Vertex Specification

Revision as of 15:13, 1 September 2012 by Alfonse (Talk | contribs) (link update.)

Jump to: navigation, search

Vertex Specification is the process of setting up the necessary objects for rendering with a particular shader program, as well as the process of using those objects to render.


Submitting vertex data for rendering requires creating a stream of vertices, and then telling OpenGL how to interpret that stream.

Vertex Stream

In order to render at all, you must be using a shader program. This program has a list of expected Vertex Attributes. This set of attributes determines what attribute values you must send in order to properly render with this shader.

For each attribute in the shader, you must provide a list of data for that attribute. All of these lists must have the same number of elements.

The order of vertices in the stream is very important; it determines how OpenGL will render your mesh. The order of the stream can either be the order of data in the arrays, or you can specify a list of indices. The indices control what order the vertices are received in, and indices can specify the same vertex more than once.

Let's say you have the following as your array of 3d position data:

 { {1, 1, 1}, {0, 0, 0}, {0, 0, 1} }

If you simply use this as a stream as is, OpenGL will receive and process these three vertices in order (left-to-right). However, you can also specify a list of indices that will select which vertices to use and in which order.

Let's say we have the following index list:

 {2, 1, 0, 2, 1, 2}

If we render with the above attribute array, but selected by the index list, OpenGL will receive the following stream of vertex attribute data:

 { {0, 0, 1}, {0, 0, 0}, {1, 1, 1}, {0, 0, 1}, {0, 0, 0}, {0, 0, 1} }

The index list is a way of reordering the vertex attribute array data without having to actually change it. This is mostly useful as a means of data compression; in most tight meshes, vertices are used multiple times. Being able to store the vertex attributes for that vertex only once is very economical, as a vertex's attribute data is generally around 32 bytes, while indices are usually 2-4 bytes in size.

A vertex stream can of course have multiple attributes. You can take the above position array and augment it with, for example, a texture coordinate array:

 { {0, 0}, {0.5, 0}, {0, 1} }

The vertex stream you get will be as follows:

 { [{0, 0, 1}, {0, 1}], [{0, 0, 0}, {0.5, 0}], [{1, 1, 1}, {0, 0}], [{0, 0, 1}, {0, 1}], [{0, 0, 0}, {0.5, 0}], [{0, 0, 1}, {0, 1}] }
Note: Oftentimes, authoring tools will have similar attribute arrays, but the sizes will be different. These tools give each attribute array a separate index list; this makes each attribute list smaller. OpenGL (and Direct3D, if you're wondering) does not allow this. Each attribute array must be the same size, and each index corresponds to the same location in each attribute array.. You must manually convert the format exported by your authoring tool into the format described above.


The above stream is not enough to actually get anything; you must tell OpenGL how to interpret this stream in order to get proper rendering. And this means telling OpenGL what kind of primitive to interpret the stream as.

There are many ways for OpenGL to interpret a stream of 12 vertices. It can interpret the vertices as a sequence of triangles, points, or lines. It can even interpret these differently; it can interpret 12 vertices as 4 independent triangles (take every 3 verts as a triangle), as 10 dependent triangles (every group of 3 sequential vertices in the stream is a triangle), and so on.

The main article on this subject has the details.

Building the Stream

Making a vertex stream in OpenGL requires using two kinds of objects: Vertex Array Objects (VAO) and Vertex Buffer Objects (VBO). VBOs store the actual vertex and index arrays, while VAOs store the settings for interpreting the data in those arrays.

The first step is to create a VAO and bind it to the context.

Vertex Format

Each attribute in the VAO has its own binding point with its own parameters. In the Vertex Array Objects article, we used this pseudocode to explain the state that goes into an attribute binding:

struct VertexAttribute
    bool             bIsEnabled          = GL_FALSE;
    int              iSize               = 4; //This is the number of elements in this attribute, 1-4.
    unsigned int     iStride             = 0;
    VertexAttribType eType               = GL_FLOAT;
    bool             bIsNormalized       = GL_FALSE;
    bool             bIsIntegral         = GL_FALSE;
    void *           pBufferObjectOffset = 0;
    BufferObject *   pBufferObj          = 0;

Recall that this binding data is set by one of the glVertexAttribPointer family of functions.

Similarly to how pixel transfer operations have an internal format (the format of the image data in the texture or framebuffer) and an external format (the format of the image data in client memory or a buffer object), each attribute has an actual format and the format of the data you are passing.

The actual format is defined by the shading language. So the internal attribute format changes depending on which shader you use the VAO with. Every vertex shader has an expected list of attributes. Each attribute has a particular dimensionality and expected type. The type can be float or integral.

