Prosecution Insights
Last updated: November 20, 2026
Application No. 19/085,042

TEXT GUIDED IMAGE EDITOR

Non-Final OA §103
Filed
Mar 20, 2025
Priority
Mar 27, 2024 — IN 202411024596
Examiner
CHIN, MICHELLE
Art Unit
2614
Tech Center
2600 — Communications
Assignee
Fujitsu Limited
OA Round
1 (Non-Final)
86%
Grant Probability
Favorable
1-2
OA Rounds
6m
Est. Remaining
97%
With Interview

Examiner Intelligence

Grants 86% — above average
86%
Career Allowance Rate
571 granted / 667 resolved
+23.6% vs TC avg
Moderate +11% lift
Without
With
+11.4%
Interview Lift
resolved cases with interview
Typical timeline
2y 2m
Avg Prosecution
21 currently pending
Career history
701
Total Applications
across all art units

Statute-Specific Performance

§101
9.6%
-30.4% vs TC avg
§103
71.0%
+31.0% vs TC avg
§102
5.7%
-34.3% vs TC avg
§112
1.7%
-38.3% vs TC avg
Black line = Tech Center average estimate • Based on career data from 667 resolved cases

Office Action

§103
DETAILED ACTION Notice of Pre-AIA or AIA Status 1. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . Priority 2. Receipt is acknowledged of papers submitted under 35 U.S.C. 119(a)-(d), which papers have been placed of record in the file. Information Disclosure Statement 3. The information disclosure statement (IDS) submitted on 03/20/2025. 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 4. 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 (i.e., changing from AIA to pre-AIA ) 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. 5. 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. 6. 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. 7. Claim(s) 1-3, 6-15 and 17-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Aberman et al. (US 2024/0037822 A1) in view of Ravi et al. (US 2024/0362842 A1). 8. With reference to claim 1, Aberman teaches A computer-implemented method for image editing, (“FIG. 3 illustrates an example method 300 of generating, using an LLI model, an edited image that is visually similar to a source image, but that includes visual modifications consistent with an edit to an NL prompt used to generate the source image. For convenience, the operations of the flow chart are described with reference to a system that performs the operations. This system can include various components of various computer systems, such as one or more components of server computing device(s).” [0068]) Aberman also teaches the method comprising: obtaining a base prompt indicating a base image and an edit prompt indicating an edit to be made to the base image; (“Since the attention reflects the overall composition, the attention maps M, that were obtained from the generation of a source image I using an original prompt P and an LLI model, can be injected into a second generation using a modified prompt P* and the LLI model, in generating an edited image I*. This allows the synthesis of an edited image I* that is not only manipulated according to the edited prompt, but also preserves the structure of the input image I. Such an example is a specific instance of a broader set of attention-based manipulations enabling different types of intuitive editing.” [0027] “FIG. 1A schematically depicts example components and interactions that can be involved in generating a source image 103A based on processing a natural language prompt 101A using an LLI model 150 and generating, using the LLI model 150, an edited image 107A that is visually similar to the source image 103A, but that includes visual modifications consistent with an edit, to an NL prompt used to generate the source image 103A, that is reflected in prompt edit input 106A.” [0052]) Aberman further teaches repeating, for a plurality of iterations, the steps of: determining new edit embeddings based on: the base and edit embeddings, (“In the preceding editing function, τ is a timestamp/iteration parameter that determines until which step the injection is applied. Note that the composition is determined in the early steps of the diffusion process. Therefore, by limiting the number of injection steps, the composition of the newly generated image can be guided while still allowing the necessary geometry freedom for adapting to the new prompt. An additional or alternative adaptation is to assign a different number of injection timestamps for the different tokens in the prompt. In case the two words are represented using a different number of tokens, the maps can be duplicated/averaged as necessary using an alignment function, such as that described with respect to adding a new phrase.” [0034] “In interacting with the LLI model 150 in generating the edited image 107A, the edited image engine 130 can utilize the random seed(s) 104A that were utilized in generating the source image 103A, can utilize edit features that are based on the edit reflected by the prompt edit input 106A (e.g., a text embedding of a modified prompt reflected by the edit), and can utilize at least some of the cross-attention maps 105A in at least some of the iterations of generating the edited image 107A. Which cross-attention maps 105A are utilized in generating the edited image 107A, and/or which iterations the cross-attention maps are utilized in, can be dependent on the type(s) of edit(s) reflected by the prompt edit input 106A (e.g., dependent on whether the edit is of a replacement, addition, or emphasis adjustment type).” [0056] “the system can monitor for new user interface input that indicates a further edit to the source NL prompt, and that is in addition to edit(s) of prior iteration(s) of block 302. If such new user interface input is detected, the system can proceed to perform another iteration of block 302, 304, 306, 308, and 310 based on such new user interface input.” [0077]) Aberman teaches inputting the base embeddings into a diffusion model in a base reverse process arranged to update a base latent relating to the base image; and inputting the new edit embeddings into the diffusion model in an edit reverse process arranged to update an edit latent relating to an edited image, wherein cross-attention maps generated from the diffusion model in the base reverse process are input into the diffusion model in the edit reverse process; (“Implementations disclosed herein recognize that the structure and appearances of the generated image depend