OpenGL Shading Language

From OpenGL.org
Revision as of 13:07, 19 March 2011 by Alfonse (Talk | contribs) (External links: Removing useless links.)

Jump to: navigation, search
OpenGL Shading Language
Core in version 4.5
Core since version 2.0
ARB extension GL_ARB_shader_objects, GL_ARB_vertex_shader, GL_ARB_fragment_shader, GL_ARB_shading_language_100

The OpenGL Shading Language (GLSL) is the principle shading languages for OpenGL. While there are several shading languages available for use in OpenGL, GLSL is the only one that is a part of the OpenGL core.

GLSL is a C-style language. The language has undergone a number of version changes, and it shares the deprecation model of OpenGL. The current version of GLSL is 1.50.

Compilation model

GLSL is quite unique among shading languages due to its compilation model. Most shading languages use a simple, one-stage model: you give it a string representing a shader for one of the shader stages, and it gives you a shader object that you can bind to the context and render with.

GLSL uses a more complicated model based on the compilation of C programs. In C, a compilation unit, typically a single source file that may include several external header files, is compiled into an object file. A number of object files are then linked into a single program.

GLSL operates in a similar fashion. A set of source files, given as a series of strings, are compiled into a shader object. This is the analog of an object file. Unlike typical shading languages, there is no requirement that this shader object represent a full shader stage. The only limitation is that it can only compile code that is appropriate for a single, specific shader stage.

Shader objects are useless in that form, just as object files from compiled C programs are useless directly. Shader objects must be linked into a single program object. Unlike most shader languages, the program objects contain all shader stages at once. Only one program can be bound for rendering at any one time, and it is illegal for a program object to omit certain stages entirely.

Legacy Note: If the fixed-function pipeline is available, a program can omit one or more stages that the fixed-function pipeline can cover.

Terminology

Because of GLSL's unique compilation model, GLSL uses unique terminology.

According to GLSL's standard terminology, a shader is just a compiled set of strings for a particular programmable stage; it does not even need to have the complete code for that stage. A program is a fully linked program that covers multiple programmable stages.

For the sake of clarity, we will adjust this slightly. When the term shader is used, it will be synonymous with the GLSL concept of program. To refer to a GLSL shader, the term shader object will be used.

Language

GLSL is a C-style language, so it covers most of the features you would expect with such a language. Control structures (for-loops, if-else statements, etc) exist in GLSL, including the switch statement. This section will not cover the entire language in detail; the GLSL specification can handle that. This section will hit the highlights of the important differences between GLSL and C.

Version

The OpenGL Shading Language has gone though a number of revisions, some of them quite substantial. As part of the OpenGL Specification, a version of OpenGL is required to support one or more specific versions of GLSL. It may optionally support more.

To specify which version of GLSL should be used, use this directive:

 #version 1.50

This would tell the compiler to compile for version 1.50, or error if that version is not available.

You should put this before any other language features.

Extensions

OpenGL Extensions apply to GLSL as well. Unlike regular OpenGL, where extensions are implicitly always there whether you use it or not, GLSL extensions must explicitly be specified in the particular shader string being compiled.

Similar to the #version directive, the user can activate specific extensions with the "#extension" directive. The syntax is as follows:

 #extension extension_name : behavior

The "extension_name" can also be the string "all". This means it works for all extensions. The available behaviors are:

  • enable: Causes the named extension to work; if the implementation does not support the extension, it only gives a warning. Fails if used with "all".
  • require: Causes the named extension to work; if the implementation does not support the extension, it fails. It also fails if used with "all".
  • warn: Causes the named extension to work; however, using the extension will emit warnings. If used with "all", then the use of any extensions will emit warnings.
  • disable: Prevents the named extension from working at all. Thus, any use of it will be seen as undefined syntax and cause an error. If used with "all", then this prevents any extensions from working.

You should put these definitions before any other language features.

Basic types

C has a number of basic types. What C does not have the concept of is basic vector-types: a basic type that intrinsically store more than one value. The OpenGL Shading Language does define vector types.