For example, if you define an attribute in GLSL as an ivec3​, this means that the dimensionality of the attribute is 3 and the type is integral.

OpenGL is quite flexible in conversions between the data in your buffer objects and what the attribute expects. OpenGL can convert any data from your attribute format to the destination attribute format as long as it has the correct type. If the shader attribute is of integral type, you must use glVertexAttribIPointer to attach the attribute data. The same goes for floating-point attributes and glVertexAttribPointer. If you use a double-precision float attribute (dvec3​), then you must use glVertexAttribLPointer.

Other than this, OpenGL will make conversions as necessary. Normalized integers are converted on the expected range. Non-normalized integers given to floating-point attributes are converted to floating point values.

If there is a mismatch between the incoming dimensionality and the attribute's dimensionality, OpenGL will satisfy it. If the vertex data has more components than the shader attribute uses, the extra components are ignored. If the vertex data specifies fewer than the shader attribute uses, then the unfilled-in components are 0, except for the 4th component which is set to 1.

Your options for specifying your vertex format are varied.

The iSize​ value is the dimensionality of the attribute you are sending. It is an integer from 1-4.

Note: The size value you send with glVertexAttribPointer can also be GL_BGRA, which is equivalent to 4, but specifies that the order of the components in the buffer object is switched from the expected RGBA. This value is only allowed in glVertexAttribPointer. It requires that you use normalization and only use unsigned byte types, and it will force both of these. This feature is not very flexible, because it was primarily introduced for Direct3D compatibility reasons; you probably are better off not using it unless you're trying to be data compatible with D3D.

The bIsIntegral​ value specifies whether the value is an integer or float. If it is false, then it is a floating-point value; if it is true, then it is integral. As previously mentioned, this must match with the attribute definition in the shader. This value is set based on which of the attribute functions you use: "IPointer" sets it to true, the regular "Pointer" version sets it to false.

Compatibility Note: In a compatibility profile, you are required to use attribute index 0 for some attribute. If you do not, then rendering will not take place. Both the shader and the vertex array state must have an attribute index 0.

Data Conversion

The eType​ is the type of the data in the buffer object, not the attribute's type. This value and bIsNormalized​ define how the value in the buffer is interpreted and converted into the attribute's type in the shader.

The possible values of eType​ are broken into categories: integer and floating-point.

  • Integer types allow signed or unsigned forms of bytes, shorts (16-bit), and int (32-bit).
  • Floating-point types are float (32-bit), double (64-bit), or half (16-bit half float values).

Floating-point buffer types convert to floating-point attribute types directly. Integer types can be converted to floating point types in one of two ways: either directly (convert the number into a float value) or via normalization.

Normalization, activated by setting bIsNormalized​ to true, causes unsigned integers to be converted into floating point values on the range [0, 1], and signed integers are converted to [-1, 1].

  • An integer value of 0 gives a normalized float value of 0.0.
  • To get a float value of 1.0, you pass in the largest possible integer for that size (unsigned bytes == 255, signed bytes == 127, etc).
  • To get -1, you pass the most negative integer value (signed bytes == -128, signed shorts == -32768, etc).

Stride and Interleaving

The placement of vertex attribute data within the buffer is governed by the eType​, iSize​, iStride​, and pBufferObjectOffset​ fields. The type and size determine how big an individual value for the attribute is; 2 unsigned bytes takes up, well two bytes. The iStride​ specifies how many bytes it takes to get from one element to the next.

You might think that this would simply be the size of the attribute data: the type * iSize. However, OpenGL is very flexible in how you can position data in the buffer. The stride field allows you to interleave vertex data within a buffer. That is, you can put the data for multiple attributes in the same buffer.

Let's say that your vertex data is built by a hard-coded C structure:

struct Vertex
    float x, y, z;
    unsigned char r, g, b, a;
    float u, v;

You would like your vertex data in your vertex buffer to simply be an array of such structures. So when you bind your buffer and fill it with data, you give it a Vertex*​ of some particular length.

The position attribute would have a pBufferObjectOffset​ value of 0, since it is the first entry. It's size is 3 and type is GL_FLOAT. This means that the position attribute of the first vertex in the array comes from the 0'th byte in the buffer, and OpenGL will pull 12 bytes (sizeof(float) * 3) to get that data. However, you need OpenGL to jump forward 24 bytes, the size of Vertex​ to get to the next position. You do this by setting the iStride​ to 24.

