pyglet.image

Submodules

Details

Image load, capture and high-level texture functions.

Only basic functionality is described here; for full reference see the accompanying documentation.

To load an image:

from pyglet import image
pic = image.load('picture.png')

The supported image file types include PNG, BMP, GIF, JPG, and many more, somewhat depending on the operating system. To load an image from a file-like object instead of a filename:

pic = image.load('hint.jpg', file=fileobj)

The hint helps the module locate an appropriate decoder to use based on the file extension. It is optional.

Once loaded, images can be used directly by most other modules of pyglet. All images have a width and height you can access:

width, height = pic.width, pic.height

You can extract a region of an image (this keeps the original image intact; the memory is shared efficiently):

subimage = pic.get_region(x, y, width, height)

Remember that y-coordinates are always increasing upwards.

Drawing images

To draw an image at some point on the screen:

pic.blit(x, y, z)

This assumes an appropriate view transform and projection have been applied.

Some images have an intrinsic “anchor point”: this is the point which will be aligned to the x and y coordinates when the image is drawn. By default the anchor point is the lower-left corner of the image. You can use the anchor point to center an image at a given point, for example:

pic.anchor_x = pic.width // 2
pic.anchor_y = pic.height // 2
pic.blit(x, y, z)

Texture access

If you are using OpenGL directly, you can access the image as a texture:

texture = pic.get_texture()

(This is the most efficient way to obtain a texture; some images are immediately loaded as textures, whereas others go through an intermediate form). To use a texture with pyglet.gl:

from pyglet.gl import *
glEnable(texture.target)        # typically target is GL_TEXTURE_2D
glBindTexture(texture.target, texture.id)
# ... draw with the texture

Pixel access

To access raw pixel data of an image:

rawimage = pic.get_image_data()

(If the image has just been loaded this will be a very quick operation; however if the image is a texture a relatively expensive readback operation will occur). The pixels can be accessed as a string:

format = 'RGBA'
pitch = rawimage.width * len(format)
pixels = rawimage.get_data(format, pitch)

“format” strings consist of characters that give the byte order of each color component. For example, if rawimage.format is ‘RGBA’, there are four color components: red, green, blue and alpha, in that order. Other common format strings are ‘RGB’, ‘LA’ (luminance, alpha) and ‘I’ (intensity).

The “pitch” of an image is the number of bytes in a row (this may validly be more than the number required to make up the width of the image, it is common to see this for word alignment). If “pitch” is negative the rows of the image are ordered from top to bottom, otherwise they are ordered from bottom to top.

Retrieving data with the format and pitch given in ImageData.format and ImageData.pitch avoids the need for data conversion (assuming you can make use of the data in this arbitrary format).

Classes

Images

class AbstractImage(width, height)

Abstract class representing an image.

Parameters

widthint: Width of image
heightint: Height of image
anchor_xint: X coordinate of anchor, relative to left edge of image data
anchor_yint: Y coordinate of anchor, relative to bottom edge of image data

blit(x, y, z=0)

Draw this image to the active framebuffers.

The image will be drawn with the lower-left corner at (x - anchor_x, y - anchor_y, z).

blit_into(source, x, y, z)

Draw source on this image.

source will be copied into this image such that its anchor point is aligned with the x and y parameters. If this image is a 3D texture, the z coordinate gives the image slice to copy into.

Note that if source is larger than this image (or the positioning would cause the copy to go out of bounds) then you must pass a region of source to this method, typically using get_region().

blit_to_texture(target, level, x, y, z=0)

Draw this image on the currently bound texture at target.

This image is copied into the texture such that this image’s anchor point is aligned with the given x and y coordinates of the destination texture. If the currently bound texture is a 3D texture, the z coordinate gives the image slice to blit into.

get_image_data()

Get an ImageData view of this image.

Changes to the returned instance may or may not be reflected in this image.

Return type: ImageData

New in version 1.1.

get_mipmapped_texture()

Retrieve a Texture instance with all mipmap levels filled in.

Return type: Texture

New in version 1.1.

get_region(x, y, width, height)

Retrieve a rectangular region of this image.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

AbstractImage

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

save(filename=None, file=None, encoder=None)

Save this image to a file.

Parameters

filenamestr: Used to set the image file format, and to open the output file if file is unspecified.
filefile-like object or None: File to write image data to.
encoderImageEncoder or None: If unspecified, all encoders matching the filename extension are tried. If all fail, the exception from the first one attempted is raised.

anchor_x = 0

anchor_y = 0

class BufferImage(x, y, width, height)

Bases: AbstractImage

An abstract framebuffer.

get_image_data()

Get an ImageData view of this image.

