Fastest
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →https://www.patsnap.com/resources/blog/rd-blog/text-to-image-generation-control-patent-landscape/ · Patsnap · data cut-off 2026-07-31 · downloaded from the live page
Patent Landscape · Multimodal ModelsText to Image Generation Control: Who Holds the Patents Behind Prompt-Guided Diffusion
- 9,731% filing growth from 2021 to 2024. the fastest-moving stretch in the dataset, coinciding with the mainstream arrival of diffusion-based image generators.
- The top 5 assignees hold 32.0% of all 3,972 records. concentration is real but well short of a lock — the top 10 add up to only 42.3%, leaving a long tail of active filers.
- G06T image-processing claims cover 59.1% of records. spatial conditioning and artifact suppression sit inside the busiest single IPC subclass in the field.
3,972
Published Records
32%
Top-5 Share of All Records
+9731%
Filing Growth 2021→2024
US
Leading Jurisdiction
Filing growth compares 2021 (16 records) with 2024 (1,573) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 3,972 records in scope (CR5), not by the ranked leaders only.
Published byPatsnap Research··7 min readSourced from Patsnap Eureka
Overview
What this landscape covers
Text-to-image generation control spans the mechanisms that let a diffusion model turn a text prompt, plus optional spatial or identity constraints, into a specific rendered image rather than an arbitrary one. This dataset gathers 3,972 patent families filed against search terms covering classifier-free guidance, denoising steps, spatial conditioning, identity preservation and artifact suppression — the vocabulary that patent drafters use to describe how a generation process is steered rather than merely how images are produced.
Filings were negligible before 2021 and only became a dense filing category from 2022 onward, tracking the shift from research prototypes to commercial diffusion pipelines. Because publication lags filing by roughly 18 months, the most recent one to two years in any chart will always look thinner than they eventually turn out to be.
Filing activity, 2017–2026
Source: Patsnap Eureka. Assignee ranking and totals. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
10,000 free credits to start
The Numbers
Filing trends and technology composition
Two views of the same 3,972-record dataset: how filing volume moved year over year, and which IPC subclasses the claims actually sit in.
From near-zero to a 2024 peak
Filings moved from 0 in 2017 to a peak of 1,573 in 2024, with growth of +9731% between 2021 and 2024. 2025 and 2026 figures are still filling in as publications catch up to filing dates, so they should not be read as a slowdown.
Image processing dominates, but the field is multimodal
G06T (image data processing & generation) covers 59.1% of records and G06N (AI-model computing) covers 36.6%, confirming this is primarily an image-generation claim space with heavy AI-model overlap. Smaller but non-trivial shares in G10L (speech/audio, 3.5%) and G16H (healthcare informatics, 3.0%) show the control mechanisms are already being claimed outside pure image contexts.
Shares are the percentage of the 3,972 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Source: Patsnap Eureka. Filing trend and technology composition. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.
Go deeper on Text to Image Generation Control with Eureka
This page is one run against one query. Ask Eureka your own question about text to image generation control and every answer comes back with the patent numbers behind it.
Try EurekaKey Patents
The most-cited records in this space
Representative Filing
Domain-specific attribute-adapter augmenting pre-trained text-to-image diffusion models
TOYOTA RESEARCH INSTITUTE, INC.
A method for a domain-specific attribute-adapter is described. The method includes learning domain-specific attributes from a collection of domain-specific images. The method also includes encoding a latent space of a pre-trained text-to-image diffusion model according to the learned domain-specific attributes from the collection of domain-specific images. The method further includes decoding the latent space in response to a received text prompt and one or more conditions. The method also includes inferring a series of images based on decoding the latent space in response to the received text prompt and the one or more conditions.Filed by Toyota Research Institute, published 2026-03-05 — illustrates how adapter-based fine-tuning of pre-trained diffusion models is now being claimed as a discrete, licensable component rather than folded into a base model patent.
