# Upscaling Images with AI and Sub-Pixel Information > This patent describes how artificial intelligence, specifically neural networks, can make low-resolution images look sharper by cleverly guessing new pixel colors based on tiny details between existing pixels. - **Patent:** US 12737843 - **Original title:** Upsampling an image using one or more neural networks - **Granted:** 2026 - **Status:** Active - **Times cited:** 0 - **Field:** consumer_electronics, software, ai_ml, telecommunications, gaming ## What it does The patent describes using one or more neural networks to create higher-resolution images from lower-resolution ones. It achieves this by determining 'pixel weights' based on 'sub-pixel offset values.' This means the system looks at the tiny spaces *between* existing pixels in the original image. It then uses these calculated weights to intelligently fill in the new pixels needed for the larger, sharper image. For example, if you have a blurry photo, this system could use AI to add new pixels, making edges look smoother and details clearer by considering where colors should blend at a microscopic level. ## What it does NOT cover - Does not cover image upsampling methods that rely solely on traditional techniques like bilinear or bicubic interpolation without neural networks. - Does not cover neural network-based upsampling that does not determine pixel weights based on sub-pixel offset values, as specified in the abstract. - Does not cover image generation techniques that create entirely new images rather than upscaling existing ones. - Does not cover image enhancement methods that do not involve increasing resolution, such as noise reduction or color correction, unless they also involve upsampling. ## The clever bit The novelty appears to be in how the neural network specifically uses 'sub-pixel offset values' to determine 'pixel weights.' This suggests a fine-grained approach to interpolating new pixel data, going beyond simply looking at the immediate surrounding pixels to reconstruct details more accurately. ## Real-world examples 1. AI image upscalers 2. Video streaming services (e.g., Netflix, YouTube) 3. Digital photography software (e.g., Adobe Photoshop) 4. Medical imaging systems 5. Security camera systems 6. Video game upscaling technologies (e.g., DLSS, FSR) ## Why it matters High-quality images are crucial for everything from medical imaging to entertainment. AI upsampling can make old photos look new, improve video streaming quality, and enhance details in security footage. This technology helps make digital content look better on high-resolution screens and can reduce the storage or bandwidth needed for high-quality visual data. ## Frequently asked questions ### What does Upscaling Images with AI and Sub-Pixel Information cover? This patent describes how artificial intelligence, specifically neural networks, can make low-resolution images look sharper by cleverly guessing new pixel colors based on tiny details between existing pixels. ### When does this patent expire? This patent is expected to expire on September 15, 2046, when the invention enters the public domain. ### What problem does this patent solve? High-quality images are crucial for everything from medical imaging to entertainment. AI upsampling can make old photos look new, improve video streaming quality, and enhance details in security footage. This technology helps make digital content look better on high-resolution screens and can reduce the storage or bandwidth needed for high-quality visual data. ### What does this patent NOT cover? Does not cover image upsampling methods that rely solely on traditional techniques like bilinear or bicubic interpolation without neural networks. **Full plain-English explainer:** https://patentbrief.org/patent/us/12737843/upsampling-an-image-using-one-or-more-neural-networks **Original patent:** https://patents.google.com/patent/US12737843 --- _Source: PatentBrief — https://patentbrief.org. Patent facts are from public records; the plain-English explanation is PatentBrief's._ ## Related patents Semantically similar inventions in the PatentBrief corpus: - [How AI Generates Images Based on Style and Content Cues](https://patentbrief.org/patent/us/12141700/generative-adversarial-network-for-generating-images) — This patent describes an artificial intelligence system that creates new images by understanding both what the image should show (like a cat) and how it should look (like a high-quality photo), using a special internal layer to blend these instructions. - [How AI Transfers Visual Styles Between Images While Keeping Structure](https://patentbrief.org/patent/us/12282696/method-and-system-for-semantic-appearance-transfer-using-splicing-vit-features) — This patent describes an AI method to generate a new image that combines the structure of one input image with the visual style of another, semantically related image, using a pre-trained Vision Transformer model. - [How to Build 3D Models Using Pictures from Different Angles](https://patentbrief.org/patent/us/12737973/3d-model-generation-using-multiple-textures) — This patent describes a system that creates a three-dimensional digital object by combining multiple flat images taken from different viewpoints, then lets a user fine-tune how those images wrap around the object. - [How to Update AI on Small Devices with Slow Internet](https://patentbrief.org/patent/us/20250363357/systems-and-methods-for-deploying-and-updating-neural-networks-at-the-edge-of-a-) — This patent describes a method for efficiently updating artificial intelligence models on small, internet-connected devices, like smart cameras, by sending only the changes, or 'patches,' instead of the entire updated model, which saves bandwidth. - [Training AI Models Across Different Computers](https://patentbrief.org/patent/us/12574477/distributed-deep-learning-using-a-distributed-deep-neural-network) — This 2026 patent describes a way to train AI models on one computer, send a version to another computer for further training with private data, and then update the original model with the improvements.