Changes to the returned instance may or may not be reflected in this image.

Return type: ImageData

New in version 1.1.

get_region(x, y, width, height)

Retrieve a rectangular region of this image.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

AbstractImage

format = '': The format string used for image data.

gl_buffer = 1029: The OpenGL read and write target for this buffer.

gl_format = 0: The OpenGL format constant for image data.

owner = None

class BufferImageMask(x, y, width, height)

Bases: BufferImage

A single bit of the stencil buffer.

format = 'L': The format string used for image data.

gl_format = 6401: The OpenGL format constant for image data.

class ColorBufferImage(x, y, width, height)

Bases: BufferImage

A color framebuffer.

This class is used to wrap the primary color buffer (i.e., the back buffer)

blit_to_texture(target, level, x, y, z)

Draw this image on the currently bound texture at target.

This image is copied into the texture such that this image’s anchor point is aligned with the given x and y coordinates of the destination texture. If the currently bound texture is a 3D texture, the z coordinate gives the image slice to blit into.

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

format = 'RGBA': The format string used for image data.

gl_format = 6408: The OpenGL format constant for image data.

class DepthBufferImage(x, y, width, height)

Bases: BufferImage

The depth buffer.

blit_to_texture(target, level, x, y, z)

Draw this image on the currently bound texture at target.

This image is copied into the texture such that this image’s anchor point is aligned with the given x and y coordinates of the destination texture. If the currently bound texture is a 3D texture, the z coordinate gives the image slice to blit into.

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

format = 'L': The format string used for image data.

gl_format = 6402: The OpenGL format constant for image data.

class Texture(width, height, target, tex_id)

Bases: AbstractImage

An image loaded into video memory that can be efficiently drawn to the framebuffer.

Typically, you will get an instance of Texture by accessing the texture member of any other AbstractImage.

Parameters

region_classclass (subclass of TextureRegion): Class to use when constructing regions of this texture.
tex_coordstuple: 12-tuple of float, named (u1, v1, r1, u2, v2, r2, …). u, v, r give the 3D texture coordinates for vertices 1-4. The vertices are specified in the order bottom-left, bottom-right, top-right and top-left.
targetint: The GL texture target (e.g., GL_TEXTURE_2D).
levelint: The mipmap level of this texture.

region_class: alias of TextureRegion

bind(texture_unit: int = 0): Bind to a specific Texture Unit by number.

bind_image_texture(unit, level=0, layered=False, layer=0, access=35002, fmt=34836): Bind as an ImageTexture for use with a ComputeShaderProgram.

Note

OpenGL 4.3, or 4.2 with the GL_ARB_compute_shader extention is required.

blit(x, y, z=0, width=None, height=None)

Draw this image to the active framebuffers.

The image will be drawn with the lower-left corner at (x - anchor_x, y - anchor_y, z).

blit_into(source, x, y, z)

Draw source on this image.

source will be copied into this image such that its anchor point is aligned with the x and y parameters. If this image is a 3D texture, the z coordinate gives the image slice to copy into.

Note that if source is larger than this image (or the positioning would cause the copy to go out of bounds) then you must pass a region of source to this method, typically using get_region().

classmethod create(width, height, target=3553, internalformat=32856, min_filter=None, mag_filter=None, fmt=6408, blank_data=True)

Create a Texture

Create a Texture with the specified dimentions, target and format. On return, the texture will be bound.

Parameters

widthint: Width of texture in pixels.
heightint: Height of texture in pixels.
targetint: GL constant giving texture target to use, typically GL_TEXTURE_2D.
internalformatint: GL constant giving internal format of texture; for example, GL_RGBA. The internal format decides how the texture data will be stored internally. If None, the texture will be created but not initialized.
min_filterint: The minifaction filter used for this texture, commonly GL_LINEAR or GL_NEAREST
mag_filterint: The magnification filter used for this texture, commonly GL_LINEAR or GL_NEAREST
fmtint: GL constant giving format of texture; for example, GL_RGBA. The format specifies what format the pixel data we’re expecting to write to the texture and should ideally be the same as for internal format.
blank_databool: Setting to True will initialize the texture data with all zeros. Setting False, will initialize Texture with no data.

Return type

Texture

get_image_data(z=0)

Get the image data of this texture.

Changes to the returned instance will not be reflected in this texture.

Parameters

zint: For 3D textures, the image slice to retrieve.