Most-cited patent records
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20190295302A1 | Segmentation Guided Image Generation With Adversarial Networks | 392 |
| 2 | US20240386015A1 | Composite symbolic and non-symbolic artificial intelligence system for advanced reasoning and semantic search | 370 |
| 3 | US20190251952A1 | Systems and methods for neural voice cloning with a few samples | 156 |
| 4 | US20230067841A1 | Image Enhancement via Iterative Refinement based on Machine Learning Models | 102 |
| 5 | WO2024155584A1 | Systems, methods, devices, and platforms for industrial internet of things | 95 |
| 6 | US20230095092A1 | Denoising diffusion generative adversarial networks | 88 |
| 7 | US20230377226A1 | Generating images using sequences of generative neural networks | 79 |
| 8 | US11995803B1 | Training and deployment of image generation models | 71 |
| 9 | US20230230198A1 | Utilizing a generative neural network to interactively create and modify digital images based on natural lang… | 59 |
| 10 | US20200160153A1 | System and method for augmented reality using conditional cycle-consistent generative image-to-image translat… | 57 |
Citation counts reflect influence within the searched corpus and favour older, earlier-published records over recent ones.
Each row carries its publication number; clicking a row searches Eureka by that number.
Source: Patsnap Eureka. Citation counts and representative records. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Run it yourself
Put your own technology through the same analysis
Where to run it
For builders
MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →Insights
What the data says about where this field is headed
Three read-throughs from the filing trend, the citation table and the concentration figures.
Growth
+9731%
2021 → 2024 filings
The field industrialised in three years
Filings went from 16 in 2021 to 1,573 in 2024. That is not organic growth from a mature category — it is a category being created, largely in step with the public availability of diffusion-based generators. Treat 2025–2026 counts as undercounts, not evidence of a peak already passed.
Filing trend, 2021–2024
Concentration
32.0% / 42.3%
top 5 / top 10 of 3,972 records
Leadership is real but not exclusionary
The top 5 assignees combined hold 32.0% of all 3,972 records in scope, and the top 10 add up to 42.3%. That leaves well over half the corpus filed by entities outside the leading group — a workable landscape for a new entrant with a genuinely different technical angle.
Top 5 vs top 10 share of all records
Citations
392 citations
top-cited record
Foundational segmentation and voice-cloning work still anchors the field
The most-cited record in the set addresses segmentation-guided image generation with adversarial networks, and a neural voice-cloning patent from the same era also ranks highly cited. Both predate the 2021–2024 filing surge, which is the usual pattern: older filings accumulate citations simply by being available longer, not because they remain the most relevant art today.
Most-cited records table
Eureka AI Agent
Looking for what nobody has claimed yet?
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to text to image generation control, with the prior art for and against each one.
Source: Patsnap Eureka. Co-assignee relationships and derived observations. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Players
Who is filing, and where the momentum is cooling
The leader in this dataset holds 511 records against a fifth-place figure of 113 and a tenth-place figure of 65 — a steep drop-off rather than a flat field. Recent-year momentum, however, is negative across the most active filers.
Leader
511 records
vs. 113 at fifth place
One filer sits well ahead of the field
The leading assignee's record count is more than four times the fifth-placed filer's, a gap wide enough to suggest an early, sustained filing programme rather than a single burst of activity.
Assignee ranking, leader vs. fifth place
Momentum
-41% to -94% YoY
latest year vs. prior year
Every major filer slowed in the latest year
Qualcomm, GDM, Google, Nvidia, Snap and Adobe all show double-digit percentage declines in their latest-year filing counts, ranging from -41% to -94%. Given the 18-month publication lag, this looks like an artefact of incomplete recent data rather than a genuine pullback — but it means the current leaderboard understates 2025–2026 activity for all of them equally.
