The surface stores directions, not bumps
A normal map is an image that tells a game how each small patch of a surface should face the light. Its colors encode normal vectors—arrows indicating the direction that the surface is treated as pointing. During rendering, the game samples those directions and uses them to calculate brightness, highlights, and shading, as if the flat polygon had many tiny ridges and grooves.
The map is usually laid over the model like a texture. Artists can paint details into it or bake them from a more detailed version of the object, transferring the appearance of fine sculpted features onto a simpler model. The game still draws the simpler geometry; it just lights that geometry using the more varied directions in the map.
What the trick can—and cannot—change
Because the effect changes the lighting calculation rather than the outline, it can suggest pores, scratches, scales, or carved lines without requiring each detail to be modeled. That saves geometry and can make richly textured surfaces practical to render in real time.
But the illusion depends on light. A normal map cannot make a crack physically open, cast a true shadow from its own raised edge, or alter the object’s silhouette. Viewed from the side, the underlying surface remains smooth. The result is most convincing when the detail is small enough that changing how light reflects is all the eye needs to read it as texture.