The color attribute would have a pBufferObjectOffset​ value of 12, since they start 3 floats into the struct. The size is 4 and the type is GL_UNSIGNED_BYTE (and it is normalized). The stride is also 24. Remember that stride is the number of bytes from one vertex value to the next. Each attribute has the same stride: 24.

You can compute the stride of any structure that you want to use in a buffer object with the C sizeof​ feature. If you want to compute the offset of a component within a struct, for the pBufferObjectOffset​, you may use the macro offsetof​ (included in stddef.h​ header file). This would be done as follows: offsetof(Vertex, r)​.

A stride value of 0 has a special meaning: it means the attribute data is tightly packed. So a stride of zero always means that the stride is sizeof(eType) * iSize​.

It is very possible, and usually desired for performance reasons, to use a single buffer object for all attributes of a mesh's data. Indeed, thanks to the buffer object offset, it is possible to put the data for many meshes in the same buffer object. You simply use glBufferSubData or glMapBufferRange to update different sections of it, and then set the offset when you are building your VAO.

Matrix Attributes

Attributes in GLSL can be of matrix types. However, our attribute binding functions only bind up to a dimensionality of 4. OpenGL solves this problem by converting matrix GLSL attributes into multiple attribute indices.

If you directly assign an attribute index to a matrix type, it implicitly takes up more than one attribute index. How many it takes depends on the height of the matrix: a 2x2 matrix will take 2, while a 2x4 matrix will take 4. The size of each attribute is the width of the matrix.

Each bound attribute in the VAO therefore fills in a single row, starting with the top-most and progressing down. Thus, if you have a 3x3 matrix, and you assign it to attribute index 3, it will naturally take attribute indices 3, 4, and 5. Each of these indices will be 3 elements in size. Attribute 3 is the top row, 4 is the middle, and 5 is the bottom.

If you let GLSL apply attributes manually, and query them with glGetAttribLocation, then OpenGL will allocate locations for matrix attributes contiguously as above. So if you defined a 3x3 matrix, it will return one value, but the next two values are also valid, active attributes.

Index Data

If you intend to use indices for your vertex arrays, you will need an index list. This is simply a buffer object that contains a list of integer values. It is best that this not be the same buffer object you use for attributes.

To attach this to the VAO, simply call glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, bufferObj);​ with your buffer object. Once attached, any indexed drawing calls will pull from this buffer. Attempting to use indexed drawing calls without a buffer bound to GL_ELEMENT_ARRAY_BUFFER is an error.

Finished VAO

At this point, your VAO is finished. You can render with it immediately, or you can bind another VAO and build a new one (or render with it). It's probably a good idea not to change a VAO once you've created it.

Fixed Attribute Values

It is legal to render with a vertex shader and VAO pair that do not match completely in terms of attributes. Any attributes the VAO specifies that the vertex shader does not consume are ignored. Whereas any attributes that the vertex shader needs but the VAO does not provide will be taken from a set of fixed attribute data.

This fixed attribute data is not an array. Therefore, each vertex in the stream will get the same value. The initial value for these is (0.0, 0.0, 0.0, 1.0).

To change the value, you use a function of this form:

 void glVertexAttrib*(GLuint index​, Type values​);

The * is the type descriptor, using the traditional OpenGL syntax. There is also a version for setting integral attributes, glVertexAttribI*. The index​ is the attribute index to change.

Note that the fixed attribute values are not part of the VAO state. Changes to them do not affect the VAO.

Note: It is not recommended that you use these. The performance characteristics of using fixed attribute data are unknown, and it is not a high-priority case that OpenGL driver developers optimize for.


Once you have a vertex array object (and the appropriate program object used to draw with it, as well as other rendering state), rendering with it is quite easy.

Regardless of which function you use, you will be required to specify a Primitive type. This will be how OpenGL interprets the vertex data for that object.

As previously mentioned, there are two ways to send a stream of vertices: indexed and unindexed. All functions that draw indexd have the form glDraw*Elements*​, while functions that draw unindexed have the form glDraw*Arrays*​.

Basic Drawing

The basic drawing functions are these:

 void glDrawArrays( GLenum mode​, GLint first​, GLsizei count​ );
 void glDrawElements( GLenum mode​, GLsizei count​, GLenum type​, void * indices​ );

where, for glDrawArrays:

  • mode​ parameter is the Primitive type.
  • first​ and count​ values in define the range of elements to be pulled from the buffer.

as for glDrawElements:

  • count​ and indices​ parameters define the range of indices.
    • count​ defines how many indices to use.
    • indices​ defines the offset into the index buffer object (bound to GL_ELEMENT_ARRAY_BUFFER, stored in the VAO) to begin reading data.
  • type​ field describes what the type of the indices are:


The basic drawing functions are all you really need in order to send vertices for rendering. However, there are a number of ways to draw that optimize certain rendering cases.