The basic non-vector types are:

  • bool: conditional type, values may be either true or false
  • int: a signed integer
  • uint: an unsigned integer
  • float: a floating point number

Each of these types, including booleans, can have 2, 3, and 4-component vector equivalents. The n digit below can be 2, 3, or 4:

  • bvecn: a vector of booleans
  • ivecn: a vector of signed integers
  • uvecn: a vector of unsigned integers
  • vecn: a vector of floating-point numbers

Vector values can have the same math operators applied to them that scalar values do. These all perform the component-wise operations on each component. However, in order for these operators to work on vectors, the two vectors must have the same number of components.

You can access vectors using the following syntax:

vec4 someVec;
someVec.x + someVec.y;

This is called swizzling. You can use x, y, z, or w, referring to the first, second, third, and fourth components, respectively.

The reason it has that name swizzling is because the following syntax is entirely valid:

vec2 someVec;
vec4 otherVec = someVec.xyxx;
vec3 thirdVec = otherVec.zyy;

You can use any combination of up to 4 of the letters to create a vector (of the same basic type) of that length. So otherVec.zyy is a vec3, which is how we can initialize a vec3 value with it. Any combination of up to 4 letters is acceptable, so long as the source vector actually has those components. Attempting to access the 'w' component of a vec3 for example is a compile-time error.

Additionally, there are 3 sets of swizzle masks. You can use xyzw, rgba (for colors), or stpq (for texture coordinates). These three sets have no actual difference; they're just syntactic sugar.

Matrices

In addition to vectors, there are also matrix types. All matrix types are floating-point. Matrix types are as follows, where n and m can be the numbers 2, 3, or 4:

  • matnxm: A matrix with n columns and m rows. OpenGL uses column-major matrices, which is standard for mathematics users. Example: mat3x4.
  • matn: A symmetric matrix with n columns and rows. This type is equivalent to matnxn.

Swizzling does not work with matrices. You can instead access a matrix's fields with array syntax:

mat3 theMatrix;
theMatrix[1] = vec3(3.0, 3.0, 3.0); //Sets the second column to all 3.0s
theMatrix[2][0] = 16.0; //Sets the first entry of the third column to 16.0.

Structs and arrays

Constructors

All types have constructor syntax that allows you to create values of that type. Constructors use this general syntax:

 type(value, value, ...);

The type is the type you wish to create, be it scalar, vector, matrix, struct, or array.

Constructors for scalar types are special. They can

Samplers

Texture access is not as simple as reading a value from a memory address. Filtering and other processes are applied to textures, and how texture coordinates are interpreted can be part of the texture access operation. For these reason, texture access is somewhat complicated.

It starts with samplers, a special type that GLSL defines. Each sampler represents a texture that is attached to the program. Samplers have a type that defines what kind of texture can be attached to them. The following samplers types are available:

  • sampler1D
  • sampler2D
  • sampler3D
  • samplerCube
  • sampler2DRect
  • sampler1DShadow: For doing depth shadow texture accesses. Depth formats are required.
  • sampler2DShadow
  • samplerCubeShadow
  • sampler2DRectShadow
  • sampler1DArray: For array textures.
  • sampler2DArray
  • sampler1DArrayShadow
  • sampler2DArrayShadow
  • samplerBuffer: For buffer textures
  • sampler2DMS: For multisample textures.
  • sampler2DMSArray

These samplers are for textures with floating-point image formats. For textures with integral image formats, you can preceed the sampler type with "i" for signed integers or "u" for unsigned integers. However, there are no integral "shadow" samplers.

Uniforms

Stage inputs and outputs

Each stage in the shader pipeline can define a number of input values and a number of output values. There is an implementation-defined maximum number of inputs and outputs for each stage.

In a very few cases, there are also certain built-in outputs or inputs that can be used. These are generated or required by fixed-functionality processes. However, most inputs and outputs are user-defined.

GLSL creates linkage between outputs from one stage and inputs to the next stage very simply. Because the entire program consists of all stages at once, the linkage is made via matching the variable names. It is illegal for the types of the two variables to be different if they have the same name. This will cause a linker error.

