Samsung Patents Technology That Fills Missing Puzzle Pieces for Sharper Video Playback

By Patentlyze Team · Updated Jul 24, 2026

When a video decoder tries to reconstruct a frame by referencing another part of the same image, what happens if that reference hasn't been decoded yet? Samsung's new patent describes a method for filling in those missing pieces on the fly.

How Samsung fills in the blanks during video decoding

Imagine a jigsaw puzzle where you're trying to place a piece by looking at a nearby section of the board, but that nearby section hasn't been filled in yet. That's a problem modern video decoders run into constantly.

When your phone or TV decodes compressed video, it often saves space by saying "this block of pixels looks like that block over there in the same frame, just copy it." But sometimes the "that block over there" hasn't been processed yet, so the decoder is stuck.

Samsung's patent describes a way to handle exactly that situation. Instead of stopping or skipping, the decoder estimates the missing pixel values using whatever has already been reconstructed nearby, then uses that estimate to finish the job. It's a smarter fallback that keeps the decoding process moving without sacrificing image quality.

How the decoder patches unreconstructed reference samples

The patent covers an intra-prediction decoding method, meaning the decoder is referencing pixels within the same image (not from a previous or future frame) to reconstruct a block of the picture.

Here's the flow the patent describes:

• The decoder checks whether the reference block it needs is fully reconstructed.
• If yes, it uses that reference block directly to rebuild the current block.
• If some samples (individual pixel values) in the reference block are not yet decoded, the decoder estimates those missing values using already-reconstructed pixels from elsewhere in the same frame.
• It then combines the real and estimated samples into a complete reference block and proceeds with reconstruction.

The key insight is that intra-frame prediction (referencing pixels within the same image) can sometimes create circular dependencies, where block A needs block B, but block B hasn't been processed yet. Rather than treating this as an error, Samsung's method turns it into a recoverable situation by substituting estimated values.

What this means for video compression efficiency

This kind of fix matters most inside video codec standards, the rules that define how streaming video is compressed and decompressed. Codecs like VVC (Versatile Video Coding) push hard on intra-prediction to squeeze out file size, and handling edge cases like incomplete reference blocks is exactly the kind of detail that separates a working codec from one that produces glitchy frames.

For everyday viewers, a working implementation would mean fewer decoding artifacts, particularly in demanding compression scenarios like 4K and 8K streaming. It's not the kind of feature you'd ever see listed on a product box, but it's the type of foundational codec work that keeps video looking clean when compression is pushed to its limits.

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