Rendering with a different VAO from the last rendering command is usually a relatively expensive operation. So many of the optimization mechanisms are based on you storing the data for several meshes in the same buffer objects with the same vertex formats and other VAO data.


Binding a VAO is often an expensive operation. And there are many cases where you want to render a number of distinct meshes with a single draw call. All of the meshes must be in the same VAO, as must all of the index arrays if you are doing indexed rendering. Also, of course, they must use the same shader program with the same uniform values.

To render multiple primitives from a VAO at once, use this:

 void glMultiDrawArrays( GLenum mode​, GLint *first​, GLsizei *count​, GLsizei primcount​);

This function is conceptually implemented as:

void glMultiDrawArrays( GLenum mode, GLint *first, GLsizei *count, GLsizei primcount )
	for (int i = 0; i < primcount; i++)
		if (count[i] > 0)
			glDrawArrays(mode, first[i], count[i]);

Of course, you could write this function yourself. However, because it all happens in a single OpenGL call, the implementation has the opportunity to optimize this beyond what you could write.

There is an indexed form as well:

 void glMultiDrawElements( GLenum mode​, GLsizei *count​, GLenum type​, void **indices​, GLsizei primcount​ );

Similarly, this is implemented conceptually as:

void glMultiDrawElements( GLenum mode, GLsizei *count, GLenum type, void **indices, GLsizei primcount )
	for (int i = 0; i < primcount; i++)
		if (count[i]) > 0)
			glDrawElements(mode, count[i], type, indices[i]);

Multi-draw is useful for circumstances where you know that you are going to draw a lot of separate primitives of the same kind that all use the same shader. Typically, this would be a single conceptual object that you would always draw together in the same way. You simply pack all of the vertex data into the same VAO and buffer objects, using the various offsets to pick and choose between them.

Primitive Restart

Primitive restart functionality allows you to tell OpenGL that a particular index value means, not to source a vertex at that index, but to begin a new primitive of the same type with the next vertex. In essence, it is an alternative to glMultiDrawElements. This allows you to have an element buffer that contains multiple triangle strips or fans (or similar primitives where the start of a primitive has special behavior).

The way it works is with the function glPrimitiveRestartIndex. This function takes an index value. If this index is found in the index array, the system will start the primitive processing again as though a second rendering command had been issued. If you use a BaseVertex drawing function, this test is done before the base vertex is added to the restart. Using this feature also requires using glEnable(GL_PRIMITIVE_RESTART);​ to activate it, and the corresponding glDisable​ to turn it off.

Here is an example. Let's say you have an index array as follows:

 { 0 1 2 3 65535 2 3 4 5 }

If you render this as a triangle strip normally, you get 7 triangles. If you render it with glPrimitiveRestartIndex(65535)​ and the primitive restart enabled, then you will get 4 triangles:

 {0 1 2}, {1 2 3}, {2 3 4}, {3 4 5}

Primitive restart works with any of the versions of these functions.

Warning: It is technically legal to use this with non-indexed rendering. You should not do this, as it will not give you a useful result.

Base Index

All of the glVertexAttribPointer calls define the format of the vertices. That is, the way the vertex data is stored in the buffer objects. Changing this format is somewhat expensive in terms of performance.

If you have a number of meshes that all share the same vertex format, it would be useful to be able to put them all in a single set of buffer objects, one after the other. If we have two meshes, A and B, then their data would look like this:

 [A00 A01 A02 A03 A04... Ann B00 B01 B02... Bmm]

B's mesh data immediately follows A's mesh data, with no breaks inbetween.

The glDrawArrays call takes a start index. If we are using unindexed rendering, then this is all we need. We call glDrawArrays once with 0 as the start index and nn as the array count. Then we call it again with nn as the start index and mm as the array count.

Indexed rendering is often very useful, both for memory saving and performance. So it would be great if we can preserve this performance saving optimization when using indexed rendering.

In indexed rendering, each mesh also has an index buffers. glDrawElements takes an offset into the index buffer, so we can use the same mechanism to select which sets of indices to use.

The problem is the contents of these indices. The third vertex of mesh B is technically index 02. However, the actual index is determined by the location of that vertex relative to where the format was defined. And since we're trying to avoid redefining the format, the format still points to the start of the buffer. So the third vertex of mesh B is actually at index 02 + nn.