Return type

ImageData

get_region(x, y, width, height)

Retrieve a rectangular region of this image.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

AbstractImage

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

get_transform(flip_x=False, flip_y=False, rotate=0)

Create a copy of this image applying a simple transformation.

The transformation is applied to the texture coordinates only; get_image_data() will return the untransformed data. The transformation is applied around the anchor point.

Parameters

flip_xbool: If True, the returned image will be flipped horizontally.
flip_ybool: If True, the returned image will be flipped vertically.
rotateint: Degrees of clockwise rotation of the returned image. Only 90-degree increments are supported.

Return type

TextureRegion

colors = (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)

default_mag_filter = 9729

default_min_filter = 9729

images = 1

level = 0

tex_coords = (0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0)

tex_coords_order = (0, 1, 2, 3)

property uv: Tuple containing the left, bottom, right, top 2D texture coordinates.

x = 0

y = 0

z = 0

class DepthTexture(width, height, target, tex_id)

Bases: Texture

A texture with depth samples (typically 24-bit).

blit_into(source, x, y, z)

Draw source on this image.

source will be copied into this image such that its anchor point is aligned with the x and y parameters. If this image is a 3D texture, the z coordinate gives the image slice to copy into.

Note that if source is larger than this image (or the positioning would cause the copy to go out of bounds) then you must pass a region of source to this method, typically using get_region().

class TextureRegion(x, y, z, width, height, owner)

Bases: Texture

A rectangular region of a texture, presented as if it were a separate texture.

blit_into(source, x, y, z)

Draw source on this image.

source will be copied into this image such that its anchor point is aligned with the x and y parameters. If this image is a 3D texture, the z coordinate gives the image slice to copy into.

Note that if source is larger than this image (or the positioning would cause the copy to go out of bounds) then you must pass a region of source to this method, typically using get_region().

get_image_data()

Get the image data of this texture.

Changes to the returned instance will not be reflected in this texture.

Parameters

zint: For 3D textures, the image slice to retrieve.

Return type

ImageData

get_region(x, y, width, height)

Retrieve a rectangular region of this image.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

AbstractImage

class TileableTexture(width, height, target, tex_id)

Bases: Texture

A texture that can be tiled efficiently.

Use create_for_image classmethod to construct.

blit_tiled(x, y, z, width, height)

Blit this texture tiled over the given area.

The image will be tiled with the bottom-left corner of the destination rectangle aligned with the anchor point of this texture.

classmethod create_for_image(image)

get_region(x, y, width, height)

Retrieve a rectangular region of this image.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

AbstractImage

Image Sequences

class AbstractImageSequence

Abstract sequence of images.

The sequence is useful for storing image animations or slices of a volume. For efficient access, use the texture_sequence member. The class also implements the sequence interface (__len__, __getitem__, __setitem__).

get_animation(period, loop=True)

Create an animation over this image sequence for the given constant framerate.

:Parameters

periodfloat: Number of seconds to display each frame.
loopbool: If True, the animation will loop continuously.

Return type: Animation

New in version 1.1.

get_texture_sequence()

Get a TextureSequence.

Return type: TextureSequence

New in version 1.1.

class TextureSequence

Bases: AbstractImageSequence

Interface for a sequence of textures.

Typical implementations store multiple TextureRegion s within one Texture so as to minimise state changes.

get_texture_sequence()

Get a TextureSequence.

Return type: TextureSequence

New in version 1.1.

class UniformTextureSequence

Bases: TextureSequence

Interface for a sequence of textures, each with the same dimensions.

Parameters

item_widthint: Width of each texture in the sequence.
item_heightint: Height of each texture in the sequence.

property item_height

property item_width

class TextureGrid(grid)

Bases: TextureRegion, UniformTextureSequence

A texture containing a regular grid of texture regions.

To construct, create an ImageGrid first:

image_grid = ImageGrid(...)
texture_grid = TextureGrid(image_grid)

The texture grid can be accessed as a single texture, or as a sequence of TextureRegion. When accessing as a sequence, you can specify integer indexes, in which the images are arranged in rows from the bottom-left to the top-right:

# assume the texture_grid is 3x3:
current_texture = texture_grid[3] # get the middle-left image

You can also specify tuples in the sequence methods, which are addressed as row, column:

# equivalent to the previous example:
current_texture = texture_grid[1, 0]

When using tuples in a slice, the returned sequence is over the rectangular region defined by the slice:

# returns center, center-right, center-top, top-right images in that
# order:
images = texture_grid[(1,1):]
# equivalent to
images = texture_grid[(1,1):(3,3)]

get(row, column)

columns = 1

item_height = 0

item_width = 0

items = ()

rows = 1

class Texture3D(width, height, target, tex_id)

Bases: Texture, UniformTextureSequence

A texture with more than one image slice.