not only on the random seed s, but also on the interaction between the pixels to the text embedding through the diffusion process. More particularly, implementations disclosed herein recognize that through modifying the pixel-to-text interaction that occurs in cross-attention layers, prompt-to-prompt image editing capabilities are enabled that maintain the structure and composition of the source image I that is being edited. More specifically, injecting, in generating the edited image I* using the LLI model, at least some of the cross-attention maps produced in generating the source image I enables preservation of the composition and structure of the source image I. For additional context on cross-attention maps, a particular example of cross-attention in an IMAGEN LLI model (that includes text-conditioned diffusion models) is described in more detail. Implementations of IMAGEN include three text-conditioned diffusion models: a text-to-image 64×64 model, and two super-resolution models—a 64×64.fwdarw.256×256 model and a 256×256.fwdarw.1024×1024 model. These predict the noise ϵ.sub.74(z.sub.t, c, t) via a U-shaped network, for t ranging from T to 1, where z.sub.t is the latent vector and c is the text embedding.” [0020-0021] “With adding a new phrase, user interface input is provided that indicates a user has added new token(s) to the original prompt. For example, “children drawing of” can be prepended to an original prompt of “a castle next to a river”, when the user interface input indicates such prepending. For example, the user interface input can include typing “children drawing of” at the beginning of the original prompt or can be spoken user interface input such as “prepend children drawing of”. With adding a new phrase, to preserve the common details implementations can apply the attention injection only over the common token(s) from both prompts.” [0035] “some implementations disclosed herein adopt an improved inversion approach that is based on a deterministic denoising diffusion implicit model (DDIM) model rather than a denoising diffusion probabilistic model (DDPM). Those implementations can perform the diffusion process in the reverse direction,” [0044] “In order to sample from the distribution q(custom-character.sub.0), the dual “reverse process” p(custom-character.sub.t−1|custom-character.sub.t), from isotropic Gaussian noise custom-character.sub.T to data, is defined by sampling the posteriors q(custom-character.sub.t−1|custom-character.sub.t). … The σ.sub.t of each sample stage can be controlled, and in DDIMs the sampling process can be made deterministic using σ.sub.t=0 in all the steps. The reverse process can finally be trained by solving the following optimization problem:” [0049-0051] “the system can monitor for new user interface input that indicates a further edit to the source NL prompt, and that is in addition to edit(s) of prior iteration(s) of block 302. If such new user interface input is detected, the system can proceed to perform another iteration of block 302, 304, 306, 308, and 310 based on such new user interface input.” [0077]) Aberman also teaches converting the edit latent to the edited image; and outputting the edited image. (“The edited image engine 130 receives prompt edit input 106B, that is user interface input provided at the client device 110 and that specifies one or more edits to the NL prompt 101B (which can be rendered at the client device 110—optionally based on output from the caption engine 140), such as replacement input, addition input, and/or emphasis adjustment input. In response to receiving the prompt edit input 106B, the edited image engine 130 can interact with the LLI model 150 in generating an edited image 107B that is visually similar to the source image 103B but that includes visual modifications that are consistent with the edit(s), to the NL prompt 1016, that are reflected by the prompt edit input 106B.” [0060]) PNG media_image1.png 602 482 media_image1.png Greyscale Aberman does not explicitly teach a time step relating to the iteration, and a weight dependent on the time step wherein the weight controls mixing of the base and edit embeddings; This is what Ravi teaches (“The diffusion prior image editing system 102 selects a conceptual editing step 408 to inject a base image embedding 402 (e.g., the base image embedding 306 discussed above). The diffusion prior image editing system 102 then utilizes the second set of steps 412 to generate the text-edited image embedding 406. Specifically, the diffusion prior image editing system 102 injects the base image embedding 402 at the conceptual editing step 408 and then utilizes the second set of steps 412 to convert the base image embedding 402 into the text-edited image embedding 406. To illustrate, the diffusion prior image editing system 102 processes the base image embedding 402 utilizing the conceptual editing step 408 to generate an intermediate embedding. The diffusion prior image editing system 102 then processes the intermediate embedding by an additional step 414a of the diffusion prior neural network 316. The diffusion prior image editing system 102 iteratively repeats this process through the remaining steps 414b-414n to generate the text-edited image embedding 406.” [0063-0064] “the diffusion prior image editing system 102 performs the conceptual edit process by utilizing the diffusion prior model to modify a CLIP L/14 embedding z.sub.b of the input image x.sub.b to another embedding z.sub.e by conceptually imitating the process of moving the base embedding along the direction specified by the edit text y.sub.e in CLIP embedding space. The diffusion prior image editing system 102 injects x.sub.b into the diffusion prior model custom-character.sub.e at some intermediate timestep t.sub.c during sampling and runs the remaining sampling steps from t.sub.c to 0 while conditioning on the CLIP text embedding z.sub.y of the edit text prompt y.sub.e. Utilizing this approach, the resulting image embedding z.sub.e has context from both the base image and the edit prompt. … The higher the value of t.sub.c, the greater number of steps the prior gets to modify the injected embedding according to edit text and the closer the generated embedding will be to the edit text. In one or more implementations, the diffusion prior image