword>; Paragraph [0092], the first file is a data file containing...a unique model ID defining the model used as the
background set <read on mapping relationship between keyword/identifier and model data>);and storing the mapping relationship between the first model input information and the keywords in the mapping relationship library (Hibbert, Paragraph [0093], the system scans the storyboard folder and parses the filenames of each individual file in order to rebuild a storyboard in memory; Paragraph [0095], the timestamps in the files for each panel within that folder are checked and compared with a database in the system memory to find out which panel(s) is(are) affected. The updated data can then be retrieved from the affected panels; Paragraph [0170]-[0171], This lookup table links the low-polygon proxies (used in the system) to the location and filename of the corresponding final high-polygon model...by using the lookup table as a reference).
As explained in rejection of claim 4, the obviousness for combining the library selection and rendering preview system of Hibbert with the automated storyboard generation system of Price enhanced by the multi-domain style encoder of Lupascu is provided above.
Claim 6, 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Price et al. (US 9106812 81, hereinafter Price) in view of Lupascu et al. (US 20230267652 A1, hereinafter Lupascu) as applied to Claim 1, 11 above respectively and further in view of Sturlaugson et al. (US 20160358099 A1, hereinafter Sturlaugson).
Regarding Claim 6, the combination of Price and Lupascu teaches the invention
in Claim 1.
The combination does not explicitly disclose but Sturlaugson teaches wherein the
determining first model input information corresponding to the first comic storyboard
according to a mapping relationship library between dimension keywords and model
input information, and the keyword information of the one or more comic storyboards in the comic image generation dimensions, comprises: when the mapping relationship library does not store a mapping relationship that is related to a keyword comprised in
the keyword information (Sturlaugson, Paragraph [0103], "a machine learning algorithm library that includes a plurality of machine learning algorithms configured to be tested with a common interface"; Paragraph [0104], "a data input module configured to receive an input dataset and a selection of machine learning models")
, determining pieces of candidate input information corresponding to the keyword (Sturlaugson, Paragraph [0104], a data input module configured to receive an input dataset and a selection of machine learning models <read on determining pieces of candidate input information corresponding to the keyword>); with respect to anyone of the candidate input information, using a plurality of text-and-image conversion models separately to generate corresponding images according to the candidate input information, wherein different text-and-image conversion models are deployed with different text-and-image conversion algorithms (Sturlaugson, Paragraph [0103], a machine learning algorithm library that includes a plurality of machine learning algorithms <read on plurality of text-and-image conversion models with different algorithms> configured to be tested with a common interface; Paragraph [0003], a broad array of machine learning algorithms are available ... artificial neural networks, learned decision trees, and support vector machines are different classes of algorithms; Paragraph [0040], Experiment module 30 is configured to train each of the machine learning models 32 ... to produce a trained model for each machine learning model <read on generating corresponding images using different models with different algorithms>); and determining an image generation effect of the candidate input information according to the images, and determining the first model input information corresponding to the comic storyboard according to the image generation effect of the candidate input information respectively (Sturlaugson, Paragraph [0040], Experiment module 30 is configured to evaluate and/or to validate each trained model to produce a performance result for each machine learning model <read on determining an image generation effect according to images>; Paragraph [0106], an aggregation module configured to aggregate the performance results for all of the machine learning models to form performance comparison statistics).
Sturlaugson and Price are analogous since all three are dealing with automated
generation of output content by processing input data using computing systems that
select and apply appropriate algorithms or models based on input parameters.
Price provided a way of automatically analyzing screenplays to extract storyboard
content (including character descriptions, scene descriptions, and dialogue) and
generating storyboard images using libraries of visual assets accessed based on
extracted content. Sturlaugson provided a way of maintaining a machine learning algorithm library including a plurality of different machine learning algorithms, receiving a selection of machine learning models, training and evaluating each machine learning model separately to produce a performance result for each model, and aggregating the
performance results to form comparison statistics for selecting the best-performing
model. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the multi-model testing and comparison technique taught by Sturlaugson into the modified invention combining Price such that when the mapping relationship library does not store an exact match for keyword information, the system would determine multiple pieces of candidate input information, use a plurality of text-and-image conversion models with different algorithms separately to generate corresponding images for each candidate, evaluate the image generation effect of each candidate by analyzing the performance results of the generated images, and determine the model
input information by selecting the candidate that produces the best image generation
effect according to the performance comparison statistics.