We could in fact store these indices in the index buffer that way. We could go through all of mesh B's indices and add nn to them. But we don't have to.

Instead, we can use this function:

 void glDrawElementsBaseVertex( GLenum mode​, GLsizei count​,
   GLenum type​, void *indices​, GLint basevertex​);

This works as glDrawElements does, except that basevertex​ is added to each index before pulling from the vertex data. So for mesh A, we pass a base vertex of 0 (or just use glDrawElements), and for mesh B, we pass a base vertex of nn.

Note: When combining with primitive restart, the restart test happens before the base index is added to the index.


It is often useful to be able to render multiple copies of the same mesh in different locations. If you're doing this with small numbers, like 5-20 or so, multiple draw commands with shader uniform changes between them (to tell which is in which location) is reasonably fast in performance. However, if you're doing this with large numbers of meshes, like 5,000+ or so, then it can be a performance problem.

Instancing is a way to get around this. The idea is that your vertex shader has some internal mechanism for deciding where each instance of the rendered mesh goes based on a single number. Perhaps it has a table (stored in a Buffer Texture or Uniform Buffer Object) that it indexes with the instance number to get the per-instance data it needs. Or perhaps it has a simple algorithm for computing the location of an instance based on its number.

Regardless of the mechanism, it is based the shader getting an instance number that changes only when it is rendering a new instance. If you want to do instanced rendering, you call:

 void glDrawArraysInstanced( GLenum mode​, GLint first​,
   GLsizei count​, GLsizei primcount​ );
 void glDrawElementsInstanced( GLenum mode​, GLsizei count​, 
   GLenum type​, const void *indices​, GLsizei primcount​ );

It will send the same vertices primcount​ number of times, as though you called glDrawArrays/Elements​ in a loop of primcount​ length. However, the vertex shader is given a special input value: gl_InstanceID​. It will receive a value from 0 to primcount​-1 based on which instance of the mesh is being rendered. This is the only mechanism the vertex shader has for differentiating between instances; it is up to the shader itself to decide how to use this information.


Implementations of OpenGL can often find it useful to know how much vertex data is being used in a buffer object. For non-indexed rendering, this is pretty easy to determine: the first​ and count​ parameters of the Arrays functions gives you appropriate information. For indexed rendering, this is more difficult, as the index buffer can use potentially any index up to its size.

Still for optimization purposes, it is useful for implementations to know the range of indexed rendering data. Implementations may even read index data manually to determine this.

The "Range" series of glDrawElements commands allows the user to specify that this indexed rendering call will never cause indices outside of the given range of values to be sourced. The call works as follows:

 void glDrawRangeElements( GLenum mode​, GLuint start​, 
   GLuint end​, GLsizei count​, GLenum type​, void *indices​ );

Unlike the "Arrays" functions, the start​ and end​ parameters specify the minimum and maximum index values (from the element buffer) that this draw call will use (rather than a first and count-style). If you try to violate this restriction, you will get implementation-behavior (ie: rendering may work fine or you may get garbage).

There is one index that is allowed outside of the area bound by start​ and end​: the primitive restart index. If primitive restart is set and enabled, it does not have to be within the given boundary.

Implementations may have a specific "sweet spot" for the range of indices, such that using indices within this range will have better performance. They expose such values with a pair of glGetIntegerv enumerators. To get the best performance, end​ - start​ should be less than or equal to GL_MAX_ELEMENTS_VERTICES, and count​ (the number of indices to be rendered) should be less than or equal to GL_MAX_ELEMENTS_INDICES.


It is often useful to combine these optimization techniques. Primitive restart can be combined with any of them, so long as they are using indexed rendering. The primitive restart comparison test, in the case of BaseVertex calls, is done before the base index is added to the index from the mesh.

Base vertex can be combined with any one of MultiDraw, Range, or Instancing. These functions are:

 void glMultiDrawElementsBaseVertex( GLenum mode​, 
   GLsizei *count​, GLenum type​, void **indices​, 
   GLsizei primcount​, GLint *basevertex​ );
 void glDrawRangeElementsBaseVertex( GLenum mode​, 
   GLuint start​, GLuint end​, GLsizei count​, GLenum type​, 
   void *indices​, GLint basevertex​ );
 void glDrawElementsInstancedBaseVertex( GLenum mode​, 
   GLsizei count​, GLenum type​, const void *indices​, 
   GLsizei primcount​, GLint basevertex​ );

In the case of MultiDraw, the basevertex​ parameter is an array, so each primitive can have its own base index.

None of the other features can be combined with one another. So Range does not combine with MultiDraw.

See Also