Of course, it is attainable to do that by simply modelling all of the bricks. That’s more likely to be prohibitively costly for any non-trivial scene although.

Normal mapping is a quite common place to begin for this kind of impact. Normal mapping produces the looks of ridges, bumps and different floor aberrations with out the addition of additional triangles within the mesh (that are far costlier, performance-wise). It works by encoding floor normals right into a texture as RGB parts, mapping that texture to a polygonal floor, after which utilizing the sampled floor normals from the feel per-fragment to carry out lighting, as an alternative of the particular polygon floor regular (which does not differ as a lot over the floor like the traditional map’s encoded normals do).

There is a associated approach referred to as parallax mapping, which successfully does the identical factor besides does it by shifting texture coordinates round primarily based on a perform of the view angle relative to the floor regular and peak map worth for the floor as properly. Parallax mapping can sometimes obtain nice depth results than regular mapping alone.

You may modify regular mapping to assist self-shadowing.

Both of those methods — and basically all the methods that may be used to provide the scene you are envisioning with out resorting to easily modelling all of the bricks — are about faking information associated to mild propagation and simulation.

What meaning is that you simply additionally want an affordable lighting mannequin in your recreation (or modelling software) to see actually wonderful outcomes, and a few of these lighting fashions themselves contain important investments in different methods equivalent to producing offline mild probe data for computing diffuse bounces and international illumination.

Unless you spend pretty important time and power modifying your modelling instruments to assist precisely the identical rendering pipeline as the sport engine you are utilizing, and provide the modeller with related information, it is attainable that the end result within the modeller will not look the identical (or pretty much as good) because the lead to the actual recreation. Keep that in thoughts while you’re feeling probably discouraged about the best way your implementations look.