Render
Rendering means: A computer calculates a finished image from a description — such as a 3D scene or a piece of code. The term comes up in video games, animated films, websites, and increasingly in AI image generators.
A computer doesn’t always store an image as a finished picture. Often only a description exists: there’s a cube in this spot, it’s red, the light comes from the left. From this description, the machine first has to calculate what color each individual pixel ends up having. This calculation process is exactly what is called rendering. So at the start there’s an instruction, and at the end a visible image. The word comes from English and roughly means “to reproduce” or “to depict”.
Why films compute for months and games don’t
Rendering costs computing time, and surprisingly a lot of it. A single frame from an animated film can take a computer several hours. A cinema film has 24 frames per second, so over 170,000 frames for 120 minutes. That’s why studios operate so-called render farms: halls full of computers that divide the frames among themselves.
A video game has this problem in an intensified form. There, a frame is not allowed to take hours, but at most about 16 milliseconds. Only then do 60 frames per second run smoothly. This time pressure explains why game graphics look different from film graphics. Games use plenty of shortcuts and approximations, while films calculate the physics of light more precisely.
Economically, this is a major expense. Graphics cards are expensive because they’re built specifically for this kind of calculation. The fact that the same cards now also power AI models is no coincidence — both tasks consist of very many similar computational steps that can run in parallel.
From scene description to pixel
A 3D scene consists of points in space connected to form triangles. Thousands of such triangles make up a character or a car. Added to this are specifications about material and surface: matte, glossy, transparent. Light sources and a virtual camera are also defined. The renderer takes all of this and determines the color of each pixel from it.
There are two basic approaches to light calculation. The fast variant estimates brightness and shadows using simplified formulas. The precise variant is called ray tracing: for each pixel, the path of a light ray is traced backwards, from the camera to the light source. This produces realistic reflections and soft shadows — but computing time rises sharply.
It’s important to distinguish this from AI image generation. An image generator does not render in the classic sense. It has no 3D scene with triangles and lights, but instead generates pixels directly from learned patterns. Nevertheless, people colloquially often say a model has “rendered” an image. That’s understandable, but technically imprecise.
From browser tab to render farm
The term comes up well beyond the film industry. Every website gets rendered: the browser receives text code and builds the visible page from it, with fonts, images, and buttons. When a page briefly stays blank, it’s often this process getting stuck. Developers distinguish here whether rendering happens on the server or only on the user’s device.
In news from the tech industry, the word appears mainly in connection with graphics cards. Reports about new chips almost always mention figures on rendering performance and ray tracing. Cloud providers also rent out computing time for rendering by the hour, so small studios don’t need their own hall.
One final everyday case: when someone edits a video, they click Export at the end. There too, the program recalculates every single frame anew, including effects and transitions. The progress bar then shows exactly this rendering process. A three-minute video can thus take several minutes even on a normal laptop.