Computer-Generated Imagery

Computer-Generated Imagery, usually abbreviated to CGI, refers to images that a computer has generated. None of it was captured with a camera. Instead, graphic artists build a three-dimensional scene within a program: objects, characters, spaces, light sources. From this, the computer calculates a flat image, just as an imaginary camera would see it within that scene. The result can be a single still image or a sequence of images that plays as a film. CGI is thus an umbrella term and not a single piece of software or method.

Why films today can hardly do without CGI

Before CGI, everything visible on screen actually had to exist. A spaceship was a plastic model, a monster was a person in a costume. This limited what could be told, and it was expensive. A collapsing skyscraper can only be filmed once. On a computer, the same collapse can be repeated as often as needed and altered in detail.

That’s why CGI is no longer limited to fantasy and science fiction. Even in quiet dramas, backgrounds are swapped out, crowds are enlarged, or distracting street signs are removed. Many viewers don’t even notice, because the best work remains invisible. CGI is often only considered bad once it becomes noticeable.

An entire industry depends on this economically. Large effects studios in London, Vancouver, or New Zealand each employ thousands of people. Their costs account for a nine-figure sum on some blockbusters. When a studio loses contracts or prices collapse, that is business news, not film news.

From wireframe model to finished image

It begins with modelling. An object is built as a mesh of thousands of small triangles in space. Next comes texturing: surfaces are applied to this mesh, meaning color, rust, skin pores, or fabric patterns. For characters, rigging follows, an internal skeleton made of movable joints. Only with this can the character be animated at all.

The decisive step is called rendering. Here the computer calculates how light behaves in the scene. A common method is ray tracing: it traces individual rays of light and calculates where they are reflected or refracted. This is exactly where believable reflections, soft shadows, and glass that truly looks like glass come from. Rendering is the most expensive part because it consumes an enormous amount of computing time.

A single film frame can take hours on a computer. A feature film has 24 such frames per second. That’s why studios work with render farms, meaning halls full of networked computers calculating simultaneously. Video games don’t have this time budget; they must generate every frame in milliseconds. They therefore use coarser shortcuts and rely heavily on specialized graphics chips.

CGI, deepfakes, and generative AI

In everyday life, CGI is encountered more often than one might think. Many furniture images in online catalogs are not photos but rendered models. This is cheaper than photographing every color variant of a sofa. Car advertising, architectural visualizations, and explainer videos in medicine are also created this way.

In the news, the term currently appears mainly in connection with artificial intelligence. It’s important to distinguish: classic CGI builds a scene component by component, with a human controlling every detail. Image generators, by contrast, produce an image directly from a text description, without any 3D model existing. A deepfake is yet another thing entirely, namely a face swapped via AI into genuine footage.

In practice, these methods are increasingly blending together. Studios use AI to speed up intermediate steps, for example when cutting out objects or aging faces. This drives down costs while simultaneously sparking disputes over jobs. This exact conflict was a central issue in the Hollywood strikes and remains one in many of the industry’s labor negotiations.