Texture

class arcade.gl.Texture2D(ctx: Context, size: tuple[int, int], *, components: int = 4, dtype: str = 'f1', data: BufferProtocol | None = None, filter: tuple[PyGLuint, PyGLuint] | None = None, wrap_x: PyGLuint | None = None, wrap_y: PyGLuint | None = None, target=3553, depth=False, samples: int = 0, immutable: bool = False, internal_format: PyGLuint | None = None, compressed: bool = False, compressed_data: bool = False)[source]

Bases:

An OpenGL 2D texture. We can create an empty black texture or a texture from byte data. A texture can also be created with different datatypes such as float, integer or unsigned integer.

The best way to create a texture instance is through arcade.gl.Context.texture()

Supported dtype values are:

# Float formats
'f1': UNSIGNED_BYTE
'f2': HALF_FLOAT
'f4': FLOAT
# int formats
'i1': BYTE
'i2': SHORT
'i4': INT
# uint formats
'u1': UNSIGNED_BYTE
'u2': UNSIGNED_SHORT
'u4': UNSIGNED_INT
Parameters:
  • ctx – The context the object belongs to

  • size – The size of the texture

  • components – The number of components (1: R, 2: RG, 3: RGB, 4: RGBA)

  • dtype – The data type of each component: f1, f2, f4 / i1, i2, i4 / u1, u2, u4

  • data – The texture data (optional). Can be bytes or any object supporting the buffer protocol.

  • filter – The minification/magnification filter of the texture

  • wrap_x – Wrap mode x

  • wrap_y – Wrap mode y

  • target – The texture type (Ignored. Legacy)

  • depth – creates a depth texture if True

  • samples – Creates a multisampled texture for values > 0. This value will be clamped between 0 and the max sample capability reported by the drivers.

  • immutable – Make the storage (not the contents) immutable. This can sometimes be required when using textures with compute shaders.

  • internal_format – The internal format of the texture

  • compressed – Is the texture compressed?

  • compressed_data – The raw compressed data

resize(size: tuple[int, int])[source]

Resize the texture. This will re-allocate the internal memory and all pixel data will be lost.

Note

Immutable textures cannot be resized.

Parameters:

size – The new size of the texture

property ctx: Context

The context this texture belongs to.

property glo: c_uint

The OpenGL texture id

property compressed: bool

Is this using a compressed format?

property width: int

The width of the texture in pixels

property height: int

The height of the texture in pixels

property dtype: str

The data type of each component

property size: tuple[int, int]

The size of the texture as a tuple

property samples: int

Number of samples if multisampling is enabled (read only)

property byte_size: int

The byte size of the texture.

property components: int

Number of components in the texture

property component_size: int

Size in bytes of each component

property depth: bool

If this is a depth texture.

property immutable: bool

Does this texture have immutable storage?

property swizzle: str

The swizzle mask of the texture (Default 'RGBA').

The swizzle mask change/reorder the vec4 value returned by the texture() function in a GLSL shaders. This is represented by a 4 character string were each character can be:

'R' GL_RED
'G' GL_GREEN
'B' GL_BLUE
'A' GL_ALPHA
'0' GL_ZERO
'1' GL_ONE

Example:

# Alpha channel will always return 1.0
texture.swizzle = 'RGB1'

# Only return the red component. The rest is masked to 0.0
texture.swizzle = 'R000'

# Reverse the components
texture.swizzle = 'ABGR'
property filter: tuple[int, int]

Get or set the (min, mag) filter for this texture.

These are rules for how a texture interpolates. The filter is specified for minification and magnification.

Default value is LINEAR, LINEAR. Can be set to NEAREST, NEAREST for pixelated graphics.

When mipmapping is used the min filter needs to be one of the MIPMAP variants.

Accepted values:

# Enums can be accessed on the context or arcade.gl
NEAREST                # Nearest pixel
LINEAR                 # Linear interpolate
NEAREST_MIPMAP_NEAREST # Minification filter for mipmaps
LINEAR_MIPMAP_NEAREST  # Minification filter for mipmaps
NEAREST_MIPMAP_LINEAR  # Minification filter for mipmaps
LINEAR_MIPMAP_LINEAR   # Minification filter for mipmaps

Also see

property wrap_x: int

Get or set the horizontal wrapping of the texture.

This decides how textures are read when texture coordinates are outside the [0.0, 1.0] area. Default value is REPEAT.

Valid options are:

# Note: Enums can also be accessed in arcade.gl
# Repeat pixels on the y axis
texture.wrap_x = ctx.REPEAT
# Repeat pixels on the y axis mirrored
texture.wrap_x = ctx.MIRRORED_REPEAT
# Repeat the edge pixels when reading outside the texture
texture.wrap_x = ctx.CLAMP_TO_EDGE
# Use the border color (black by default) when reading outside the texture
texture.wrap_x = ctx.CLAMP_TO_BORDER
property wrap_y: int

Get or set the horizontal wrapping of the texture.

