---

fl::PixelFormat Class Reference

#include <image.h>

Inheritance diagram for fl::PixelFormat:

List of all members.

Public Methods
virtual Image	filter (const Image &image)
	Return an Image in this format.
void	fromAny (const Image &image, Image &result) const
virtual bool	operator== (const PixelFormat &that) const
	Checks if this and that describe the same actual format.
bool	operator!= (const PixelFormat &that) const
virtual unsigned int	getRGBA (void *pixel) const=0
	Return value is always assumed to be non-linear sRGB. Same for other RGB methods below.
virtual void	getRGBA (void *pixel, float values[]) const
	"values" must have at least four elements. Each returned value is in [0,1].
virtual void	getXYZ (void *pixel, float values[]) const
virtual unsigned int	getYUV (void *pixel) const
virtual unsigned char	getGray (void *pixel) const
virtual void	getGray (void *pixel, float &gray) const
virtual unsigned char	getAlpha (void *pixel) const
	Returns fully opaque by default. PixelFormats that actually have an alpha channel must override this to return correct value.
virtual void	setRGBA (void *pixel, unsigned int rgba) const=0
virtual void	setRGBA (void *pixel, float values[]) const
	Each value must be in [0,1]. Values outside this range will be clamped and modified directly in the array.
virtual void	setXYZ (void *pixel, float values[]) const
virtual void	setYUV (void *pixel, unsigned int yuv) const
virtual void	setGray (void *pixel, unsigned char gray) const
virtual void	setGray (void *pixel, float gray) const
virtual void	setAlpha (void *pixel, unsigned char alpha) const
	Ignored by default. Formats that actually have an alpha channel must override this method.
Public Attributes
int	depth
	Number of bytes in one pixel, including any padding.
int	precedence
	Imposes a (partial?) order on formats according to information content. Bigger numbers have more information.
bool	monochrome
	Indicates that this format has no color components.
bool	hasAlpha
	Indicates that this format has a real alpha channel (as apposed to a default alpha value).

---

Detailed Description

A PixelFormat wraps access to an element of an Image. A pixel itself is sort of the combination of a pointer to memory and a PixelFormat. A PixelFormat describes the entire collection of pixels in an image, and we use it to interpret each pixel in the image. PixelFormat extends Filter so it can be used directly to convert image format. The "from*" methods implement an n^2 set of direct conversions between selected formats. These conversions occur frequently when displaying images on an X windows system. For the present, all formats exept for XYZ make sRGB assumptions (see www.srgb.com). In addition, all integer values are non-linear (with gamma = 2.2 as per sRGB spec), and all floating point values are linear. We can add parameters to the formats if we need to distinguish more color spaces. Naming convention: <color space><basic C type for one channel> Color space names refer to sequence of channels (usually bytes) in memory, rather than in machine words (eg: registers in the processor). The leftmost letter in a name refers to the lowest numbered address. The "Bits" formats also follow this convention with two exceptions: 1) The bitmasks are necessarily machine words. 2) The order of channels is actually arbitrary. TODO: Finish thinking about the naming convention and then actually apply it. Fix code to work on both endians (right now it is just little endian). think about adding arbitrary information channels besides alpha (eg: depth). alpha gets special treatment because it has a specific, well-defined effect on how pixels are combined. add alpha blend methods (operator<< and >>) to Pixel

---

Member Function Documentation

Image PixelFormat::filter (  const Image &    image )  [virtual]

Return an Image in this format. Implements fl::Filter. Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatRGBAChar, fl::PixelFormatRGBABits, fl::PixelFormatYVYUChar, and fl::PixelFormatVYUYChar.

void PixelFormat::fromAny (  const Image &    image, Image &    result )  const

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

unsigned char PixelFormat::getAlpha (  void *    pixel )  const [virtual]

Returns fully opaque by default. PixelFormats that actually have an alpha channel must override this to return correct value. Reimplemented in fl::PixelFormatRGBAChar, fl::PixelFormatRGBABits, and fl::PixelFormatRGBAFloat.

void PixelFormat::getGray (  void *    pixel, float &    gray )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

unsigned char PixelFormat::getGray (  void *    pixel )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatYVYUChar, and fl::PixelFormatVYUYChar.

void PixelFormat::getRGBA (  void *    pixel, float    values[] )  const [virtual]

"values" must have at least four elements. Each returned value is in [0,1]. Reimplemented in fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatRGBAFloat, and fl::PixelFormatHLSFloat.

virtual unsigned int fl::PixelFormat::getRGBA (  void *    pixel )  const [pure virtual]

Return value is always assumed to be non-linear sRGB. Same for other RGB methods below. Implemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatRGBAChar, fl::PixelFormatRGBABits, fl::PixelFormatRGBAFloat, fl::PixelFormatYVYUChar, fl::PixelFormatVYUYChar, and fl::PixelFormatHLSFloat.

void PixelFormat::getXYZ (  void *    pixel, float    values[] )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

unsigned int PixelFormat::getYUV (  void *    pixel )  const [virtual]

See PixelFormatYVYUChar::setRGBA() for more details on the conversion matrix. Reimplemented in fl::PixelFormatYVYUChar, and fl::PixelFormatVYUYChar.

bool fl::PixelFormat::operator!= (  const PixelFormat &    that )  const [inline]

bool PixelFormat::operator== (  const PixelFormat &    that )  const [virtual]

Checks if this and that describe the same actual format. Reimplemented in fl::PixelFormatRGBAChar, and fl::PixelFormatRGBABits.

void PixelFormat::setAlpha (  void *    pixel, unsigned char    alpha )  const [virtual]

Ignored by default. Formats that actually have an alpha channel must override this method. Reimplemented in fl::PixelFormatRGBAChar, fl::PixelFormatRGBABits, and fl::PixelFormatRGBAFloat.

void PixelFormat::setGray (  void *    pixel, float    gray )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

void PixelFormat::setGray (  void *    pixel, unsigned char    gray )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

void PixelFormat::setRGBA (  void *    pixel, float    values[] )  const [virtual]

Each value must be in [0,1]. Values outside this range will be clamped and modified directly in the array. Reimplemented in fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatRGBAFloat, and fl::PixelFormatHLSFloat.

virtual void fl::PixelFormat::setRGBA (  void *    pixel, unsigned int    rgba )  const [pure virtual]

Implemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, fl::PixelFormatGrayDouble, fl::PixelFormatRGBAChar, fl::PixelFormatRGBABits, fl::PixelFormatRGBAFloat, fl::PixelFormatYVYUChar, fl::PixelFormatVYUYChar, and fl::PixelFormatHLSFloat.

void PixelFormat::setXYZ (  void *    pixel, float    values[] )  const [virtual]

Reimplemented in fl::PixelFormatGrayChar, fl::PixelFormatGrayFloat, and fl::PixelFormatGrayDouble.

void PixelFormat::setYUV (  void *    pixel, unsigned int    yuv )  const [virtual]

Reimplemented in fl::PixelFormatYVYUChar, and fl::PixelFormatVYUYChar.

---

Member Data Documentation

int fl::PixelFormat::depth

Number of bytes in one pixel, including any padding.

bool fl::PixelFormat::hasAlpha

Indicates that this format has a real alpha channel (as apposed to a default alpha value).

bool fl::PixelFormat::monochrome

Indicates that this format has no color components.

int fl::PixelFormat::precedence

Imposes a (partial?) order on formats according to information content. Bigger numbers have more information.

---

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

• image.h
• Pixel.cc

---