Examples¶

fromPILimportImagewithImage.open("hopper.jpg")asim:im.rotate(45).show()

fromPILimportImageimportglob,ossize=128,128forinfileinglob.glob("*.jpg"):file,ext=os.path.splitext(infile)withImage.open(infile)asim:im.thumbnail(size)im.save(file+".thumbnail","JPEG")

Functions¶

PIL.Image.open(fp:StrOrBytesPath|IO[bytes],mode:Literal['r']='r',formats:list[str]|tuple[str,...]|None=None)→ImageFile.ImageFile[source]¶

Opens and identifies the given image file.

This is a lazy operation; this function identifies the file, butthe file remains open and the actual image data is not read fromthe file until you try to process the data (or call theload() method). Seenew(). SeeFile handling in Pillow.

Parameters:

• fp – A filename (string), os.PathLike object or a file object.The file object must implementfile.read,file.seek, andfile.tell methods,and be opened in binary mode. The file object will also seek to zerobefore reading.

• mode – The mode. If given, this argument must be “r”.

• formats – A list or tuple of formats to attempt to load the file in.This can be used to restrict the set of formats checked.PassNone to try all supported formats. You can print the set ofavailable formats by runningpython3-mPIL or usingthePIL.features.pilinfo() function.

Returns:

AnImage object.

Raises:

• FileNotFoundError – If the file cannot be found.

• PIL.UnidentifiedImageError – If the image cannot be opened andidentified.

• ValueError – If themode is not “r”, or if aStringIOinstance is used forfp.

• TypeError – Ifformats is notNone, a list or a tuple.

Warning

To protect against potential DOS attacks caused by “decompression bombs” (i.e. malicious fileswhich decompress into a huge amount of data and are designed to crash or cause disruption by using upa lot of memory), Pillow will issue aDecompressionBombWarning if the number of pixels in animage is over a certain limit,MAX_IMAGE_PIXELS.

This threshold can be changed by settingMAX_IMAGE_PIXELS. It can be disabledby settingImage.MAX_IMAGE_PIXELS=None.

If desired, the warning can be turned into an error withwarnings.simplefilter('error',Image.DecompressionBombWarning) or suppressed entirely withwarnings.simplefilter('ignore',Image.DecompressionBombWarning). See alsothe logging documentation to have warnings output to the logging facility instead of stderr.

If the number of pixels is greater than twiceMAX_IMAGE_PIXELS, then aDecompressionBombError will be raised instead.

The Image class¶

classPIL.Image.Image[source]¶

This class represents an image object. To createImage objects, use the appropriate factoryfunctions. There’s hardly ever any reason to call the Image constructordirectly.

• open()

• new()

• frombytes()

An instance of theImage class has the followingmethods. Unless otherwise stated, all methods return a new instance of theImage class, holding the resulting image.

Image.alpha_composite(im:Image,dest:Sequence[int]=(0,0),source:Sequence[int]=(0,0))→None[source]¶

‘In-place’ analog of Image.alpha_composite. Composites an imageonto this image.

Parameters:

• im – image to composite over this one

• dest – Optional 2 tuple (left, top) specifying the upperleft corner in this (destination) image.

• source – Optional 2 (left, top) tuple for the upper leftcorner in the overlay source image, or 4 tuple (left, top, right,bottom) for the bounds of the source rectangle

Performance Note: Not currently implemented in-place in the core layer.

Image.apply_transparency()→None[source]¶

If a P mode image has a “transparency” key in the info dictionary,remove the key and instead apply the transparency to the palette.Otherwise, the image is unchanged.

Image.convert(mode:str|None=None,matrix:tuple[float,...]|None=None,dither:Dither|None=None,palette:Palette=Palette.WEB,colors:int=256)→Image[source]¶

Returns a converted copy of this image. For the “P” mode, thismethod translates pixels through the palette. If mode isomitted, a mode is chosen so that all information in the imageand the palette can be represented without a palette.

This supports all possible conversions between “L”, “RGB” and “CMYK”. Thematrix argument only supports “L” and “RGB”.

When translating a color image to grayscale (mode “L”),the library uses the ITU-R 601-2 luma transform:

L=R*299/1000+G*587/1000+B*114/1000

The default method of converting a grayscale (“L”) or “RGB”image into a bilevel (mode “1”) image uses Floyd-Steinbergdither to approximate the original image luminosity levels. Ifdither isNone, all values larger than 127 are set to 255 (white),all other values to 0 (black). To use other thresholds, use thepoint() method.

