# Textures in Ray-tracing

Earth texture bitmap (

large)

## Noise generator (sparse convolution)

Convolution radial function ƒ = (cos(r/π)+1)/2 *(kernel radius is 1)*.

Constant number of samples in one square/cube cell 1x1/1x1x1 is generated.

### Plain noise

Cell: 50x50, samples: 2/cell = 2π/supp(ƒ)

Cell: 50x50, samples: 3/cell

Cell: 50x50, samples: 6/cell

Cell: 50x50, samples: 10/cell

Cell: 20x20, samples: 6/cell

Cell: 8.3x8.3, samples: 6/cell

Cell: 3.3x3.3, samples: 6/cell

### Turbulence

Sum of a sequence of noise values with increasing frequency and decreasing amplitude.

Cell: 50x50, 2 comp, coeff: 0.5

Cell: 50x50, 4 comp, coeff: 0.5

Cell: 42.5x42.5, 6 comp, coeff: 0.7

Cell: 42.5x42.5, 7 comp, coeff: 0.7/0.8

## Noise-based textures

Color 2D texture, spherical mapping

2D bump-map, spherical mapping

Anisotropic bump-map, spherical mapping

Color + bump map, spherical mapping

3D cell texture (hexagonal)

## Water surface simulation

Empirical modelling using continuous noise functions to modulate surface normals.
Caustics were computed by light-tracing preprocessing steps. One rectangular uniform
light source (on vertical wall behind the pool) was sampled by 11.8 million of
light-rays per image. Light map on the pool bottom had 512x512px resolution.

## Flame simulation

Empirical 2D flame simulation is based on deforming a base flame image by a
suitable continuous noise function. Animation is introduced using 3D noise,
drifting slowly along a sloped direction.

Flame animation, snapshots from one animation sequence

Copyright (C) 1996-2017 J.Pelikán,
last change: 2018-03-28 10:19:31 +0200 (Wed, 28 Mar 2018)