editing system 102 controls the injection timestep the using a conceptual edit strength parameter c∈[0,1] and t.sub.c=T×c. The higher the value of c, the more the base embedding z.sub.b will be modified.” [0071-0072]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 9. With reference to claim 2, Aberman does not explicitly teach the weight is dependent on the time step such that: at earlier time steps, the base and edit embeddings mix less than at later time steps. This is what Ravi teaches (“The diffusion prior image editing system 102 selects a conceptual editing step 408 to inject a base image embedding 402 (e.g., the base image embedding 306 discussed above). The diffusion prior image editing system 102 then utilizes the second set of steps 412 to generate the text-edited image embedding 406. Specifically, the diffusion prior image editing system 102 injects the base image embedding 402 at the conceptual editing step 408 and then utilizes the second set of steps 412 to convert the base image embedding 402 into the text-edited image embedding 406. To illustrate, the diffusion prior image editing system 102 processes the base image embedding 402 utilizing the conceptual editing step 408 to generate an intermediate embedding. The diffusion prior image editing system 102 then processes the intermediate embedding by an additional step 414a of the diffusion prior neural network 316. The diffusion prior image editing system 102 iteratively repeats this process through the remaining steps 414b-414n to generate the text-edited image embedding 406.” [0063-0064] “the diffusion prior image editing system 102 performs the conceptual edit process by utilizing the diffusion prior model to modify a CLIP L/14 embedding z.sub.b of the input image x.sub.b to another embedding z.sub.e by conceptually imitating the process of moving the base embedding along the direction specified by the edit text y.sub.e in CLIP embedding space. The diffusion prior image editing system 102 injects x.sub.b into the diffusion prior model custom-character.sub.e at some intermediate timestep t.sub.c during sampling and runs the remaining sampling steps from t.sub.c to 0 while conditioning on the CLIP text embedding z.sub.y of the edit text prompt y.sub.e. Utilizing this approach, the resulting image embedding z.sub.e has context from both the base image and the edit prompt. … The higher the value of t.sub.c, the greater number of steps the prior gets to modify the injected embedding according to edit text and the closer the generated embedding will be to the edit text. In one or more implementations, the diffusion prior image editing system 102 controls the injection timestep the using a conceptual edit strength parameter c∈[0,1] and t.sub.c=T×c. The higher the value of c, the more the base embedding z.sub.b will be modified.” [0071-0072]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 10. With reference to claim 3, Aberman does not explicitly teach the weight is further dependent on a time invariant parameter which controls how much the base and edit embeddings mix. This is what Ravi teaches (“The diffusion prior image editing system 102 can control the diffusion prior neural network 316 based on a conceptual edit strength parameter 314. The conceptual edit strength parameter 314 includes a metric, measure, or weight indicating a balance or trade-off between input signals. For example, the conceptual edit strength parameter 314 can include a balancing metric or weight relative to the edit text 308 and or the base digital image 302.” [0051] “FIG. 6B illustrates the user interface 602 generated by the diffusion prior image editing system 102 in response to user interaction with the generate image element 614. In particular, FIG. 6B illustrates the user interface 602 that includes a modified digital image 616. As shown, the modified digital image 616 reflects the base digital image 606 and the edit text according to the conceptual edit strength parameter and the structural edit strength parameter indicated by the conceptual weight element 610 and the structural weight element 612. Although not illustrated, the diffusion prior image editing system 102 can iteratively generate modified digital images as the diffusion prior image editing system 102 receives additional user interaction via the user interface 602. For example, in response to selection of a different base digital image, selection of different edit text, selection of a different conceptual edit strength parameter, and/or selection of a different structural edit strength parameter, the diffusion prior image editing system 102 can generate modified digital images and provide the modified digital images for display via the user interface 602.” [0095]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 11. With reference to claim 6, Aberman teaches the inputting of the base embeddings into the diffusion model in the base reverse process and the inputting of the new edit embeddings into the diffusion model in the edit reverse process overlap in time. (“Implementations disclosed herein recognize that the structure and appearances of the generated image depend not only on the random seed s, but also on the interaction between the pixels to the text embedding through the diffusion process. More particularly, implementations disclosed herein recognize that through modifying the pixel-to-text interaction that occurs in cross-attention layers, prompt-to-prompt image editing capabilities are enabled that maintain the structure and composition of the source image I that is being edited. More specifically, injecting, in generating the edited image I* using the LLI model, at least some of the cross-attention maps produced in generating the source image I enables preservation of the composition and structure of the source image I. For additional context on cross-attention maps, a particular example of cross-attention in an IMAGEN LLI model (that includes text-conditioned diffusion models) is described in more detail. Implementations of IMAGEN include three text-conditioned diffusion models: a text-to-image 64×64 model, and two super-resolution models—a 64×64.fwdarw.256×256 model and a 256×256.fwdarw.1024×1024 model. These predict the noise ϵ.sub.74(z.sub.t, c, t) via a U-shaped network, for t ranging from T to 