English: A demonstration of global illumination computation using the classical radiosity method. Surfaces are divided into 16x16 meshes (16x32 for the sphere). Top: direct light only (ceiling and shadows are completely black). Bottom: full radiosity solution (at least 20 diffuse bounces contribute to the indirect lighting). The final rendering uses ray casting, and colors the pixel using the radiosity value of the first patch hit by the ray. The images were generated using Python code, using the "numba" library to speed up some code. A single Jacobi iteration was used to generate the top image, and the bottom image used matrix inversion (via the numpy library) to solve the radiosity equation. Sampling was used to estimate form factors rather than analytical formulas. The images were rendered at 2048x2048 and downsampled to 512x512 to anti-alias the edges. The albedo for all surfaces is 0.85 (only really relevant for the bottom image).