GPU Types (gpu.types)

class gpu.types.Buffer(format, dimensions, data)

For Python access to GPU functions requiring a pointer.

Parameters

format – Format type to interpret the buffer. Possible values are FLOAT, INT, UINT, UBYTE, UINT_24_8 and 10_11_11_REV.
dimensions (int) – Array describing the dimensions.
data (sequence) – Optional data array.

return the buffer as a list

dimensions: Undocumented, consider contributing.

class gpu.types.GPUBatch(type, buf, elem=None)

Reusable container for drawable geometry.

Parameters

type (str) – The primitive type of geometry to be drawn. Possible values are POINTS, LINES, TRIS, LINE_STRIP, LINE_LOOP, TRI_STRIP, TRI_FAN, LINES_ADJ, TRIS_ADJ and LINE_STRIP_ADJ.
buf (gpu.types.GPUVertBuf) – Vertex buffer containing all or some of the attributes required for drawing.
elem (gpu.types.GPUIndexBuf) – An optional index buffer.

draw(program=None)

Run the drawing program with the parameters assigned to the batch.

Parameters: program (gpu.types.GPUShader) – Program that performs the drawing operations. If None is passed, the last program set to this batch will run.

program_set(program)

Assign a shader to this batch that will be used for drawing when not overwritten later. Note: This method has to be called in the draw context that the batch will be drawn in. This function does not need to be called when you always set the shader when calling gpu.types.GPUBatch.draw().

Parameters: program (gpu.types.GPUShader) – The program/shader the batch will use in future draw calls.

vertbuf_add(buf)

Add another vertex buffer to the Batch. It is not possible to add more vertices to the batch using this method. Instead it can be used to add more attributes to the existing vertices. A good use case would be when you have a separate vertex buffer for vertex positions and vertex normals. Current a batch can have at most 16 vertex buffers.

Parameters: buf (gpu.types.GPUVertBuf) – The vertex buffer that will be added to the batch.

class gpu.types.GPUFrameBuffer(depth_slot=None, color_slots=None)

This object gives access to framebuffer functionallities. When a ‘layer’ is specified in a argument, a single layer of a 3D or array texture is attached to the frame-buffer. For cube map textures, layer is translated into a cube map face.

Parameters

depth_slot (gpu.types.GPUTexture, dict or Nonetype) – GPUTexture to attach or a dict containing keywords: ‘texture’, ‘layer’ and ‘mip’.
color_slots (tuple or Nonetype) – Tuple where each item can be a GPUTexture or a dict containing keywords: ‘texture’, ‘layer’ and ‘mip’.

bind(): Context manager to ensure balanced bind calls, even in the case of an error.

clear(color=None, depth=None, stencil=None)

Fill color, depth and stencil textures with specific value. Common values: color=(0.0, 0.0, 0.0, 1.0), depth=1.0, stencil=0.

Parameters

color (sequence of 3 or 4 floats) – float sequence each representing (r, g, b, a).
depth (float) – depth value.
stencil (int) – stencil value.

read_color(x, y, xsize, ysize, channels, slot, format, data=data)

Read a block of pixels from the frame buffer.

Parameters

y (x,) – Lower left corner of a rectangular block of pixels.
ysize (xsize,) – Dimensions of the pixel rectangle.
channels (int) – Number of components to read.
slot (int) – The framebuffer slot to read data from.
format – The format that describes the content of a single channel. Possible values are FLOAT, INT, UINT, UBYTE, UINT_24_8 and 10_11_11_REV.
data (gpu.types.Buffer) – Optional Buffer object to fill with the pixels values.

Returns

The Buffer with the read pixels.

Return type

gpu.types.Buffer

read_depth(x, y, xsize, ysize, data=data)

Read a pixel depth block from the frame buffer.

Parameters

y (x,) – Lower left corner of a rectangular block of pixels.
ysize (xsize,) – Dimensions of the pixel rectangle.
data (gpu.types.Buffer) – Optional Buffer object to fill with the pixels values.

Returns

The Buffer with the read pixels.

Return type

gpu.types.Buffer

viewport_get(): Returns position and dimension to current viewport.

viewport_set(x, y, xsize, ysize)

Set the viewport for this framebuffer object. Note: The viewport state is not saved upon framebuffer rebind.

Parameters

y (x,) – lower left corner of the viewport_set rectangle, in pixels.
ysize (xsize,) – width and height of the viewport_set.

is_bound: Checks if this is the active framebuffer in the context.

class gpu.types.GPUIndexBuf(type, seq)

Contains an index buffer.