Use create_for_images or create_for_image_grid classmethod to construct.

classmethod create_for_image_grid(grid, internalformat=6408)

classmethod create_for_images(images, internalformat=6408, blank_data=True)

item_height = 0

item_width = 0

items = ()

Patterns

class ImagePattern

Abstract image creation class.

create_image(width, height)

Create an image of the given size.

Parameters

widthint: Width of image to create
heightint: Height of image to create

Return type

AbstractImage

class CheckerImagePattern(color1=(150, 150, 150, 255), color2=(200, 200, 200, 255))

Bases: ImagePattern

Create an image with a tileable checker image.

create_image(width, height)

Create an image of the given size.

Parameters

widthint: Width of image to create
heightint: Height of image to create

Return type

AbstractImage

class SolidColorImagePattern(color=(0, 0, 0, 0))

Bases: ImagePattern

Creates an image filled with a solid color.

create_image(width, height)

Create an image of the given size.

Parameters

widthint: Width of image to create
heightint: Height of image to create

Return type

AbstractImage

Data

class ImageData(width, height, fmt, data, pitch=None)

Bases: AbstractImage

An image represented as a string of unsigned bytes.

Parameters

datastr: Pixel data, encoded according to format and pitch.
formatstr: The format string to use when reading or writing data.
pitchint: Number of bytes per row. Negative values indicate a top-to-bottom arrangement.

blit(x, y, z=0, width=None, height=None)

Draw this image to the active framebuffers.

The image will be drawn with the lower-left corner at (x - anchor_x, y - anchor_y, z).

blit_to_texture(target, level, x, y, z, internalformat=None)

Draw this image to to the currently bound texture at target.

This image’s anchor point will be aligned to the given x and y coordinates. If the currently bound texture is a 3D texture, the z parameter gives the image slice to blit into.

If internalformat is specified, glTexImage is used to initialise the texture; otherwise, glTexSubImage is used to update a region.

create_texture(cls, rectangle=False)

Create a texture containing this image.

Parameters

clsclass (subclass of Texture): Class to construct.
rectanglebool: Unused. kept for compatibility.

New in version 1.1.

Return type

cls or cls.region_class

get_data(fmt=None, pitch=None)

Get the byte data of the image.

Parameters

fmtstr: Format string of the return data.
pitchint: Number of bytes per row. Negative values indicate a top-to-bottom arrangement.

New in version 1.1.

Return type: sequence of bytes, or str

get_image_data()

Get an ImageData view of this image.

Changes to the returned instance may or may not be reflected in this image.

Return type: ImageData

New in version 1.1.

get_mipmapped_texture()

Return a Texture with mipmaps.

If set_mipmap_Image has been called with at least one image, the set of images defined will be used. Otherwise, mipmaps will be automatically generated.

Return type: Texture

New in version 1.1.

get_region(x, y, width, height)

Retrieve a rectangular region of this image data.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

ImageDataRegion

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

set_data(fmt, pitch, data)

Set the byte data of the image.

Parameters

fmtstr: Format string of the return data.
pitchint: Number of bytes per row. Negative values indicate a top-to-bottom arrangement.
datastr or sequence of bytes: Image data.

New in version 1.1.

set_mipmap_image(level, image)

Set a mipmap image for a particular level.

The mipmap image will be applied to textures obtained via get_mipmapped_texture.

Parameters

levelint: Mipmap level to set image at, must be >= 1.
imageAbstractImage: Image to set. Must have correct dimensions for that mipmap level (i.e., width >> level, height >> level)

property format

Format string of the data. Read-write.

Type: str

class CompressedImageData(width, height, gl_format, data, extension=None, decoder=None)

Bases: AbstractImage

Image representing some compressed data suitable for direct uploading to driver.

blit_to_texture(target, level, x, y, z)

Draw this image on the currently bound texture at target.

This image is copied into the texture such that this image’s anchor point is aligned with the given x and y coordinates of the destination texture. If the currently bound texture is a 3D texture, the z coordinate gives the image slice to blit into.

get_mipmapped_texture()

Retrieve a Texture instance with all mipmap levels filled in.

Return type: Texture

New in version 1.1.

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

set_mipmap_data(level, data)

Set data for a mipmap level.

