Raster image

A raster image is an image built from a uniform grid of tiny colored squares. These squares are called pixels, the smallest unit that makes up a digital image. Each individual pixel stores only a single color. Only when millions of these dots come together does our eye perceive a photo. You can think of it like a mosaic: up close you see individual tiles, from a distance a coherent picture. Almost everything you see as an image on a screen is stored this way.

Why raster images become blurry when enlarged

A raster image has a fixed number of dots. A phone photo with twelve megapixels consists of about twelve million of them. This number no longer changes, no matter what you do with the image later. If you enlarge the image significantly, the existing dots are simply stretched larger. No new details are added, because they were never stored in the first place.

This is exactly where a raster image differs from a vector graphic. A vector graphic doesn’t store dots but rather calculation instructions like “circle with radius 5 at this position.” Such images can be scaled to any size without losing sharpness. Company logos and maps are therefore usually created as vector graphics. This doesn’t work for a photo, because real-world scenes can’t be broken down into simple shapes.

A second point is file size. Because every single pixel has to be stored, raster images quickly become very large. That’s why compression methods are used, meaning tricks to store the same information using less storage space. The best-known one is called JPEG, and it discards details that are barely noticeable to the eye in the process.

How dots and numbers become a photo

Each pixel is stored as a small group of numbers. Typically there are three values for red, green, and blue. Each value usually ranges between 0 and 255. The screen mixes the color of the dot from these three numbers. Pure yellow, for example, results from a lot of red, a lot of green, and no blue.

An image is thus basically nothing more than a huge table full of numbers. This table has a width, a height, and three color channels. This is exactly the form in which AI systems are also presented with an image. A program that recognizes traffic signs doesn’t see a sign, but rather computes with rows of numbers. Patterns such as edges or areas of color emerge for the model from differences between neighboring numbers.

Image generators, too, ultimately produce exactly such a table of numbers. They often start with pure random noise and gradually change the values until an image emerges. A common misconception is that such programs paste images together from a database. In reality, every single dot is calculated anew.

Raster images in everyday life and in tech news

The file extensions JPG, PNG, GIF, and WebP all stand for raster images. Screenshots, phone photos, and scanned documents also belong to this category. If a website loads slowly, it’s often due to image files of this kind that are too large. That’s why modern websites deliver versions with different levels of resolution depending on screen size.

In news reports, the term usually comes up indirectly. When there’s talk of resolution, megapixels, or “upscaling,” it’s about raster images. During upscaling, an AI model fills in missing dots and estimates plausible details to add. This often looks convincing, but it’s an invention, not proof. In court or in forensics, this is a well-known problem.

In practical terms, this means one thing above all for you: it’s better to save important images in high resolution right away. Lost details cannot be recovered afterward. And if you need a logo that looks equally good on a poster and a business card, a vector file is the better choice.