Processing

General image processing utilities.

Real-time image processing methods, such as box blur, approximate gaussian blur, flood-filling. An implementation of the Convolution pyramids paper also offer a Poisson filling tool and a laplacian integrator.

Classes
class	BilateralBlur
	Apply an approximate bilateral blur to a texture. This can be used to blur while preserving edges, or even to upscale and blur a small texture. The approximate implementation in Rendu is based on the one demonstrated in G3D (McGuire M., Mara M., Majercik Z., http://casual-effects.com/g3d, 2017), and relies on a separable Gaussian blur with adjusted weights and an extra step distance. More...
class	BoxBlur
	Applies a box blur of fixed radius 2. Correspond to uniformly averaging values over a 5x5 square window. More...
class	ConvolutionPyramid
	Implements a multiscale scheme for approximating convolution with large filters. This is the basis of the technique described in Convolution Pyramids, Farbman et al., 2011. A set of filter parameters can be estimated through an offline optimization for each desired task: gradient field integration, seamless image cloning, background filling, or scattered data interpolation. More...
class	FloodFiller
	Perform an approximate flood fill on the GPU, outputing a color filled image or a distance map. Implement the method described in Jump Flooding in GPU with Applications to Voronoi Diagram and Distance Transform, Rong et al., 2006. More...
class	GaussianBlur
	Applies an approximate gaussian blur using a dual filtering approach. More...
class	LaplacianIntegrator
	Compute the laplacian field of a RGB image before reconstructing the initial image through integration, using a filter as described in Convolution Pyramids, Farbman et al., 2011. More...
class	PoissonFiller
	Solve a membrane interpolation ("Poisson filling") problem, using a filter as described in Convolution Pyramids, Farbman et al., 2011. More...
class	SSAO
	Computes screen space ambient occlusion from a depth and view-space normal buffer (brought to [0,1]). More...