Extreme Color Compression: A Hybrid JPEG-XL and JPEG-2000 Pipeline.

Every photograph we take is a conversation between light and memory. When a camera captures a scene, it records two distinct kinds of information: the brightness of the world, which defines shapes and edges, and the color, which adds the richness of the sky, the skin, and the foliage. For decades, scientists and engineers have known that human eyes are far more sensitive to the first kind of information than the second. We can spot a jagged rock or a distant face in the dark, but we struggle to distinguish subtle shades of blue or green in the same low light. This biological reality has long guided how we store images, yet a new study suggests we have been far too conservative in how much color data we keep.

Researchers have traditionally treated color and brightness as partners that must be compressed together, often using a single set of rules for both. But a new approach, developed by independent researcher Leslie Dalton, proposes a radical separation. By treating the brightness of an image with one modern standard and its color with another, the study demonstrates that we can shrink the file size of color information by thousands of times without the human eye noticing a difference. This method, which the author calls "Extreme Color Compression," challenges the assumption that high-fidelity color is necessary for a high-quality image, showing instead that the structural skeleton of a photo matters far more than the paint on its surface.

To test this idea, the researcher gathered eight hundred high-resolution photographs of everyday scenes, ranging from people and animals to landscapes and flowers. The process began by breaking each image into its three fundamental components: a channel for brightness, and two separate channels for color. The brightness channel was compressed using a powerful, modern tool called JPEG-XL, which is designed to preserve fine details. The two color channels, however, were subjected to a different treatment using an older standard known as JPEG-2000. The goal was to see how much color data could be discarded before the image looked broken. The researcher tested compression levels that were staggering in their intensity, reducing the color information by factors of 250, 1,000, and even 10,000 times.

The results were surprising. Even when the color data was compressed by a factor of 3,000, the images remained visually indistinguishable from the originals to a human observer. When the researcher zoomed in on the pictures, inspecting them under magnification that revealed individual pixels, the structural details remained sharp and clear. The colors, though heavily compressed, still looked natural. It was only when the compression reached extreme levels, such as 4,000 or 10,000 times, that the images began to show signs of trouble, with patches of color simply vanishing or becoming flat. Yet, even at these extreme levels, the underlying structure of the image—the shapes of the objects and the texture of the surfaces—remained intact because the brightness channel had been preserved with high care.

The study also measured the images using objective computer tools designed to mimic human vision. These tools confirmed what the human reviewers saw: the structural quality of the images stayed remarkably high even as the color data was stripped away. At a compression ratio of 1,000, the images retained a level of quality that the computer models rated as excellent. The most striking finding, however, was the impact on storage space. In a typical image, the color data takes up a significant portion of the file. In this new method, the color data became a tiny add-on. For one specific image of a parade, the brightness channel took up a small amount of space, but the color channels, after being compressed 3,000 times, required less than one kilobyte of storage each. In total, the color information made up only about one and a half percent of the final file size, yet the image still looked colorful and complete.

This work suggests that the way we currently archive images is inefficient. For years, the standard practice has been to compress brightness and color together, often sacrificing some detail in both to save space. This new approach argues that we should treat them differently, recognizing that our eyes rely on brightness for structure and color for atmosphere. By using a specialized tool for the brightness and an aggressive, flexible tool for the color, it is possible to store images in a fraction of the space they currently require. The study does not claim that color is unimportant; in many contexts, from safety signs to biological warnings, color is essential. But for the vast majority of everyday photographs, the research indicates that we can afford to be much more generous with how much color data we throw away.

The implications are practical and immediate. For institutions that store millions of images, such as medical archives or digital libraries, this method could drastically reduce the cost of storage without sacrificing the ability to recognize what is in the picture. The researcher provided the computer code used to perform these tests, allowing others to verify the findings. The process involves taking an image, splitting it apart, compressing the parts differently, and then stitching them back together. It is a simple procedure that leverages the unique way human vision works, proving that sometimes, to see the world clearly, we do not need to keep every single shade of color. We only need to keep the light.