1, where z.sub.t is the latent vector and c is the text embedding. … In generating an image using the IMAGEN LLI model, and/or other LLI model(s), each diffusion step or iteration t includes predicting the noise f from a noisy image z.sub.t and text embedding ψ(custom-character) using a U-shaped network. At the final diffusion step, this process yields the generated image I=z.sub.0. Notably, the interaction between the two modalities occurs during the noise prediction, where the embeddings of the visual and textual features are fused using cross-attention layers that produce spatial attention maps for each textual token.” [0020-0022] “The internal randomness that is used in each of the diffusion processes, which can be reflected by random seed(s), can be fixed/the same in each process. This is due to the nature of diffusion models, where even for the same prompt, two random seeds produce drastically different outputs.” [0029] “a diffusion step can be applied on both z.sub.t−1 and z*.sub.t in the same batch (i.e., in parallel), and so there is only one step overhead with respect to the original inference of the diffusion model.” [0031] “With adding a new phrase, user interface input is provided that indicates a user has added new token(s) to the original prompt. For example, “children drawing of” can be prepended to an original prompt of “a castle next to a river”, when the user interface input indicates such prepending. For example, the user interface input can include typing “children drawing of” at the beginning of the original prompt or can be spoken user interface input such as “prepend children drawing of”. With adding a new phrase, to preserve the common details implementations can apply the attention injection only over the common token(s) from both prompts.” [0035] “some implementations disclosed herein adopt an improved inversion approach that is based on a deterministic denoising diffusion implicit model (DDIM) model rather than a denoising diffusion probabilistic model (DDPM). Those implementations can perform the diffusion process in the reverse direction,” [0044] “In order to sample from the distribution q(custom-character.sub.0), the dual “reverse process” p(custom-character.sub.t−1|custom-character.sub.t), from isotropic Gaussian noise custom-character.sub.T to data, is defined by sampling the posteriors q(custom-character.sub.t−1|custom-character.sub.t). … The σ.sub.t of each sample stage can be controlled, and in DDIMs the sampling process can be made deterministic using σ.sub.t=0 in all the steps. The reverse process can finally be trained by solving the following optimization problem:” [0049-0051] “the system can monitor for new user interface input that indicates a further edit to the source NL prompt, and that is in addition to edit(s) of prior iteration(s) of block 302. If such new user interface input is detected, the system can proceed to perform another iteration of block 302, 304, 306, 308, and 310 based on such new user interface input.” [0077]) 12. With reference to claim 7, Aberman teaches the base prompt is derived from an image. (“The NL prompt 101B can be provided by the client device and based on user interface input (e.g., user interface input that is a user-curated caption for the real image 102B) and/or can be provided by a caption engine 140 that automatically generates the NL prompt by processing the real image 102B using a caption model.” [0058]) 13. With reference to claim 8, Aberman teaches the base prompt and/or the edit prompt are obtained from a user input. (“The NL prompt 101B can be provided by the client device and based on user interface input (e.g., user interface input that is a user-curated caption for the real image 102B) and/or can be provided by a caption engine 140 that automatically generates the NL prompt by processing the real image 102B using a caption model.” [0058]) 14. With reference to claim 9, Aberman teaches the base prompt comprises a textual description relating to the base image. (“assume the source NL prompt is “a furry bear watching a bird”, the source image reflects a furry bear that is watching a red bird, and the source image is generated based on processing, using an LLI model, “a furry bear watching a bird” and a random seed.” [0008] “As a working example of various implementations disclosed herein, let I be a source image that was generated by an LLI model (e.g., a text-guided diffusion model) using the a prompt P and a random seed s.” [0019] “user interface input is provided that indicates a user desire to strengthen or weaken the extent to which token(s) of the original prompt are affecting the original source image. For example, the original prompt can be “a fluffy red ball”, and the user may want an edited image where the ball is more fluffy or less fluffy than it is in the original image.” [0037]) 15. With reference to claim 10, Aberman teaches the edit prompt comprises a textual description relating to the edited image. (“At block 304, the system generates edit features based on the edit to the source NL prompt, that is reflected by the user interface input received at block 302. For example, where the edit is a replacement, the system can generate edit features that include a text embedding of a modified prompt that conforms to the source NL prompt, but replaces the subset of tokens of the source NL prompt with the edited tokens. As another example, where the edit is an addition, the system can generate edit features that include a text embedding of a modified prompt that includes the source NL prompt and the additional tokens. As yet another example, where the edit is an adjustment of emphasis on emphasis token(s), the system can generate edit features that include scaled attention map(s) for the one or more emphasis token(s).” [0071] “the system generates an edited image based on processing, using an LLI model, (A) edit features generated based on edit to the source NL prompt (generated at block 304), (B) the source seed(s) (identified at block 306), and (C) at least some of the cross-attention maps (identified at block 306).” [0073]) 16. With reference to claim 11, Aberman does not explicitly teach the base and edit embeddings comprise vector representations of the base and edit prompts, respectively. This is what Ravi teaches (the base image