This decides how textures are read when texture coordinates are outside the [0.0, 1.0] area. Default value is REPEAT.

Valid options are:

# Note: Enums can also be accessed in arcade.gl
# Repeat pixels on the x axis
texture.wrap_x = ctx.REPEAT
# Repeat pixels on the x axis mirrored
texture.wrap_x = ctx.MIRRORED_REPEAT
# Repeat the edge pixels when reading outside the texture
texture.wrap_x = ctx.CLAMP_TO_EDGE
# Use the border color (black by default) when reading outside the texture
texture.wrap_x = ctx.CLAMP_TO_BORDER
property anisotropy: float

Get or set the anisotropy for this texture.

property compare_func: str | None

Get or set the compare function for a depth texture:

texture.compare_func = None  # Disable depth comparison completely
texture.compare_func = '<='  # GL_LEQUAL
texture.compare_func = '<'   # GL_LESS
texture.compare_func = '>='  # GL_GEQUAL
texture.compare_func = '>'   # GL_GREATER
texture.compare_func = '=='  # GL_EQUAL
texture.compare_func = '!='  # GL_NOTEQUAL
texture.compare_func = '0'   # GL_NEVER
texture.compare_func = '1'   # GL_ALWAYS
read(level: int = 0, alignment: int = 1) bytes[source]

Read the contents of the texture.

Parameters:
  • level – The texture level to read

  • alignment – Alignment of the start of each row in memory in number of bytes. Possible values: 1,2,4

write(data: ByteString | memoryview | array | Array | Buffer, level: int = 0, viewport=None) None[source]

Write byte data from the passed source to the texture.

The data value can be either an arcade.gl.Buffer or anything that implements the Buffer Protocol.

The latter category includes bytes, bytearray, array.array, and more. You may need to use typing workarounds for non-builtin types. See Writing Raw Bytes to GL Buffers & Textures for more information.

Parameters:
  • dataBuffer or buffer protocol object with data to write.

  • level – The texture level to write

  • viewport – The area of the texture to write. 2 or 4 component tuple

build_mipmaps(base: int = 0, max_level: int = 1000) None[source]

Generate mipmaps for this texture.

The default values usually work well.

Mipmaps are successively smaller versions of an original texture with special filtering applied. Using mipmaps allows OpenGL to render scaled versions of original textures with fewer scaling artifacts.

Mipmaps can be made for textures of any size. Each mipmap version halves the width and height of the previous one (e.g. 256 x 256, 128 x 128, 64 x 64, etc) down to a minimum of 1 x 1.

Note

Mipmaps will only be used if a texture’s filter is configured with a mipmap-type minification:

# Set up linear interpolating minification filter
texture.filter = ctx.LINEAR_MIPMAP_LINEAR, ctx.LINEAR
Parameters:
  • base – Level the mipmaps start at (usually 0)

  • max_level – The maximum number of levels to generate

Also see: https://www.khronos.org/opengl/wiki/Texture#Mip_maps

delete()[source]

Destroy the underlying OpenGL resource.

Don’t use this unless you know exactly what you are doing.

static delete_glo(ctx: Context, glo: gl.GLuint)[source]

Destroy the texture.

This is called automatically when the object is garbage collected.

Parameters:
  • ctx – OpenGL Context

  • glo – The OpenGL texture id

use(unit: int = 0) None[source]

Bind the texture to a channel,

Parameters:

unit – The texture unit to bind the texture.

bind_to_image(unit: int, read: bool = True, write: bool = True, level: int = 0)[source]

Bind textures to image units.

Note that either or both read and write needs to be True. The supported modes are: read only, write only, read-write

Parameters:
  • unit – The image unit

  • read – The compute shader intends to read from this image

  • write – The compute shader intends to write to this image

  • level – The mipmap level to bind

get_handle(resident: bool = True) int[source]

Get a handle for bindless texture access.

Once a handle is created its parameters cannot be changed. Attempting to do so will have no effect. (filter, wrap etc). There is no way to undo this immutability.

Handles cannot be used by shaders until they are resident. This method can be called multiple times to move a texture in and out of residency:

>> texture.get_handle(resident=False)
4294969856
>> texture.get_handle(resident=True)
4294969856

Ths same handle is returned if the handle already exists.

Note

Limitations from the OpenGL wiki

The amount of storage available for resident images/textures may be less than the total storage for textures that is available. As such, you should attempt to minimize the time a texture spends being resident. Do not attempt to take steps like making textures resident/un-resident every frame or something. But if you are finished using a texture for some time, make it un-resident.

Parameters:

resident – Make the texture resident.