The WebGLRenderingContext.vertexAttribPointer() method of the WebGL API binds the buffer currently bound to gl.ARRAY_BUFFER to a generic vertex attribute of the current vertex buffer object and specifies its layout.

index

A GLuint specifying the index of the vertex attribute that is to be modified.

size

A GLint specifying the number of components per vertex attribute. Must be 1, 2, 3, or 4.

type

A GLenum specifying the data type of each component in the array. Possible values:

• gl.BYTE: signed 8-bit integer, with values in [-128, 127]
• gl.SHORT: signed 16-bit integer, with values in [-32768, 32767]
• gl.UNSIGNED_BYTE: unsigned 8-bit integer, with values in [0, 255]
• gl.UNSIGNED_SHORT: unsigned 16-bit integer, with values in [0,65535]
• gl.FLOAT: 32-bit IEEE floating point number

When using a WebGL 2 context, the following values are available additionally:

• gl.HALF_FLOAT: 16-bit IEEE floating point number
• gl.INT: 32-bit signed binary integer
• gl.UNSIGNED_INT: 32-bit unsigned binary integer
• gl.INT_2_10_10_10_REV: 32-bit signed integer with values in [-512, 511]
• gl.UNSIGNED_INT_2_10_10_10_REV: 32-bit unsigned integer with values in [0, 1023]

normalized

A GLboolean specifying whether integer data values should be normalized into a certain range when being cast to a float.

• For types gl.BYTE and gl.SHORT, normalizes the values to [-1, 1] if true.
• For types gl.UNSIGNED_BYTE and gl.UNSIGNED_SHORT, normalizes the values to [0, 1] if true.
• For types gl.FLOAT and gl.HALF_FLOAT, this parameter has no effect.

stride

A GLsizei specifying the offset in bytes between the beginning of consecutive vertex attributes. Cannot be negative or larger than 255. If stride is 0, the attribute is assumed to be tightly packed, that is, the attributes are not interleaved but each attribute is in a separate block, and the next vertex' attribute follows immediately after the current vertex.

offset

A GLintptr specifying an offset in bytes of the first component in the vertex attribute array. Must be a multiple of the byte length of type.

None (undefined).

• A gl.INVALID_VALUE error is thrown if stride or offset are negative.
• A gl.INVALID_OPERATION error is thrown if stride and offset are not multiples of the size of the data type.
• A gl.INVALID_OPERATION error is thrown if no WebGLBuffer is bound to the ARRAY_BUFFER target.
• When using a WebGL 2 context, a gl.INVALID_OPERATION error is thrown if this vertex attribute is defined as an integer in the vertex shader (e.g. uvec4 or ivec4, instead of vec4).

Let's assume we want to render some 3D geometry, and for that we will need to supply our vertices to the Vertex Shader. Each vertex has a few attributes, like position, normal vector, or texture coordinate, that are defined in an ArrayBuffer and will be supplied to the Vertex Buffer Object (VBO). First, we need to bind the WebGLBuffer we want to use to gl.ARRAY_BUFFER, then, with this method, gl.vertexAttribPointer(), we specify in what order the attributes are stored, and what data type they are in. In addition, we need to include the stride, which is the total byte length of all attributes for one vertex. Also, we have to call gl.enableVertexAttribArray() to tell WebGL that this attribute should be filled with data from our array buffer.

Usually, your 3D geometry is already in a certain binary format, so you need to read the specification of that specific format to figure out the memory layout. However, if you are designing the format yourself, or your geometry is in text files (like Wavefront .obj files) and must be converted into an ArrayBuffer at runtime, you have free choice on how to structure the memory. For highest performance, interleave the attributes and use the smallest data type that still accurately represents your geometry.

The maximum number of vertex attributes depends on the graphics card, and you can call gl.getParameter(gl.MAX_VERTEX_ATTRIBS) to get this value. On high-end graphics cards, the maximum is 16, on lower-end graphics cards, the value will be lower.

For each attribute, you must specify its index. This is independent from the location inside the array buffer, so your attributes can be sent in a different order than how they are stored in the array buffer. You have two options:

• Either you specify the index yourself. In this case, you call gl.bindAttribLocation() to connect a named attribute from the vertex shader to the index you want to use. This must be done before calling gl.linkProgram(). You can then provide this same index to gl.vertexAttribPointer().
• Alternatively, you use the index that is assigned by the graphics card when compiling the vertex shader. Depending on the graphics card, the index will vary, so you must call gl.getAttribLocation() to find out the index, and then provide this index to gl.vertexAttribPointer(). If you are using WebGL 2, you can specify the index yourself in the vertex shader code and override the default used by the graphics card, e.g. layout(location = 3) in vec4 position; would set the "position" attribute to index 3.

While the ArrayBuffer can be filled with both integers and floats, the attributes will always be converted to a float when they are sent to the vertex shader. If you need to use integers in your vertex shader code, you can either cast the float back to an integer in the vertex shader (e.g. (int) floatNumber), or use gl.vertexAttribIPointer() from WebGL2.

The vertex shader code may include a number of attributes, but we don't need to specify the values for each attribute. Instead, we can supply a default value that will be identical for all vertices. We can call gl.disableVertexAttribArray() to tell WebGL to use the default value, while calling gl.enableVertexAttribArray() will read the values from the array buffer as specified with gl.vertexAttribPointer().

Similarly, if our vertex shader expects e.g. a 4-component attribute with vec4 but in our gl.vertexAttribPointer() call we set the size to 2, then WebGL will set the first two components based on the array buffer, while the third and fourth components are taken from the default value.

