Samsung Patents a Two-Pass Lens Shading Fix That Cleans Up Video Stabilization Artifacts

By Patentlyze Team · Updated Jun 5, 2026

When your phone stabilizes a shaky video by cropping and shifting the frame, it quietly breaks the color correction it just applied. Samsung's new patent fixes that by running lens shading correction twice — once before stabilization, once after.

What Samsung's two-stage lens shading correction actually does

Imagine you're recording a video while walking, and your phone's camera is working hard to keep the footage smooth. To do that, it digitally shifts and crops the frame dozens of times per second. The problem? Any color or brightness corrections it applied to the original full frame no longer line up correctly once that frame gets cropped and repositioned.

That mismatch can leave your video with subtle color casts or uneven brightness — especially around the edges of the frame. It's the kind of artifact that's hard to name but easy to notice, and it gets worse when you're moving a lot.

Samsung's patent describes a fix: apply lens shading correction twice. First, you correct the raw full-frame image before stabilization. Then, after the frame has been cropped and shifted by the stabilizer, you apply a second, targeted correction to clean up whatever the stabilization step disturbed. Two passes instead of one.

How the two LSC tables work around VDIS cropping

The patent describes a pipeline with four key stages running inside the camera processor:

• Raw image capture: The camera module captures a full-resolution raw image containing all pixels from the sensor.
• First lens shading correction (LSC): A correction table of per-pixel gain values is applied to compensate for the natural light falloff a lens creates — brighter in the center, darker at the edges. This is standard camera processing.
• VDIS correction: The motion sensor's data is used to perform Video Digital Image Stabilization (VDIS) — essentially cropping and repositioning the frame to counteract detected hand shake. This produces the first corrected image, but the crop means the pixel positions from the original LSC map no longer correspond correctly to the new frame boundaries.
• Second lens shading correction: A second LSC table — with gain values mapped to the stabilized frame's pixel positions — is applied to the VDIS-corrected image, generating the final output.

The key insight is that the second LSC table only needs to cover the pixels that actually appear in the stabilized (cropped) frame, not the full sensor. This makes the second pass computationally lighter while still fixing the artifacts the stabilization step introduced. The motion sensor data is central to the whole process — it informs both where to crop and how to calculate the second correction map.

What this means for Galaxy camera video quality

Lens shading correction is one of those invisible but essential camera pipeline steps that most people never think about — until it goes wrong. In video mode, where VDIS is constantly repositioning the frame, a single-pass correction can leave visible color or exposure inconsistencies, particularly at the edges. This two-pass approach is a practical, low-overhead fix for a real artifact that affects video quality on every modern smartphone.

For Samsung specifically, video performance is a competitive battleground for the Galaxy S and Galaxy Z series. A cleaner stabilization pipeline — one that doesn't trade color accuracy for smooth footage — is the kind of incremental improvement that shows up in camera benchmark scores and real-world reviews. You probably won't notice the mechanism, but you'd notice the absence of the artifacts it prevents.

Editorial commentary on a publicly published patent application. Not legal advice.