logo
Expand description

The library interface is awfully abstract, because it strives to efficiently, and very accurately, support several pixel types. It also allows replacing some parts of the algorithm with different implementations (if you need higher accuracy or higher speed).

Structs

Configuration for the comparison

Abstract wrapper for images. See Dssim::create_image()

L*a*b*b, but using float units (values are 100× smaller than in usual integer representation)

Detailed comparison result

Result of comparison as f64

Traits

Component-wise averaging of pixel values used by Downsample to support arbitrary pixel types

You can customize how images are downsampled

See GammaPixel & ToRGBAPLU

Downsampling should be done in linear RGB color space.

Convert image to L*a*b* planar

RGBA Premultiplied Linear-light Unit scale

Functions

Create new context for a comparison

Type Definitions

RGBA, but: premultiplied alpha, linear (using sRGB primaries, but not its gamma curve), f32 unit scale 0..1

RGB, but: linear (using sRGB primaries, but not its gamma curve), f32 unit scale 0..1