A digital material attempts to replicate the look and feel of its physical version as closely as possible. To achieve this it is not sufficient to simply take a high quality photograph of the material as vital info is lost when using such a single-angle approach.
To convert the look of a material to digital extra information is required for the surface structure, reflectivity, and other aspects. Each of these characteristics are captured in separate data channels, which are then combined to create a digital twin of the material.
You could even think of a digital material like a lasagna whereby the different ingredients must be layered in the correct order and quantity to create a delicious dish, or in our case an accurate representation of the real fabric.
Example of the layers that make up a digital material.
In practice, a scanned digital material consists of a set of images, each describing a different characteristic of the material. These are then plugged into their respective data channels, so that when rendering time comes, the software can interpret each channel to correctly visualize the material.
A second important aspect of digital materials is "tile-ability". Most material capturing methods are limited in the size of physical material that can be scanned.
As a result, to be able to cover large digital surfaces the scanned area is copy-pasted next to itself both horizontally and vertically. To create a realistic appearance, the material needs to be edited so that no seams or repetitions are visible (except of course in the case of pattern designs).