Get the newsletter
Samsung · Cell phone · Filed Feb 20, 2026 · USPTO published Oct 1, 2026 · Oct 1 release

Samsung Patents a Camera Color Fix That Reads the Time of Day

Your phone's camera often guesses wrong about lighting, turning a warm golden-hour photo into something flat and off-color. Samsung is patenting a way to cut down on those mistakes by feeding the camera the time of day, not just what it sees in the frame.

A detailed diagram of a neural network processing an input raw image and time-capture features to estimate illumination. Drawing from patent filing US 2026/0303986 A1.
A detailed diagram of a neural network processing an input raw image and time-capture features to estimate illumination.All 13 drawings ↓
Publication number
US 2026/0303986 A1(US20260303986A1)
Applicant
SAMSUNG ELECTRONICS CO., LTD.
Filed · Published
Feb 20, 2026 · Oct 1, 2026
Inventors
Mahmoud Afifi, Luxi Zhao, Abhijith Punnappurath, Mohamed Ashraf Hassan Mohamed Abdelsalam, Ran Zhang, Michael Brown
Status when we published
Waiting for an examiner
Document
20 claims
US classification
348/223.1
Parent application
Claims priority from a provisional application 63777494 (filed Mar 25, 2025)

How Samsung's camera uses your clock to fix photo colors

Most phone cameras figure out the color of the light in a scene by analyzing the image itself, which can go badly wrong in unusual lighting like candles, fluorescent offices, or a sunset. The camera ends up making skin tones look green or a cozy indoor scene look cold and clinical.

Samsung's patent proposes adding a simple piece of information the camera already has access to: when the photo was taken. If your phone knows it's 6:45 p.m. in October, it can reasonably expect warm, low-angle sunlight rather than harsh noon sun. That time-of-day clue gets fed into a small AI model alongside the usual pixel color data, and together they drive a better color correction decision.

The result, if it works as described, is that the colors in your photos more closely match what your eyes actually saw in the moment, especially for those lighting situations that trip up cameras most often.

From the filing · CLAIM 1
… generating a time-capture feature based on the metadata; generating a histogram feature based on the raw image; obtaining illuminant chromaticity based on a neural network by using the histogram feature and the time-capture feature as inputs; …

Translation: An AI looks at both the time the photo was taken and pixel data to figure out the lighting.

How the neural network combines time and pixel data

The patent describes a pipeline that runs inside a phone's camera before an image is processed into a final photo.

  • Raw image capture: The camera captures a raw, unprocessed image (the data straight from the sensor, before any color adjustments).
  • Metadata extraction: The system pulls the capture time from the image's metadata, then converts it into a time-capture feature, a numerical representation of when the shot was taken that the neural network can read.
  • Histogram feature: Separately, the raw image is summarized as a color histogram (a statistical count of how many pixels fall into each color range), which describes the overall color distribution of the scene without storing every pixel.
  • Neural network inference: Both inputs go into a neural network (a small AI model trained on many lighting examples), which outputs an estimate of the scene's illuminant chromaticity, the color temperature of the dominant light source.
  • ISP adjustment: That estimate is converted to an RGB color value and used to tune the Image Signal Processor (the hardware chip that turns raw sensor data into a viewable photo), correcting the white balance before the image is saved.

The core idea is that time of day carries real predictive power about what kind of light is probably hitting a scene, and that prior knowledge should make the AI's color estimate more reliable.

From the filing · THE ABSTRACT
… adjusting one or more white balance parameters of an Image Signal Processor (ISP) based on the RGB illuminant color.

Translation: The camera updates its color settings to make whites look natural based on that lighting calculation.

What this means for everyday phone photography

For most phone users, white balance is invisible until it fails, and then it's very visible: an orange cast at a dinner table, a sickly green under office lights, a blown-out blue sky. Cameras already apply white balance corrections automatically, but they can only work from what's in the frame, which leaves them guessing in edge cases.

Adding time as an input is a small architectural change that costs almost nothing at runtime, since the timestamp is already stored in every photo's metadata. Whether Samsung ships this in a future Galaxy camera pipeline isn't certain from the patent alone, but the approach points toward cameras that treat context, not just pixels, as a useful signal for image quality.

Samsung's 142nd filing we've tracked since May in our camera sensor push watchlist follows the voltage trick for less noise and the two-pass light capture chip.

Source: US 2026/0303986 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 Samsung patent news breaks.

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.

The drawings · 13 sheets

13 drawing sheets from US 2026/0303986 A1 · click any drawing to enlarge

User device, network, cloud computing., drawing sheet 1 from US 2026/0303986 A1
Sheet 1 User device, network, cloud computing.
Computer system components., drawing sheet 2 from US 2026/0303986 A1
Sheet 2 Computer system components.
Neural network for image illumination estimation., drawing sheet 3 from US 2026/0303986 A1
Sheet 3 Neural network for image illumination estimation.
Input processing for neural network., drawing sheet 4 from US 2026/0303986 A1
Sheet 4 Input processing for neural network.
Steps for predicting illuminant color., drawing sheet 5 from US 2026/0303986 A1
Sheet 5 Steps for predicting illuminant color.
Input interface for image processing., drawing sheet 6 from US 2026/0303986 A1
Sheet 6 Input interface for image processing.
Steps for predicting illuminant color., drawing sheet 7 from US 2026/0303986 A1
Sheet 7 Steps for predicting illuminant color.
Input interface for image processing., drawing sheet 8 from US 2026/0303986 A1
Sheet 8 Input interface for image processing.
Show the other 5 sheets
Samsung Patents a Camera Color Fix That Reads the Time of Day patent drawing, sheet 9
Sheet 9
Samsung Patents a Camera Color Fix That Reads the Time of Day patent drawing, sheet 10
Sheet 10
Samsung Patents a Camera Color Fix That Reads the Time of Day patent drawing, sheet 11
Sheet 11
Samsung Patents a Camera Color Fix That Reads the Time of Day patent drawing, sheet 12
Sheet 12
Samsung Patents a Camera Color Fix That Reads the Time of Day patent drawing, sheet 13
Sheet 13
Patent filing page

Watchlists this filing belongs to

All watchlists →
Reader comments

No comments yet

Start the discussion

Real name or a handle, either is fine. Comments are read by a person before they appear, so allow a little time. Keep it about the filing.