Namespace Egami.Quantization

Classes

ErrorDiffusionKernel

A custom error-diffusion kernel for Kernel: how much of each pixel's error goes to which neighbours. Row 0 of Weights is the current row, with the current pixel at column Origin; the following rows are the rows below. Only entries to the right of the current pixel on row 0, and every entry below, may be non-zero, since the rest have been mapped already.

IndexedImage

An image as palette indices: one byte per pixel and the palette they index, for formats and APIs that want indexed color, such as a PDF indexed image. The indices are pooled; dispose the image to return them.

Palette

An ordered list of up to 256 colors to map images onto: a brand palette, a display's inks, or one built from an image with Wu's quantizer. Use it as FixedPalette for the GIF, PNG, BMP and TGA writers, with Remap(Palette, DitherMode, float), or to find the nearest color yourself.

PaletteOptions

How formats with at most 256 colors (GIF, PNG with a palette, and indexed BMP and TGA) build their palette when an image has more colors than fit. Images with few enough colors always keep them exactly, without dithering.

Palettes

Ready-made palettes.

Enums

DitherMode

How the error of mapping pixels to a palette is spread: error diffusion (Floyd–Steinberg and eight other published kernels, and Riemersma's Hilbert-curve walk) or an ordered pattern (Bayer 2, 4, 8 and 16, and 3 × 3). The values are fixed.