embedding 306 can include a vector representation of the base digital image 302 upon processing of the base digital image 302 utilizing the trained text-image encoder 304 (e.g., a CLIP L/14 image embedding). As shown in FIG. 3, the diffusion prior image editing system 102 also utilizes the trained text-image encoder 304 to generate an edit text embedding 312 from the edit text 308. As mentioned above, the edit text 308 includes a description of a desired modification, characteristic, or feature of a digital image. The diffusion prior image editing system 102 can utilize the trained text-image encoder 304 to generate the edit text embedding 312 that reflects characteristics or features of the edit text 308. Accordingly, the edit text embedding 312 comprises an encoding, embedding, or representation of the edit text 308. In particular, the edit text embedding 312 can include a vector representation of the edit text 308 upon processing of the edit text 308 utilizing the trained text-image encoder 304 (e.g., a CLIP L/14 text embedding).” [0048-0049]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 17. With reference to claim 12, Aberman teaches the steps of obtaining the base and edit prompts and converting them to embeddings is only performed once per edit. (“In interacting with the LLI model 150 in generating the edited image 107A, the edited image engine 130 can utilize the random seed(s) 104A that were utilized in generating the source image 103A, can utilize edit features that are based on the edit reflected by the prompt edit input 106A (e.g., a text embedding of a modified prompt reflected by the edit), and can utilize at least some of the cross-attention maps 105A in at least some of the iterations of generating the edited image 107A.” [0056] “the system generates edit features based on the edit to the source NL prompt, that is reflected by the user interface input received at block 302. For example, where the edit is a replacement, the system can generate edit features that include a text embedding of a modified prompt that conforms to the source NL prompt, but replaces the subset of tokens of the source NL prompt with the edited tokens. As another example, where the edit is an addition, the system can generate edit features that include a text embedding of a modified prompt that includes the source NL prompt and the additional tokens.” [0071]) 18. With reference to claim 13, Aberman does not explicitly teach the weight is dependent on the time step such that the base embeddings have a greater weight at earlier time steps than at later time steps. This is what Ravi teaches (“The diffusion prior image editing system 102 selects a conceptual editing step 408 to inject a base image embedding 402 (e.g., the base image embedding 306 discussed above). The diffusion prior image editing system 102 then utilizes the second set of steps 412 to generate the text-edited image embedding 406. Specifically, the diffusion prior image editing system 102 injects the base image embedding 402 at the conceptual editing step 408 and then utilizes the second set of steps 412 to convert the base image embedding 402 into the text-edited image embedding 406. To illustrate, the diffusion prior image editing system 102 processes the base image embedding 402 utilizing the conceptual editing step 408 to generate an intermediate embedding. The diffusion prior image editing system 102 then processes the intermediate embedding by an additional step 414a of the diffusion prior neural network 316. The diffusion prior image editing system 102 iteratively repeats this process through the remaining steps 414b-414n to generate the text-edited image embedding 406.” [0063-0064] “the diffusion prior image editing system 102 performs the conceptual edit process by utilizing the diffusion prior model to modify a CLIP L/14 embedding z.sub.b of the input image x.sub.b to another embedding z.sub.e by conceptually imitating the process of moving the base embedding along the direction specified by the edit text y.sub.e in CLIP embedding space. The diffusion prior image editing system 102 injects x.sub.b into the diffusion prior model custom-character.sub.e at some intermediate timestep t.sub.c during sampling and runs the remaining sampling steps from t.sub.c to 0 while conditioning on the CLIP text embedding z.sub.y of the edit text prompt y.sub.e. Utilizing this approach, the resulting image embedding z.sub.e has context from both the base image and the edit prompt. … The higher the value of t.sub.c, the greater number of steps the prior gets to modify the injected embedding according to edit text and the closer the generated embedding will be to the edit text. In one or more implementations, the diffusion prior image editing system 102 controls the injection timestep the using a conceptual edit strength parameter c∈[0,1] and t.sub.c=T×c. The higher the value of c, the more the base embedding z.sub.b will be modified.” [0071-0072]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 19. With reference to claim 14, Aberman teaches the base image comprises an image of a human face. (“As a non-limiting example, assume the source NL prompt is “a furry bear watching a bird”, the source image reflects a furry bear that is watching a red bird, and the source image is generated based on processing, using an LLI model, “a furry bear watching a bird” and a random seed. The edit to the source NL prompt can include a replacement of a subset of tokens of the source NL prompt with replacement token(s) (e.g., replacing “bird” with “butterfly”), an addition of token(s) to the source NL prompt (e.g., adding “blue” before “bird”), and/or an adjustment of emphasis on token(s) of the source NL prompt (e.g., increasing emphasis on “fuzzy”).” [0008]) Aberman teaches the non-limiting example that an image can be a human face. 20. With reference to claim 15, Aberman teaches the edit comprises a change in facial expression of the human face. (“As a non-limiting example, assume the source NL prompt is “a furry bear watching a bird”, the source image reflects a furry bear that is watching a red bird, and the source image is generated based on processing, using an LLI model, “a furry bear watching a bird” and a random seed. The edit to the source NL prompt can include a replacement of a subset