Regarding Claim 15, it recites limitations similar in scope to the limitations of Claim 6 and therefore is rejected under the same rationale.
Response to Arguments
Applicant’s arguments with respect to claim 1, 11, 20, filed on 5/11/2026, with respect to rejection under 35 USC § 103 have been considered but is not persuasive.
Applicant asserts that Lupascu's 'multi-domain style encoder 308' is not a 'mapping relationship library'".
In response to the argument, Claim 1 does not define "mapping relationship library" as a discrete lookup-table data structure or as a repository separate from a neural network. The specification also does not restrict "mapping relationship library" to any particular data-structure implementation. Secondary, Lupascu's multi-domain style encoder 308 — particularly in the form of the CLIP model expressly identified in Lupascu Paragraph [0069] — is precisely such a facility. The CLIP model is not merely an "encoding layer"; it is a pretrained model whose learned parameters constitute a stored, indexable set of associations between text/image inputs (across the multiple "domains" — i.e., dimensions — of text and image) and the corresponding embedding output. A pretrained multimodal embedding model that stores learned mappings between keywords/prompts (across text and image domains) and encoding outputs is functionally a mapping relationship library within the BRI of the claim. Third, Applicant states the multi-domain encoder "takes in data in various formats ... and encodes the disparate information into a common encoding." That is precisely the functional role recited by the claim for the "mapping relationship library between dimension keywords and model input information" — receiving keywords in multiple dimensions and yielding the model input information used downstream. Hence, the reference structurally and functionally performs the claimed function. Therefore, applicant remark cannot be considered persuasive.
Applicant further asserts that Lupascu does not disclose any 'mapping relationship' between 'dimension keywords' and 'model input information.
In response to the argument, Applicant's argument draws a distinction without a technical basis. Applicant concedes that Lupascu's encoder "can transform the style parameters into the style encodings." A deterministic transformation from a keyword input (style parameter) to a corresponding model-consumable output (style encoding) is, by definition, a mapping. A mapping relationship, in its ordinary and customary sense, is an association between an input in one domain and an output in another, which is exactly what Lupascu Paragraph [0067] describes: "the artistic content generation system 106 utilizes the multi-domain style encoder 308 to determine style encodings 320 from the style parameters associated with the style digital image 316 and the style text prompt 318." "Determining style encodings 320 from the style parameters" is the very definition of a mapping between the style parameters (keywords) and the style encodings (model input information). The claim recites "a mapping relationship library between dimension keywords and model input information." The claim does not require the mapping to be static, hand-coded, or stored in a specific data-structure format such as a hash table. Under BRI, a learned, parameterized mapping from multi-domain inputs to encodings (as in a pretrained CLIP model or a cross-lingual-multimodal-embedding model, per Lupascu [0069]) satisfies the claimed "mapping relationship." Hence, the combination of prior arts fully anticipate all the limitations.. Therefore, applicant remark cannot be considered persuasive.
In regard to Claims 2-9, 12-19, they directly/indirectly depends on independent Claim 1, 11 respectively. Applicant does not argue anything other than the independent claim 1, 11. The limitations in those claims in conjunction with combination previously established as explained.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20250329065 A1 IMAGE GENERATION APPARATUS, IMAGE GENERATION METHOD, AND NON-TRANSITORY COMPUTER-READABLE MEDIUM
US 20240371049 A1 METHOD, COMPUTER DEVICE, AND NON-TRANSITORY COMPUTER-READABLE RECORDING MEDIUM FOR GENERATING IMAGE
US 20120013640 A1 GRAPHICAL REPRESENTATION OF EVENTS
US 11989922 B2 Automated image analysis and indexing
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 YUJANG TSWEI whose telephone number is (571)272-6669. The examiner can normally be reached 8:30am-5:30pm EST.
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 on (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.
/YuJang Tswei/Primary Examiner, Art Unit 2614
Prosecution Insights