Recent-year YoY by assignee
Collaboration
10 pairs
co-assignee pairings identified
Co-filing is sparse and concentrated
Only ten co-assignee pairs appear in the dataset, and the strongest pairing appears eleven times — evidence that most filing in this space is done solo rather than through joint-venture or cross-licensing structures.
Co-assignee pair analysis
Recent-year filing momentum by assignee
| Assignee | Recent year | YoY |
|---|---|---|
| Qualcomm Inc. | 16 | -41% |
| GDM Holding LLC | 16 | -65% |
| Google LLC | 12 | -87% |
| Nvidia Corp. | 6 | -91% |
| Snap Inc. | 5 | -67% |
| Adobe Inc. | 3 | -94% |
| Samsung Electronics Co., Ltd. | 3 | -92% |
| Lemon Inc. (Cayman Islands) | 1 | -94% |
Source: Patsnap Eureka. Assignee-level momentum. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
What's Next
Where to take this next
The dataset points to specific next questions rather than a single conclusion.
Watch the adapter-fine-tuning layer
Filings that bolt domain-specific control onto a pre-trained diffusion model, rather than claiming a new base model, are a growing filing pattern and sit on comparatively less-crowded claim space.
Explore adapter claims in EurekaTrack the leader's next moves
A filer with more than four times the fifth-placed assignee's record count is worth monitoring directly for continuation filings and licensing signals rather than inferring intent from aggregate counts.
Set up assignee monitoring in EurekaRe-run the trend once 2025 settles
Because publication lags filing by roughly 18 months, any conclusion about a slowdown drawn from 2025–2026 data alone should be revisited once those years are more complete.
Track filing trends in EurekaSource: Patsnap Eureka. Forward-looking reading of the same dataset. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
FAQ
Common questions about this landscape
In this dataset the leading assignee holds 511 records, well ahead of the fifth-placed filer at 113 and the tenth-placed filer at 65. The gap between first and fifth place is steep, suggesting a long-running, deliberate filing programme rather than a single wave of applications. That said, the top 10 assignees combined account for only 42.3% of all 3,972 records in scope, so no single company controls the field outright.
Filings grew from 16 in 2021 to 1,573 in 2024, a +9731% increase over that three-year span, tracking the emergence of commercially available diffusion-based image generators. Growth data for 2025 and 2026 should be read cautiously: patent publication typically lags filing by around 18 months, so the most recent years in any chart are always undercounted at the time of publication. The 2024 figure is the most recent year that can be treated as reasonably complete.
G06T, the IPC subclass covering image data processing and generation, appears in 59.1% of the 3,972 records, making it the single largest technical category. G06N, covering AI-model computing, follows at 36.6%, and G06F (general digital data processing) and G06V (image/video recognition) both sit above 20%. Because a single record can carry multiple IPC classes, these shares add up to more than 100% and should not be summed into a combined total.
The smaller IPC categories in this dataset point to under-claimed territory: healthcare informatics (G16H, 3.0% of records) and speech/audio synthesis (G10L, 3.5%) both show real but comparatively thin filing activity relative to the core image-processing classes. Combined with the fact that co-assignee filing pairs number only ten across the whole dataset, there is room for entrants focused on cross-modal conditioning or domain-specific adapters that has not been heavily claimed by the leading filers.
Not outright. The top 5 assignees combined hold 32.0% of all 3,972 records, which is meaningful concentration but leaves roughly two-thirds of the filing landscape outside that group. A new entrant should still run freedom-to-operate analysis against the specific claims of the leading filers' most-cited records, since citation count is a signal of influence within the corpus rather than a guarantee that a given claim is still being actively enforced or extended.
Answers are grounded in the same dataset. Derived from a Patsnap search on Text to Image Generation Control covering 2015–2026, data cut-off 2026-07-31. Counts reflect published records only and shift as new filings publish.Run this in Eureka MCP
Research Text to Image Generation Control in depth with Eureka
Go past this page: query the whole text to image generation control corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company's registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.