Patents

US 12770656 B2: Image diffusion software for text-guided video editing

Granted
2026-10-20
Filed
2025-04-16 application 19180333
Type
Utility · 30 claims
Elsewhere
Google Patents ↗ · USPTO patent document ↗ · USPTO Patent Center record ↗

Abstract

The invention provides computer-readable media (CRM) configured to adapt a text-to-image (T2I) diffusion model for video editing by using spectral decomposition to achieve controlled spectral shifts in the model's weights. This CRM maintains constant singular vectors while selectively adjusting singular values according to a text prompt. A spectral shift regularizer constrains these adjustments, particularly limiting changes to larger singular values to ensure minimal deviation from the original model's structure. This approach enables efficient, prompt-driven video editing by modifying specific visual elements according to the prompt while preserving original video context. By employing selective spectral adjustments, the CRM significantly reduces adaptation time and computational demands, making it suitable for real-time, resource-sensitive applications such as dynamic video editing in streaming services.

Source: PTBLXML, file of 2026-10-20 (ipgb20261020.xml).

Assignees

AssigneeTypePlaceAlso on furp.orgSource
University of Central Florida Research Foundation, Inc. 971 patentsUS company or organizationOrlando, FLUNIVERSITY OF CENTRAL FLORIDA RESEARCH FOUNDATION, INC. State business registration EXACT high confidence: identical full name and state
UNIVERSITY OF CENTRAL FLORIDA RESEARCH FOUNDATION INC Nonprofit EXACT high confidence: identical full name and state
PTBLXML, file of 2026-10-20

Inventors

InventorPlacePatentsPerson pageSource
Nazmul KarimOrlando, FL2PTBLXML, file of 2026-10-20
Nazanin RahnavardOrlando, FL2PTBLXML, file of 2026-10-20
Umar KhalidOrlando, FL2PTBLXML, file of 2026-10-20
Chen ChenOrlando, FLPTBLXML, file of 2026-10-20

CPC categories

SubclassCategoryFirst groupSource
G11BInformation storage based on relative movement between record carrier and transducerG11B27/031PTBLXML, file of 2026-10-20
G06FElectric digital data processingG06F40/40PTBLXML, file of 2026-10-20

Cooperative Patent Classification, as currently assigned (one row per subclass).

Assignments and security interests

No recorded assignments or security interests on file for this patent.

USPTO bulk data: PatentsView granted patents, disambiguated (PVGPATDIS) for patents, assignees, inventors, places and CPC categories; the USPTO Patent Assignment Dataset (ECORSEXC) for assignments and security interests. Patents filed in the last 20 years. No claims or full text. Links to other records are by name (and state): EXACT means the identical full name and state; FUZZY means a looser name match, with the confidence shown.