Inputs and outputs are defined as follows:

 in vec3 position;
 in vec3 normal;
 out vec3 vertNormal;

The "position" and "normal" are inputs, while "vertNormal" is an output of this stage.

The term "centroid" can be used in front of inputs and outputs as well. The syntax "centroid in" means . The syntax "centroid out" means .

Interpolation

The outputs from a vertex shader (or geometry shader) are interpolated across the surface of a primitive to compute the corresponding input values for the fragment shader. There are some controls that can be applied to the outputs that allow the user to decide how this interpolation takes place.

The default is to use the interpolation mechanism specified by OpenGL. This can be overridden by placing an interpolation qualifier on the output:

  • flat: No interpolation at all. The fragment shader value will receive a constant value over the surface of the triangle, based on the value of the provoking vertex.
  • smooth: Perspective-correct interpolation.
  • noperspective: Pure linear interpolation.

Multisampling

Multisampling causes multiple samples from a single pixel to be processed. Exactly which samples within a pixel are process is implementation defined. This can cause the interpolated value to fall outside of the actual portion of the primitive being rasterized.

This is usually fine. Sometimes, it is important that interpolated values never fall outside of the boundaries of the primitive. This would be the case if certain values would produce undefined results.

The way to handle this is to use "centroid" before the "out" and "in" value. This forces the multisampling algorithm to ensure that the interpolated value is within the boundaries of the primitive.

This does cause performance loss to some degree, so it should only be done where you have actually seen a visual glitch.

Using GLSL shaders

Building shaders

Attributes and draw buffers

For the stages at the start and end of the pipeline (vertex and fragment, respectively), the initial input values and final output values do not come from or go to shader stages. The input values to a vertex shader come from vertex data specified in a vertex array object, pulled from vertex buffer objects. The output values of a fragment shader are piped to particular buffers for the currently bound framebuffer; either the default framebuffer or a framebuffer object.

Because of this, there is a mapping layer for the program's inputs and outputs. The vertex shader's input names are mapped to attribute indices, while the fragment shader's output names are mapped to draw buffer indices. This mapping can be created before the program is linked. If it is not, or if the mapping does not cover all of the inputs and outputs, then the linker will automatically define what indices are mapped to which unmapped input or output names. This auto-generated mapping can be queried by the user after the program is linked.


Setting uniforms

Setting samplers

Error Checking

This short piece of code shows the process of loading a vertex and fragment shader from 2 text files. Then it compiles them and also checks for errors. The idea here is to encourage newcomers to GLSL to always check for errors. It is in C++ but that doesn't matter.

Note that the process of loading and compiling shaders hasn't changed much over the different GL versions.

	char *vertexsource;
	char *fragmentsource;
	int isCompiled_VS, isCompiled_FS;
	int IsLinked;
	int maxLength;
	char *vertexInfoLog;
	char *fragmentInfoLog;
	char *shaderProgramInfoLog;

	/* Read our shaders into the appropriate buffers */
	/* You need to code your own version of ReadTextFile() */
	/* Also, the file extension that you give to your vertex and fragment shader doesn't matter.
	/* Don't forget, OpenGL doesn't care about your files. */
        /* Let's assuming that ReadTextFile allocates a char array and fills it with text */
	vertexsource = ReadTextFile("myfile.vert");
	fragmentsource = ReadTextFile("myfile.frag");

	/* Create an empty vertex shader handle */
	vertexshader = glCreateShader(GL_VERTEX_SHADER);

	/* Send the vertex shader source code to GL */
	/* Note that the source code is NULL character terminated. */
	/* GL will automatically detect that therefore the length info can be 0 in this case (the last parameter) */
	glShaderSource(vertexshader, 1, (const GLchar**)&vertexsource, 0);

	/* Compile the vertex shader */
	glCompileShader(vertexshader);

	glGetShaderiv(vertexshader, GL_COMPILE_STATUS, &IsCompiled_VS);
	if(IsCompiled_VS == FALSE)
	{
		glGetShaderiv(vertexshader, GL_INFO_LOG_LENGTH, &maxLength);

