Get the newsletter
Microsoft · Camera · Filed Dec 12, 2024 · USPTO published Jun 18, 2026 · Jun 18 release

Microsoft's New Patent Switches Its Camera's AI Off When Your Battery Needs a Break

Your laptop's camera could take better photos when the device has headroom to spare — and fall back to a faster, cheaper method when it doesn't. Microsoft's latest patent describes exactly that kind of conditional switching.

By · Updated · 3 min read

Microsoft Patent: Adaptive AI vs. Hardware Image Enhancement, figure from US 2026/0172697 A1
Figure from the official USPTO publication.
Publication number
US 2026/0172697 A1(US20260172697A1)
Applicant
Microsoft Technology Licensing, LLC
Filed · Published
Dec 12, 2024 · Jun 18, 2026
Inventors
Nicholas David BURTON, Maria Isabel FIGUERA ALEGRE, Esin GULDOGAN, Vivian DSOUZA
Status when we published
Examiner raised objections; the company can respond
Document
20 claims
US classification
348/241
Examiner
DAGNEW, MEKONNEN D (Art Unit 2638)

How Microsoft's camera decides which processor to use

Imagine your laptop's camera always ran its most powerful AI photo-processing engine, no matter what else you were doing. On a plugged-in desktop that sounds fine, but on a thin laptop running on battery during a crowded video call, it could slow everything down or drain your battery faster.

Microsoft's patent describes a system that reads sensor data — things like battery level, temperature, and ambient light — and uses those readings to decide which image-processing method to use. When conditions are good (plenty of power, device running cool), it routes your camera image through an AI chip that produces noticeably sharper results. When conditions are tight, it falls back to a dedicated hardware pipeline that's faster and less demanding.

The result is that you get the best image quality the device can afford at any given moment, rather than always using the expensive path or always using the cheap one.

How sensors trigger the ISP-to-NPU handoff

The patent describes a device with two distinct image-processing paths. The first is an ISP (Image Signal Processing) hardware pipeline — a fixed-function chip that applies traditional corrections like noise reduction and color balance very quickly and at low power cost. The second is an NPU (Neural Processing Unit) running an AI-based enhancement algorithm that can produce higher-quality results but demands more resources.

A sensor subsystem continuously collects two categories of data:

  • Environmental parameters — things like ambient light levels or scene conditions
  • Device operating parameters — things like battery charge, thermal state, or current CPU/GPU load

Those readings are compared against a set of adaptive device operation criteria — essentially a threshold profile. If the sensor values don't meet the criteria (say, the battery is low or the chip is hot), the raw camera image goes through the ISP pipeline and comes out as a competent but modest "first processed image." If the criteria are met, the same raw image goes through the NPU's AI algorithm instead and comes out as a higher-quality "second processed image."

The switching logic is designed to happen automatically, without the user doing anything.

What this means for cameras on thin, fanless Windows PCs

Most camera enhancement decisions today are static — either the AI pipeline is always on or it's always off, often set by a power profile the user picked manually. A device that reads its own sensors and makes that call dynamically is a meaningfully different approach, especially for thin Windows laptops and tablets where thermal and battery constraints change constantly throughout the day.

For you as a user, this could mean sharper video call images when your laptop has headroom, without the thermal throttling or battery drain that would come from running the AI path all the time. It's the kind of infrastructure change that nobody notices when it works — which is usually the point.

Source: US 2026/0172697 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.

Get our take in your Top Stories

Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Microsoft patent news breaks.

Add Patentlyze on Google One click.
There are more where this came from

We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.

More from Microsoft

All 646 Microsoft patents →

More in Camera

All 268 Camera patents →