Parameters

type (str) – The primitive type this index buffer is composed of. Possible values are POINTS, LINES, TRIS and LINE_STRIP_ADJ.
seq (1D or 2D sequence) – Indices this index buffer will contain. Whether a 1D or 2D sequence is required depends on the type. Optionally the sequence can support the buffer protocol.

class gpu.types.GPUOffScreen(width, height)

This object gives access to off screen buffers.

Parameters

width (int) – Horizontal dimension of the buffer.
height (int) – Vertical dimension of the buffer.

bind(): Context manager to ensure balanced bind calls, even in the case of an error.

draw_view3d(scene, view_layer, view3d, region, view_matrix, projection_matrix, do_color_management=False)

Draw the 3d viewport in the offscreen object.

Parameters

scene (bpy.types.Scene) – Scene to draw.
view_layer (bpy.types.ViewLayer) – View layer to draw.
view3d (bpy.types.SpaceView3D) – 3D View to get the drawing settings from.
region (bpy.types.Region) – Region of the 3D View (required as temporary draw target).
view_matrix (mathutils.Matrix) – View Matrix (e.g. camera.matrix_world.inverted()).
projection_matrix (mathutils.Matrix) – Projection Matrix (e.g. camera.calc_matrix_camera(...)).
do_color_management (bool) – Color manage the output.

free(): Free the offscreen object. The framebuffer, texture and render objects will no longer be accessible.

unbind(restore=True)

Unbind the offscreen object.

Parameters: restore (bool) – Restore the OpenGL state, can only be used when the state has been saved before.

color_texture

OpenGL bindcode for the color texture.

Type: int

height

Height of the texture.

Type: int

texture_color

The color texture attached.

Type: gpu.types.GPUTexture

width

Width of the texture.

Type: int

class gpu.types.GPUShader(vertexcode, fragcode, geocode=None, libcode=None, defines=None, name='pyGPUShader')

GPUShader combines multiple GLSL shaders into a program used for drawing. It must contain at least a vertex and fragment shaders.

The GLSL #version directive is automatically included at the top of shaders, and set to 330. Some preprocessor directives are automatically added according to the Operating System or availability: GPU_ATI, GPU_NVIDIA and GPU_INTEL.

The following extensions are enabled by default if supported by the GPU: GL_ARB_texture_gather, GL_ARB_texture_cube_map_array and GL_ARB_shader_draw_parameters.

For drawing user interface elements and gizmos, use fragOutput = blender_srgb_to_framebuffer_space(fragOutput) to transform the output sRGB colors to the frame-buffer color-space.

Parameters

vertexcode (str) – Vertex shader code.
fragcode – Fragment shader code.
geocode – Geometry shader code.
libcode – Code with functions and presets to be shared between shaders.
defines – Preprocessor directives.
name – Name of shader code, for debugging purposes.

attr_from_name(name)

Get attribute location by name.

Parameters: name (str) – The name of the attribute variable whose location is to be queried.
Returns: The location of an attribute variable.
Return type: int

bind(): Bind the shader object. Required to be able to change uniforms of this shader.

calc_format()

Build a new format based on the attributes of the shader.

Returns: vertex attribute format for the shader
Return type: gpu.types.GPUVertFormat

uniform_block(name, ubo)

Specify the value of an uniform buffer object variable for the current GPUShader.

Parameters

name (str) – name of the uniform variable whose UBO is to be specified.
ubo – Uniform Buffer to attach.

uniform_block_from_name(name)

Get uniform block location by name.

Parameters: name (str) – Name of the uniform block variable whose location is to be queried.
Returns: The location of the uniform block variable.
Return type: int

uniform_bool(name, seq)

Specify the value of a uniform variable for the current program object.

Parameters

name (str) – Name of the uniform variable whose value is to be changed.
seq (sequence of bools) – Value that will be used to update the specified uniform variable.

uniform_float(name, value)

Specify the value of a uniform variable for the current program object.

Parameters

name (str) – Name of the uniform variable whose value is to be changed.
value (single number or sequence of numbers) – Value that will be used to update the specified uniform variable.

uniform_from_name(name)

Get uniform location by name.

Parameters: name (str) – Name of the uniform variable whose location is to be queried.
Returns: Location of the uniform variable.
Return type: int

uniform_int(name, seq)

Specify the value of a uniform variable for the current program object.

