v1::dither Class Reference

#include <hikogui/random/dither.hpp>

Public Member Functions
	dither (dither const &)=default
	dither (dither &&)=default
dither &	operator= (dither const &)=default
dither &	operator= (dither &&)=default
	dither (int num_bits) noexcept
	Create a dither object.
f32x4	next () noexcept
	Get 8 floating point number to add to a samples.
f32x4	next (f32x4 samples) noexcept
	Add dither to the given samples.
float	next (float sample) noexcept
	Add dither to the given samples.

Detailed Description

An object that create dither values to add to samples before rounding.

Dither is created by adding two 8 bit RPDF into a 9 bit TPDF. Then this 9 bit TPDF is converted to floating point, which can be added to the original floating point sample.

We start of with 128 bit from an xorshift128p random number generated that is split into 8 bit chunks, made into TPDF and converted to 8 floating point values.

Constructor & Destructor Documentation

dither()

v1::dither::dither ( int num_bits )

inlinenoexcept

Create a dither object.

Parameters

num_bits

Number of significant fraction bits, excluding the sign bit. For 24 bit signed PCM samples this value is 23.

Member Function Documentation

next() [1/3]

f32x4 v1::dither::next ( )

inlinenoexcept

Get 8 floating point number to add to a samples.

The dither is a TPDF with the maximum being 2 quantization steps.

next() [2/3]

f32x4 v1::dither::next ( f32x4 samples )

inlinenoexcept

Add dither to the given samples.

Parameters

samples

The samples to add dithering to.

Returns

The sample with included dither.

next() [3/3]

float v1::dither::next ( float sample )

inlinenoexcept

Add dither to the given samples.

Parameters

sample

The sample to add dithering to.

Returns

The sample with included dither.

---

The documentation for this class was generated from the following file:

• hikogui/src/hikogui/random/dither.hpp