Before we can start creating artificial images, we need a solid grasp of what digital images actually are. Just like building anything requires understanding the materials, generating synthetic images requires understanding their fundamental components: pixels and color. Think of this as learning the alphabet before you can write stories. When we generate images, we are essentially deciding what color each tiny dot in the image should be.

Pixels: The Atoms of an Image

At its core, a digital image is simply a grid of tiny squares or dots called pixels, short for "picture elements". Each pixel is the smallest individual unit in an image that can be assigned a color or intensity. Imagine a mosaic made of tiny colored tiles; a digital image is similar, but the tiles are pixels arranged in a rectangular grid.

The resolution of an image refers to the number of pixels it contains, typically expressed as width × height (e.g., 1920 × 1080 pixels). A higher resolution means more pixels, allowing for greater detail. When we generate a synthetic image, we need to decide on its resolution – how many pixels wide and tall it will be.

A simple representation of an image as a 3x3 grid of pixels. Each pixel holds color information.

Representing Intensity: Grayscale

The simplest type of image is a grayscale image. In a grayscale image, each pixel doesn't have a distinct color but rather a single value representing its intensity or brightness. This value typically ranges from 0 (representing black) to a maximum value (often 255, representing white), with all the numbers in between corresponding to different shades of gray.

This range (0-255) comes from using 8 bits to store the intensity value for each pixel ( $2^8 = 256$ possible values). While other ranges exist, 8-bit grayscale is very common. Generating a synthetic grayscale image involves assigning an appropriate intensity value (like 0, 128, 255, or anything in between) to each pixel in the grid.

Adding Color: The RGB Model

Most images we interact with are color images. To represent color, we need more than just a single intensity value per pixel. The most common way computers handle color is using the RGB color model.

In the RGB model, the color of each pixel is determined by combining three primary colors: Red, Green, and Blue. It's an additive model, meaning colors are created by adding different amounts of red, green, and blue light.

Mixing Red and Green light produces Yellow.
Mixing Green and Blue light produces Cyan.
Mixing Blue and Red light produces Magenta.
Mixing equal amounts of all three produces shades of Gray or White.
Mixing none of them results in Black.

Each pixel in an RGB image stores three values: one for the intensity of Red, one for Green, and one for Blue. Similar to grayscale, these values often range from 0 to 255 for each channel (using 8 bits per channel). This is known as 24-bit color ( $3 \text{ channels} \times 8 \text{ bits/channel} = 24 \text{ bits}$ ), allowing for over 16 million ( $256 \times 256 \times 256$ ) possible colors for each pixel!

An example showing the Red, Green, and Blue intensity values (on a 0-255 scale) needed to create a specific shade of orange ( $R=253, G=126, B=20$ ).

When generating synthetic color images, we usually work within the RGB model, defining the R, G, and B values for every pixel to create the desired visual appearance.

Image Data Structure

So, how is this information stored computationally? Digital images are typically represented as multi-dimensional arrays (or tensors) of numbers.

A grayscale image can be represented as a 2D array (matrix) where the dimensions are Height × Width. Each element (h, w) in the array stores the intensity value (e.g., 0-255) of the pixel at row h and column w.
An RGB color image is usually represented as a 3D array with dimensions Height × Width × Channels. Here, the third dimension typically has a size of 3, representing the Red, Green, and Blue channels. So, an element (h, w, c) would store the intensity value for a specific channel c (0 for Red, 1 for Green, 2 for Blue) at the pixel located at row h and column w.

Understanding this array structure is fundamental because when we use programming libraries (like NumPy in Python) to generate or manipulate images, we are directly working with these numerical arrays. Defining the dimensions (resolution) and filling the array elements with appropriate intensity or color values is the essence of programmatic image creation.

Knowing these basics of pixels, resolution, and color models (especially RGB) provides the foundation needed to start thinking about how to generate simple synthetic images, which we'll explore next. We'll be manipulating these pixel values directly or using tools that handle the underlying array structures for us.