of tokens of the source NL prompt with replacement token(s) (e.g., replacing “bird” with “butterfly”), an addition of token(s) to the source NL prompt (e.g., adding “blue” before “bird”), and/or an adjustment of emphasis on token(s) of the source NL prompt (e.g., increasing emphasis on “fuzzy”). Implementations can generate an edited image by processing, using the LLI model, feature(s) generated based on the edit to the source NL prompt, the source random seed and, in at least some of the iterations of processing, at least a portion of the cross-attention maps that were generated in generating the source image. Utilization of the cross-attention maps, in combination with the source random seed, in generating the edited image results in an edited image that is visually similar to the source image, but that includes visual modifications that are consistent with the edit. For instance, if “bird” is replaced with “butterfly”, the edited image can replace the “red bird” of the source image with a “butterfly”, but otherwise be very visually similar.” [0008-0009] “As a working example of various implementations disclosed herein, let I be a source image that was generated by an LLI model (e.g., a text-guided diffusion model) using the a prompt P and a random seed s. Some implementations seek to edit the source image I guided only by an edited prompt P*, resulting in an edited image I*. For example, consider a source image I that is generated from the prompt P “my new bicycle”, and assume that the user wants to edit the color of the bicycle, its material, or even replace it with a scooter while preserving the appearance and structure of the source image I. An intuitive interface for the user is to directly change the text prompt P by further describing the appearance of the bike (e.g., adding “green” before “bicycle”), or replacing it with another word (e.g., replacing “bicycle” with “scooter”). As opposed to some prior techniques, various implementations disclosed herein avoid relying on any user-defined mask (e.g., defined through interaction with the source image I) to assist or signify where the edit to the source image I should occur.” [0019]) 21. With reference to claim 17, Aberman does not explicitly teach the edit latent is converted to the edited image using a latent to image decoder. This is what Ravi teaches (“Upon obtaining the deterministic noised latent z.sub.t.sub.s, in one or more implementations the diffusion prior image editing system 102 runs the sampling process of the diffusion decoder starting from t.sub.s until 0, conditioned on the conceptually edited embedding z.sub.e to get the final edited latent {circumflex over (z)}.sub.e. The generated VAE latent can then be passed through the pre-trained and fixed VAE decoder to get the final edited image x.sub.e as x.sub.e=VAE.sub.dec({circumflex over (z)}.sub.e).” [0087]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 22. With reference to claim 18, Aberman does not explicitly teach the new edit embeddings are determined by an embedding mixer unit which takes the base and edit embeddings, the time step and the weight as inputs and outputs the new edit embeddings. This is what Ravi teaches (“The diffusion prior image editing system 102 can control the diffusion prior neural network 316 based on a conceptual edit strength parameter 314. The conceptual edit strength parameter 314 includes a metric, measure, or weight indicating a balance or trade-off between input signals. For example, the conceptual edit strength parameter 314 can include a balancing metric or weight relative to the edit text 308 and or the base digital image 302.” [0051] “The diffusion prior image editing system 102 selects a conceptual editing step 408 to inject a base image embedding 402 (e.g., the base image embedding 306 discussed above). The diffusion prior image editing system 102 then utilizes the second set of steps 412 to generate the text-edited image embedding 406. Specifically, the diffusion prior image editing system 102 injects the base image embedding 402 at the conceptual editing step 408 and then utilizes the second set of steps 412 to convert the base image embedding 402 into the text-edited image embedding 406. To illustrate, the diffusion prior image editing system 102 processes the base image embedding 402 utilizing the conceptual editing step 408 to generate an intermediate embedding. The diffusion prior image editing system 102 then processes the intermediate embedding by an additional step 414a of the diffusion prior neural network 316. The diffusion prior image editing system 102 iteratively repeats this process through the remaining steps 414b-414n to generate the text-edited image embedding 406.” [0063-0064] “the diffusion prior image editing system 102 performs the conceptual edit process by utilizing the diffusion prior model to modify a CLIP L/14 embedding z.sub.b of the input image x.sub.b to another embedding z.sub.e by conceptually imitating the process of moving the base embedding along the direction specified by the edit text y.sub.e in CLIP embedding space. The diffusion prior image editing system 102 injects x.sub.b into the diffusion prior model custom-character.sub.e at some intermediate timestep t.sub.c during sampling and runs the remaining sampling steps from t.sub.c to 0 while conditioning on the CLIP text embedding z.sub.y of the edit text prompt y.sub.e. Utilizing this approach, the resulting image embedding z.sub.e has context from both the base image and the edit prompt. … The higher the value of t.sub.c, the greater number of steps the prior gets to modify the injected embedding according to edit text and the closer the generated embedding will be to the edit text. In one or more implementations, the diffusion prior image editing system 102 controls the injection timestep the using a conceptual edit strength parameter c∈[0,1] and t.sub.c=T×c. The higher the value of c, the more the base embedding z.sub.b will be modified.” [0071-0072]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Ravi into Aberman, in order to allow the amount of editing to be controlled during iterative image generation. 