The default value is vec4(0.0, 0.0, 0.0, 1.0) by default but we can specify a different default value with gl.vertexAttrib[1234]f[v]().

For example, your vertex shader may be using a position and a color attribute. Most meshes have the color specified at a per-vertex level, but some meshes are of a uniform shade. For those meshes, it is not necessary to place the same color for each vertex into the array buffer, so you use gl.vertexAttrib4fv() to set a constant color.

You can call gl.getVertexAttrib() and gl.getVertexAttribOffset() to get the current parameters for an attribute, e.g. the data type or whether the attribute should be normalized. Keep in mind that these WebGL functions have a slow performance and it is better to store the state inside your JavaScript application. However, these functions are great for debugging a WebGL context without touching the application code.

This example shows how to send your vertex attributes to the shader program. We use an imaginary data structure where the attributes of each vertex are stored interleaved with a length of 20 bytes per vertex:

1. position: We need to store the X, Y and Z coordinates. For highest precision, we use 32-bit floats; in total this uses 12 bytes.
2. normal vector: We need to store the X, Y and Z components of the normal vector, but since precision is not that important, we use 8-bit signed integers. For better performance, we align the data to 32 bits by also storing a fourth zero-valued component, bringing the total size to 4 bytes. Also, we tell WebGL to normalize the values because our normals are always in range [-1, 1].
3. texture coordinate: We need to store the U and V coordinates; for this 16-bit unsigned integers offer enough precision, the total size is 4 bytes. We also tell WebGL to normalize the values to [0, 1].

For example, the following vertex:

{
  "position": [1.0, 2.0, 1.5],
  "normal": [1.0, 0.0, 0.0],
  "texCoord": [0.5, 0.25]
}

Will be stored in the array buffer as follows:

First, we dynamically create the array buffer from JSON data using a DataView. Note the use of true because WebGL expects our data to be in little-endian.

// Load geometry with fetch() and Response.json()
const response = await fetch("assets/geometry.json");
const vertices = await response.json();

// Create array buffer
const buffer = new ArrayBuffer(20 * vertices.length);
// Fill array buffer
const dv = new DataView(buffer);
vertices.forEach((vertex, i) => {
  dv.setFloat32(20 * i, vertex.position[0], true);
  dv.setFloat32(20 * i + 4, vertex.position[1], true);
  dv.setFloat32(20 * i + 8, vertex.position[2], true);
  dv.setInt8(20 * i + 12, vertex.normal[0] * 0x7f);
  dv.setInt8(20 * i + 13, vertex.normal[1] * 0x7f);
  dv.setInt8(20 * i + 14, vertex.normal[2] * 0x7f);
  dv.setInt8(20 * i + 15, 0);
  dv.setUint16(20 * i + 16, vertex.texCoord[0] * 0xffff, true);
  dv.setUint16(20 * i + 18, vertex.texCoord[1] * 0xffff, true);
});

For higher performance, we could also do the previous JSON to ArrayBuffer conversion on the server-side, e.g. with Node.js. Then we could load the binary file and interpret it as an array buffer:

const response = await fetch("assets/geometry.bin");
const buffer = await response.arrayBuffer();

First, we create a new Vertex Buffer Object (VBO) and supply it with our array buffer:

//Bind array buffer to a Vertex Buffer Object
const vbo = gl.createBuffer();
gl.bindBuffer(gl.ARRAY_BUFFER, vbo);
gl.bufferData(gl.ARRAY_BUFFER, buffer, gl.STATIC_DRAW);

Then, we specify the memory layout of the array buffer, either by setting the index ourselves:

//Describe the layout of the buffer:
//1. position, not normalized
gl.vertexAttribPointer(0, 3, gl.FLOAT, false, 20, 0);
gl.enableVertexAttribArray(0);
//2. normal vector, normalized to [-1, 1]
gl.vertexAttribPointer(1, 4, gl.BYTE, true, 20, 12);
gl.enableVertexAttribArray(1);
//3. texture coordinates, normalized to [0, 1]
gl.vertexAttribPointer(2, 2, gl.UNSIGNED_SHORT, true, 20, 16);
gl.enableVertexAttribArray(2);

//Set the attributes in the vertex shader to the same indices
gl.bindAttribLocation(shaderProgram, 0, "position");
gl.bindAttribLocation(shaderProgram, 1, "normal");
gl.bindAttribLocation(shaderProgram, 2, "texUV");
//Since the attribute indices have changed, we must re-link the shader
//Note that this will reset all uniforms that were previously set.
gl.linkProgram(shaderProgram);

Or we can use the index provided by the graphics card instead of setting the index ourselves; this avoids the re-linking of the shader program.

const locPosition = gl.getAttribLocation(shaderProgram, "position");
gl.vertexAttribPointer(locPosition, 3, gl.FLOAT, false, 20, 0);
gl.enableVertexAttribArray(locPosition);

const locNormal = gl.getAttribLocation(shaderProgram, "normal");
gl.vertexAttribPointer(locNormal, 4, gl.BYTE, true, 20, 12);
gl.enableVertexAttribArray(locNormal);

const locTexUV = gl.getAttribLocation(shaderProgram, "texUV");
gl.vertexAttribPointer(locTexUV, 2, gl.UNSIGNED_SHORT, true, 20, 16);
gl.enableVertexAttribArray(locTexUV);

• Vertex Specification on the OpenGL wiki
• WebGL2RenderingContext.vertexAttribIPointer()