Supplied data gives a compressed image for the given mipmap level. The image must be of the correct dimensions for the level (i.e., width >> level, height >> level); but this is not checked. If any mipmap levels are specified, they are used; otherwise, mipmaps for mipmapped_texture are generated automatically.

Parameters

levelint: Level of mipmap image to set.
datasequence: String or array/list of bytes giving compressed image data. Data must be in same format as specified in constructor.

class ImageDataRegion(x, y, width, height, image_data)

Bases: ImageData

get_data(fmt=None, pitch=None)

Get the byte data of the image.

Parameters

fmtstr: Format string of the return data.
pitchint: Number of bytes per row. Negative values indicate a top-to-bottom arrangement.

New in version 1.1.

Return type: sequence of bytes, or str

get_region(x, y, width, height)

Retrieve a rectangular region of this image data.

Parameters

xint: Left edge of region.
yint: Bottom edge of region.
widthint: Width of region.
heightint: Height of region.

Return type

ImageDataRegion

set_data(fmt, pitch, data)

Set the byte data of the image.

Parameters

fmtstr: Format string of the return data.
pitchint: Number of bytes per row. Negative values indicate a top-to-bottom arrangement.
datastr or sequence of bytes: Image data.

New in version 1.1.

Other Classes

class BufferManager

Manages the set of framebuffers for a context.

Use get_buffer_manager() to obtain the instance of this class for the current context.

get_buffer_mask()

Get a free bitmask buffer.

A bitmask buffer is a buffer referencing a single bit in the stencil buffer. If no bits are free, ImageException is raised. Bits are released when the bitmask buffer is garbage collected.

Return type: BufferImageMask

get_color_buffer()

Get the color buffer.

Return type: ColorBufferImage

get_depth_buffer()

Get the depth buffer.

Return type: DepthBufferImage

static get_viewport()

Get the current OpenGL viewport dimensions.

Return type: 4-tuple of float.
Returns: Left, top, right and bottom dimensions.

class ImageGrid(image, rows, columns, item_width=None, item_height=None, row_padding=0, column_padding=0)

Bases: AbstractImage, AbstractImageSequence

An imaginary grid placed over an image allowing easy access to regular regions of that image.

The grid can be accessed either as a complete image, or as a sequence of images. The most useful applications are to access the grid as a TextureGrid:

image_grid = ImageGrid(...)
texture_grid = image_grid.get_texture_sequence()

or as a Texture3D:

image_grid = ImageGrid(...)
texture_3d = Texture3D.create_for_image_grid(image_grid)

get_image_data()

Get an ImageData view of this image.

Changes to the returned instance may or may not be reflected in this image.

Return type: ImageData

New in version 1.1.

get_texture(rectangle=False)

A Texture view of this image.

Parameters

rectanglebool: Unused. Kept for compatibility.

New in version 1.1.4..

Return type

Texture

New in version 1.1.

get_texture_sequence()

Get a TextureSequence.

Return type: TextureSequence

New in version 1.1.

Functions

create(width, height, pattern=None)

Create an image optionally filled with the given pattern.

Parameters

widthint: Width of image to create
heightint: Height of image to create
patternImagePattern or None: Pattern to fill image with. If unspecified, the image will initially be transparent.

Return type

AbstractImage

Note

You can make no assumptions about the return type; usually it will be ImageData or CompressedImageData, but patterns are free to return any subclass of AbstractImage.

get_buffer_manager()

Get the buffer manager for the current OpenGL context.

Return type: BufferManager

load(filename, file=None, decoder=None)

Load an image from a file.

Parameters

filenamestr: Used to guess the image format, and to load the file if file is unspecified.
filefile-like object or None: Source of image data in any supported format.
decoderImageDecoder or None: If unspecified, all decoders that are registered for the filename extension are tried. If none succeed, the exception from the first decoder is raised.

Return type

AbstractImage

Note

You can make no assumptions about the return type; usually it will be ImageData or CompressedImageData, but decoders are free to return any subclass of AbstractImage.

load_animation(filename, file=None, decoder=None)

Load an animation from a file.

Currently, the only supported format is GIF.

Parameters

filenamestr: Used to guess the animation format, and to load the file if file is unspecified.
filefile-like object or None: File object containing the animation stream.
decoderImageDecoder or None: If unspecified, all decoders that are registered for the filename extension are tried. If none succeed, the exception from the first decoder is raised.

Return type

Animation

get_max_texture_size(): Query the maximum texture size available

Exceptions

class ImageException

class ImageEncodeException

class ImageDecodeException