23. Claim 19 is similar in scope to claim 1, and thus is rejected under similar rationale. Aberman additionally teaches A computer program which, when run on a computer, causes the computer to carry out a method (“Computing device 510 typically includes at least one processor 514 which communicates with a number of peripheral devices via bus subsystem 512. These peripheral devices can include a storage subsystem 524, including, for example, a memory subsystem 525 and a file storage subsystem 526, user interface output devices 520, user interface input devices 522, and a network interface subsystem 516.” [0089] “Other implementations can include a non-transitory computer readable storage medium storing instructions executable by one or more processor(s) (e.g., a central processing unit(s) (CPU(s)), graphics processing unit(s) (GPU(s)), and/or tensor processing unit(s) (TPU(s))) to perform a method such as one or more of the methods described herein. Yet other implementations can include a system of one or more computers that include one or more processors operable to execute stored instructions to perform a method such as one or more of the methods described herein.” [0112]) 24. Claim 20 is similar in scope to claim 1, and thus is rejected under similar rationale. Aberman additionally teaches An information processing apparatus comprising a memory and a processor connected to the memory, wherein the processor is configured to perform a method (“Computing device 510 typically includes at least one processor 514 which communicates with a number of peripheral devices via bus subsystem 512. These peripheral devices can include a storage subsystem 524, including, for example, a memory subsystem 525 and a file storage subsystem 526, user interface output devices 520, user interface input devices 522, and a network interface subsystem 516.” [0089] “Other implementations can include a non-transitory computer readable storage medium storing instructions executable by one or more processor(s) (e.g., a central processing unit(s) (CPU(s)), graphics processing unit(s) (GPU(s)), and/or tensor processing unit(s) (TPU(s))) to perform a method such as one or more of the methods described herein. Yet other implementations can include a system of one or more computers that include one or more processors operable to execute stored instructions to perform a method such as one or more of the methods described herein.” [0112]) 25. Claim(s) 4, 5 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Aberman et al. (US 2024/0037822 A1) and Ravi et al. (US 2024/0362842 A1), as applied to claim 1 above, and further in view of Hinz et al. (US 2024/0320873 A1) 26. With reference to claim 4, Aberman teaches the converting of the base and edit prompts to the base and edit embeddings (“the edit includes a replacement, of a subset of tokens of the source NL prompt, with one or more replacement tokens that differ from the subset of tokens of the source NL prompt. In some versions of those implementations, the one or more features generated based on the edit to the source NL prompt include a text embedding of a modified prompt that conforms to the source NL prompt, but replaces the subset of tokens of the source NL prompt with the edited tokens. In some of those versions, injecting, in the at least some of the iterations of generating the edited image using the LLI model, at least a portion of the source cross-attention maps includes: using the entirety of the source cross-attention maps in processing the text embedding, of the modified prompt, in the at least some of the iterations. … the edit includes an addition, of one or more additional tokens, to the source NL prompt. In some versions of those implementations, the one or more features generated based on the edit to the source NL prompt include a text embedding of a modified prompt that includes the source NL prompt and the additional tokens.” [0099-0100]) The combination of Aberman and Ravi does not explicitly teach converting the base and edit prompts to base and edit tokens and converting the base and edit tokens to base and edit embeddings. This is what Hinz teaches (“the system encodes the text description of the low-resolution image to obtain a text embedding. In some cases, the operations of this step refer to, or may be performed by, a text encoder as described with reference to FIGS. 4-5, 8-10, and 20-21. In some cases, the text description is the text prompt, and the text embedding is the text embedding used by the image generation model. In some cases, the text encoder generates the text embedding via a pretrained encoder (such as the pretrained encoder described with reference to FIGS. 8-10). In some cases, the pretrained encoder is a CLIP model. In some cases, the pretrained encoder pads the text prompt to C words. In some cases, C=77. In some cases, the pretrained encoder tokenizes the text prompt to produce a conditioning vector c∈custom-character.sup.c×1024. In some cases, the pretrained encoder embeds the tempt prompt to obtain word embeddings. In some cases, the text encoder uses features from a penultimate layer of a frozen CLIP feature extractor in order to leverage the pretraining of the pretrained encoder. In some cases, the text encoder uses a learned encoder (such as the learned encoder described with reference to FIGS. 8-10) to process the output of the pretrained encoder to obtain text embedding t=T(CLIP)(c))∈custom-character.sup.c×1024.” [0244-0245]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Hinz into the combination of Aberman and Ravi, in order to improve the compatibility and processing of text prompts in a diffusion based image editing system. 