		/* The maxLength includes the NULL character */
		vertexInfoLog = new char[maxLength];

		glGetShaderInfoLog(vertexshader, maxLength, &maxLength, vertexInfoLog);

		/* Handle the error in an appropriate way such as displaying a message or writing to a log file. */
		/* In this simple program, we'll just leave */
		delete [] vertexInfoLog;
		return;
	}

	/* Create an empty fragment shader handle */
	fragmentshader = glCreateShader(GL_FRAGMENT_SHADER);

	/* Send the fragment shader source code to GL */
	/* Note that the source code is NULL character terminated. */
	/* GL will automatically detect that therefore the length info can be 0 in this case (the last parameter) */
	glShaderSource(fragmentshader, 1, (const GLchar**)&fragmentsource, 0);

	/* Compile the fragment shader */
	glCompileShader(fragmentshader);

	glGetShaderiv(fragmentshader, GL_COMPILE_STATUS, &IsCompiled_FS);
	if(IsCompiled_FS == FALSE)
	{
		glGetShaderiv(fragmentshader, GL_INFO_LOG_LENGTH, &maxLength);

		/* The maxLength includes the NULL character */
		fragmentInfoLog = new char[maxLength];

		glGetShaderInfoLog(fragmentshader, maxLength, &maxLength, fragmentInfoLog);

		/* Handle the error in an appropriate way such as displaying a message or writing to a log file. */
		/* In this simple program, we'll just leave */
		delete [] fragmentInfoLog;
		return;
	}

	/* If we reached this point it means the vertex and fragment shaders compiled and are syntax error free. */
	/* We must link them together to make a GL shader program */
	/* GL shader programs are monolithic. It is a single piece made of 1 vertex shader and 1 fragment shader. */
	/* Assign our program handle a "name" */
	shaderprogram = glCreateProgram();

	/* Attach our shaders to our program */
	glAttachShader(shaderprogram, vertexshader);
	glAttachShader(shaderprogram, fragmentshader);

	/* Link our program */
	/* At this stage, the vertex and fragment programs are inspected, optimized and a binary code is generated for the shader. */
	/* The binary code is uploaded to the GPU, if there is no error. */
	glLinkProgram(shaderprogram);

	/* Again, we must check and make sure that it linked. If it fails, it would mean either there is a mismatch between the vertex */
	/* and fragment shaders. It might be that you have surpassed your GPU's abilities. Perhaps too many ALU operations or */
	/* too many texel fetch instructions or too many interpolators or dynamic loops. */

	glGetProgramiv(shaderprogram, GL_LINK_STATUS, (int *)&IsLinked);
	if(IsLinked == FALSE)
	{
		/* Noticed that glGetProgramiv is used to get the length for a shader program, not glGetShaderiv. */
		glGetProgramiv(shaderprogram, GL_INFO_LOG_LENGTH, &maxLength);

		/* The maxLength includes the NULL character */
		shaderProgramInfoLog = new char[maxLength];

		/* Notice that glGetProgramInfoLog, not glGetShaderInfoLog. */
		glGetProgramInfoLog(shaderprogram, maxLength, &maxLength, shaderProgramInfoLog);

		/* Handle the error in an appropriate way such as displaying a message or writing to a log file. */
		/* In this simple program, we'll just leave */
		free(shaderProgramInfoLog);
		return;
	}

	/* Deallocating these */
	delete [] vertexsource;
	delete [] fragmentsource;

	/* In your rendering code, you just need to call glUseProgram, call the various glUniform** to update your uniforms */
	/* and then render. */
	/* Load the shader into the rendering pipeline */
	glUseProgram(shaderprogram);

	/* When the user shuts down your program, you should deallocate all your GL resources. */
	/* Unbind your shader. */
	glUseProgram(0);
	/* Let's detach */
	glDetachShader(shaderprogram, vertexshader);
	glDetachShader(shaderprogram, fragmentshader);
	/* Delete the shaders */
	glDeleteShader(vertexshader);
	glDeleteShader(fragmentshader);
	/* Delete the shader object */
	glDeleteProgram(shaderprogram);

See also

External links