Parameters

name (str) – name of the uniform variable whose value is to be changed.
seq (sequence of numbers) – Value that will be used to update the specified uniform variable.

uniform_sampler(name, texture)

Specify the value of a texture uniform variable for the current GPUShader.

Parameters

name (str) – name of the uniform variable whose texture is to be specified.
texture (gpu.types.GPUTexture) – Texture to attach.

uniform_vector_float(location, buffer, length, count)

Set the buffer to fill the uniform.

Parameters

location (int) – Location of the uniform variable to be modified.
buffer (sequence of floats) – The data that should be set. Can support the buffer protocol.
length (int) –
Size of the uniform data type:
- 1: float
- 2: vec2 or float[2]
- 3: vec3 or float[3]
- 4: vec4 or float[4]
- 9: mat3
- 16: mat4
count (int) – Specifies the number of elements, vector or matrices that are to be modified.

uniform_vector_int(location, buffer, length, count): See GPUShader.uniform_vector_float(…) description.

name

The name of the shader object for debugging purposes (read-only).

Type: str

program

The name of the program object for use by the OpenGL API (read-only).

Type: int

class gpu.types.GPUTexture(size, layers=0, is_cubemap=False, format='RGBA8', data=None)

This object gives access to off GPU textures.

Parameters

size (tuple or int) – Dimensions of the texture 1D, 2D, 3D or cubemap.
layers (int) – Number of layers in texture array or number of cubemaps in cubemap array
is_cubemap (int) – Indicates the creation of a cubemap texture.
format (str) – Internal data format inside GPU memory. Possible values are: RGBA8UI, RGBA8I, RGBA8, RGBA32UI, RGBA32I, RGBA32F, RGBA16UI, RGBA16I, RGBA16F, RGBA16, RG8UI, RG8I, RG8, RG32UI, RG32I, RG32F, RG16UI, RG16I, RG16F, RG16, R8UI, R8I, R8, R32UI, R32I, R32F, R16UI, R16I, R16F, R16, R11F_G11F_B10F, DEPTH32F_STENCIL8, DEPTH24_STENCIL8, SRGB8_A8, RGB16F, SRGB8_A8_DXT1, SRGB8_A8_DXT3, SRGB8_A8_DXT5, RGBA8_DXT1, RGBA8_DXT3, RGBA8_DXT5, DEPTH_COMPONENT32F, DEPTH_COMPONENT24, DEPTH_COMPONENT16,
data (gpu.types.Buffer) – Buffer object to fill the texture.

clear(format='FLOAT', value=(0.0, 0.0, 0.0, 1.0))

Fill texture with specific value.

Parameters

format – The format that describes the content of a single item. Possible values are FLOAT, INT, UINT, UBYTE, UINT_24_8 and 10_11_11_REV.
value (sequence of 1, 2, 3 or 4 values) – sequence each representing the value to fill.

read(): Creates a buffer with the value of all pixels.

format

Format of the texture.

Type: str

height

Height of the texture.

Type: int

width

Width of the texture.

Type: int

class gpu.types.GPUUniformBuf(data)

This object gives access to off uniform buffers.

Parameters: data (gpu.types.Buffer) – Buffer object.

update(data): Update the data of the uniform buffer object.

class gpu.types.GPUVertBuf(len, format)

Contains a VBO.

Parameters

len – Amount of vertices that will fit into this buffer.
format – Vertex format.

attr_fill(id, data)

Insert data into the buffer for a single attribute.

Parameters

id (int or str) – Either the name or the id of the attribute.
data (sequence of floats, ints, vectors or matrices) – Sequence of data that should be stored in the buffer

class gpu.types.GPUVertFormat

This object contains information about the structure of a vertex buffer.

attr_add(id, comp_type, len, fetch_mode)

Add a new attribute to the format.

Parameters

id (str) – Name the attribute. Often position, normal, …
comp_type (str) – The data type that will be used store the value in memory. Possible values are I8, U8, I16, U16, I32, U32, F32 and I10.
len (int) – How many individual values the attribute consists of (e.g. 2 for uv coordinates).
fetch_mode (str) – How values from memory will be converted when used in the shader. This is mainly useful for memory optimizations when you want to store values with reduced precision. E.g. you can store a float in only 1 byte but it will be converted to a normal 4 byte float when used. Possible values are FLOAT, INT, INT_TO_FLOAT_UNIT and INT_TO_FLOAT.