27. With reference to claim 5, Aberman teaches the new edit embeddings are further based on a mask vector and an index vector computed based on the base and edit tokens. (“IMAGEN and/or other LLI model(s) condition on the text prompt in the noise prediction of each diffusion step through two types of attention layers: i) cross-attention layers and ii) hybrid attention that acts both as self-attention and cross-attention by concatenating the text embedding sequence to the key-value pairs of each self-attention layer. Both of them can be referred to as cross-attention since various implementations can intervene only in the cross-attention part of the hybrid attention. That is, only the last channels, which refer to text tokens, are modified in the hybrid attention modules.” [0025] “An additional or alternative adaptation is to assign a different number of injection timestamps for the different tokens in the prompt. In case the two words are represented using a different number of tokens, the maps can be duplicated/averaged as necessary using an alignment function, such as that described with respect to adding a new phrase. With adding a new phrase, user interface input is provided that indicates a user has added new token(s) to the original prompt. For example, “children drawing of” can be prepended to an original prompt of “a castle next to a river”, when the user interface input indicates such prepending. For example, the user interface input can include typing “children drawing of” at the beginning of the original prompt or can be spoken user interface input such as “prepend children drawing of”. With adding a new phrase, to preserve the common details implementations can apply the attention injection only over the common token(s) from both prompts. For example, the attention injection can be applied only over “a castle next to a river” in the preceding example. More formally, an alignment function A can be utilized that receives a token index from edited prompt custom-character* and outputs the corresponding token index in custom-character* or none if there isn't a match.” [0034-0035]) 28. With reference to claim 16, Aberman teaches the converting of the base and edit prompts to the base and edit embeddings (“the edit includes a replacement, of a subset of tokens of the source NL prompt, with one or more replacement tokens that differ from the subset of tokens of the source NL prompt. In some versions of those implementations, the one or more features generated based on the edit to the source NL prompt include a text embedding of a modified prompt that conforms to the source NL prompt, but replaces the subset of tokens of the source NL prompt with the edited tokens. In some of those versions, injecting, in the at least some of the iterations of generating the edited image using the LLI model, at least a portion of the source cross-attention maps includes: using the entirety of the source cross-attention maps in processing the text embedding, of the modified prompt, in the at least some of the iterations. … the edit includes an addition, of one or more additional tokens, to the source NL prompt. In some versions of those implementations, the one or more features generated based on the edit to the source NL prompt include a text embedding of a modified prompt that includes the source NL prompt and the additional tokens.” [0099-0100]) The combination of Aberman and Ravi does not explicitly teach converting the base and edit prompts to base and edit tokens via a tokenizer unit and converting the base and edit tokens to base and edit embeddings via a text encoder unit. This is what Hinz teaches (“the system encodes the text description of the low-resolution image to obtain a text embedding. In some cases, the operations of this step refer to, or may be performed by, a text encoder as described with reference to FIGS. 4-5, 8-10, and 20-21. In some cases, the text description is the text prompt, and the text embedding is the text embedding used by the image generation model. In some cases, the text encoder generates the text embedding via a pretrained encoder (such as the pretrained encoder described with reference to FIGS. 8-10). In some cases, the pretrained encoder is a CLIP model. In some cases, the pretrained encoder pads the text prompt to C words. In some cases, C=77. In some cases, the pretrained encoder tokenizes the text prompt to produce a conditioning vector c∈custom-character.sup.c×1024. In some cases, the pretrained encoder embeds the tempt prompt to obtain word embeddings. In some cases, the text encoder uses features from a penultimate layer of a frozen CLIP feature extractor in order to leverage the pretraining of the pretrained encoder. In some cases, the text encoder uses a learned encoder (such as the learned encoder described with reference to FIGS. 8-10) to process the output of the pretrained encoder to obtain text embedding t=T(CLIP)(c))∈custom-character.sup.c×1024.” [0244-0245]) Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Hinz into the combination of Aberman and Ravi, in order to improve the compatibility and processing of text prompts in a diffusion based image editing system. Conclusion 29. Any inquiry concerning this communication or earlier communications from the examiner should be directed to Michelle Chin whose telephone number is (571)270-3697. The examiner can normally be reached on Monday-Friday 8:00 AM-4:30 PM. Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http:/Awww.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:/Awww.uspto.gov/patents/apply/patent- center for more information about Patent Center and https:/Awww.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. /MICHELLE CHIN/ Primary Examiner, Art Unit 2614
Read full office action

Prosecution Timeline

Mar 20, 2025
Application Filed
Nov 04, 2026
Non-Final Rejection mailed — §103 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12802340
VISIBLE ELEMENT DETERMINATION METHOD AND APPARATUS, STORAGE MEDIUM, AND ELECTRONIC DEVICE
3y 4m to grant Granted Nov 17, 2026
Patent 12802331
PRECISION MODULATED SHADING
2y 7m to grant Granted Nov 17, 2026
Patent 12793293
METHOD FOR POST-PROCESSING DETECTED TUBES AND LINES IN MEDICAL IMAGES
2y 3m to grant Granted Nov 10, 2026
Patent 12784323
TILE-BASED IMMEDIATE MODE RENDERER GRAPHICS PIPELINE WITH PER-TILE DEPTH PRE-PASSES
2y 7m to grant Granted Nov 03, 2026
Patent 12776083
STICKER GENERATION METHOD AND DEVICE
2y 2m to grant Granted Oct 27, 2026
Study what changed to get past this examiner. Based on 5 most recent grants.

Strategy Recommendation AI-generated — please review before filing

Get a prosecution strategy drawn from examiner precedents, rejection analysis, and claim mapping.
Typically takes 5-10 seconds — AI-generated, attorney review required before filing

Prosecution Projections

1-2
Expected OA Rounds
86%
Grant Probability
97%
With Interview (+11.4%)
2y 2m (~6m remaining)
Median Time to Grant
Low
PTA Risk
Based on 667 resolved cases by this examiner. Grant probability derived from career allowance rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month