A raw digital photo resembles a mosaic: each pixel holds only one base color — red, green, or blue. To compress such a file, it's split into brightness (a detailed black-and-white image) and color components. But the color dots on the sensor are offset from each other, and when color is reconstructed from brightness, jitter and errors creep in.
A new method, CfL-STT, works like a precision stencil alignment: it shifts the color grid to match the brightness map using simple linear interpolation. This eliminates fine noise and makes the layers nearly independent — now they can be compressed with minimal redundancy.
Tests on real photos showed an unexpected result: a simple geometric trick consistently beats the best algorithms, reducing size by 10–20%. And the compression is fully reversible — the original image is restored bit-for-bit. Such precision is critical for medicine, science, and cinematography, where even a microscopic flaw is unacceptable.
The method surprises with its simplicity: it uses basic linear interpolation — essentially drawing straight lines between points — but precise layer alignment achieves results comparable to computationally expensive algorithms.