---

fl::Pixel Class Reference

#include <image.h>

List of all members.

Public Methods
	Pixel ()
	Pixel (unsigned int rgba)
	Pixel (const Pixel &that)
	Pixel (const PixelFormat &format, void *pixel)
unsigned int	getRGBA () const
void	getRGBA (float values[]) const
void	getXYZ (float values[]) const
void	setRGBA (unsigned int rgba) const
void	setRGBA (float values[]) const
void	setXYZ (float values[]) const
Pixel &	operator= (const Pixel &that)
	Set self to have color contained in that.
Pixel &	operator+= (const Pixel &that)
	Set self to sum of respective color channels.
Pixel	operator+ (const Pixel &that) const
	Add respective color channels.
Pixel	operator * (const Pixel &that) const
	Multiply respective color channels.
Pixel	operator * (float scalar) const
	Scale each channel.
Pixel	operator/ (float scalar) const
	Scale each channel.
Pixel	operator<< (const Pixel &that)
	Alpha blend that into this. The alpha value the governs the blending comes from that.
Public Attributes
const PixelFormat *	format
void *	pixel
	always points to target data: either the union below or some point in an Image buffer
union {
unsigned char   graychar
float   grayfloat
double   graydouble
unsigned int   rgbchar
float   rgbafloat [4]
}	data

---

Detailed Description

Provides convenient access to the functions of PixelFormat for a specific datum. All linear operations between pixels take place in RGB space. It would be better colorwise to do them in XYZ space, but most formats are closer to RGB than XYZ numerically (ie: require less conversion), so it is cheaper to do them in RGB.

There are two possible ways to think of this class: 1) It provides a shorthand for applying the color access functions of PixelFormat to a particular (x,y) position in an image. E.g.: image(x,y).getRGBA () <=> image.format->getRGBA (image(x,y)) assuming the image(x,y) is defined to return a void * to the correct address. 1.1) It provides a way of treating the operation image(x,y) as a reference to the pixel without actually knowing its format. 2) It provides an abstract way to do numerical operations on pixels, such as scaling, adding, and alpha-blending. As such, it should really be an extension of Vector<float>. The current implementation kind of munges these two ideas together, creating something that is rather inefficient for either purpose. Job 1 should be moved into class Image in the form of more color accessor functions. This makes most sense from an efficiency standpoint. It is not clear to me yet what to do about job 1.1 and job 2. Currently, the only code that makes serious use of Pixel uses it for both these jobs. However, it does not do this efficiently. Pixel would do jobs 1.1 and 2 more efficiently if we relieve it of the burden of storing intermediate results. We should instead store the intermediate results in Vector<float> as RGB values.

---

Constructor & Destructor Documentation

Pixel::Pixel (    )

Pixel::Pixel (  unsigned int    rgba )

Pixel::Pixel (  const Pixel &    that )

Pixel::Pixel (  const PixelFormat &    format, void *    pixel )

---

Member Function Documentation

void Pixel::getRGBA (  float    values[] )  const

unsigned int Pixel::getRGBA (    )

void Pixel::getXYZ (  float    values[] )  const

Pixel Pixel::operator * (  float    scalar )  const

Scale each channel.

Pixel Pixel::operator * (  const Pixel &    that )  const

Multiply respective color channels.

Pixel Pixel::operator+ (  const Pixel &    that )  const

Add respective color channels.

Pixel & Pixel::operator+= (  const Pixel &    that )

Set self to sum of respective color channels.

Pixel Pixel::operator/ (  float    scalar )  const

Scale each channel.

Pixel Pixel::operator<< (  const Pixel &    that )

Alpha blend that into this. The alpha value the governs the blending comes from that.

Pixel & Pixel::operator= (  const Pixel &    that )

Set self to have color contained in that.

void Pixel::setRGBA (  float    values[] )  const

void Pixel::setRGBA (  unsigned int    rgba )  const

void Pixel::setXYZ (  float    values[] )  const

---

Member Data Documentation

union { ... } fl::Pixel::data

const PixelFormat* fl::Pixel::format

unsigned char fl::Pixel::graychar

double fl::Pixel::graydouble

float fl::Pixel::grayfloat

void* fl::Pixel::pixel

always points to target data: either the union below or some point in an Image buffer

float fl::Pixel::rgbafloat[4]

unsigned int fl::Pixel::rgbchar

---

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

• image.h
• Pixel.cc

---