When converting from “RGBA” to “P” without amatrix argument,this passes the operation toquantize(),anddither andpalette are ignored.

When converting from “PA”, if an “RGBA” palette is present, the alphachannel from the image will be used instead of the values from the palette.

Parameters:

• mode – The requested mode. See:Modes.

• matrix – An optional conversion matrix. If given, thisshould be 4- or 12-tuple containing floating point values.

• dither – Dithering method, used when converting frommode “RGB” to “P” or from “RGB” or “L” to “1”.Available methods areDither.NONE orDither.FLOYDSTEINBERG(default). Note that this is not used whenmatrix is supplied.

• palette – Palette to use when converting from mode “RGB”to “P”. Available palettes arePalette.WEB orPalette.ADAPTIVE.

• colors – Number of colors to use for thePalette.ADAPTIVEpalette. Defaults to 256.

Return type:

Image

Returns:

AnImage object.

The following example converts an RGB image (linearly calibrated according toITU-R 709, using the D65 luminant) to the CIE XYZ color space:

rgb2xyz=(0.412453,0.357580,0.180423,0,0.212671,0.715160,0.072169,0,0.019334,0.119193,0.950227,0)out=im.convert("RGB",rgb2xyz)

Image.copy()→Image[source]¶

Copies this image. Use this method if you wish to paste thingsinto an image, but still retain the original.

Return type:

Image

Returns:

AnImage object.

Image.crop(box:tuple[float,float,float,float]|None=None)→Image[source]¶

Returns a rectangular region from this image. The box is a4-tuple defining the left, upper, right, and lower pixelcoordinate. SeeCoordinate system.

Note: Prior to Pillow 3.4.0, this was a lazy operation.

Parameters:

box – The crop rectangle, as a (left, upper, right, lower)-tuple.

Return type:

Image

Returns:

AnImage object.

This crops the input image with the provided coordinates:

fromPILimportImagewithImage.open("hopper.jpg")asim:# The crop method from the Image module takes four coordinates as input.# The right can also be represented as (left+width)# and lower can be represented as (upper+height).(left,upper,right,lower)=(20,20,100,100)# Here the image "im" is cropped and assigned to new variable im_cropim_crop=im.crop((left,upper,right,lower))

Image.draft(mode:str|None,size:tuple[int,int]|None)→tuple[str,tuple[int,int,float,float]]|None[source]¶

Configures the image file loader so it returns a version of theimage that as closely as possible matches the given mode andsize. For example, you can use this method to convert a colorJPEG to grayscale while loading it.

If any changes are made, returns a tuple with the chosenmode andbox with coordinates of the original image within the altered one.

Note that this method modifies theImage objectin place. If the image has already been loaded, this method has noeffect.

Note: This method is not implemented for most images. It iscurrently implemented only for JPEG and MPO images.

Parameters:

• mode – The requested mode.

• size – The requested size in pixels, as a 2-tuple:(width, height).

Image.effect_spread(distance:int)→Image[source]¶

Randomly spread pixels in an image.

Parameters:

distance – Distance to spread pixels.

Image.entropy(mask:Image|None=None,extrema:tuple[float,float]|None=None)→float[source]¶

Calculates and returns the entropy for the image.

A bilevel image (mode “1”) is treated as a grayscale (“L”)image by this method.

If a mask is provided, the method employs the histogram forthose parts of the image where the mask image is non-zero.The mask image must have the same size as the image, and beeither a bi-level image (mode “1”) or a grayscale image (“L”).

Parameters:

• mask – An optional mask.

• extrema – An optional tuple of manually-specified extrema.

Returns:

A float value representing the image entropy

Image.filter(filter:ImageFilter.Filter|type[ImageFilter.Filter])→Image[source]¶

Filters this image using the given filter. For a list ofavailable filters, see theImageFilter module.

Parameters:

filter – Filter kernel.

Returns:

AnImage object.

This blurs the input image using a filter from theImageFilter module:

fromPILimportImage,ImageFilterwithImage.open("hopper.jpg")asim:# Blur the input image using the filter ImageFilter.BLURim_blurred=im.filter(filter=ImageFilter.BLUR)

Image.frombytes(data:bytes|bytearray|SupportsArrayInterface,decoder_name:str='raw',*args:Any)→None[source]¶

Loads this image with pixel data from a bytes object.

This method is similar to thefrombytes() function,but loads data into this image instead of creating a new image object.

Image.getbands()→tuple[str,...][source]¶

Returns a tuple containing the name of each band in this image.For example,getbands on an RGB image returns (“R”, “G”, “B”).

Returns:

A tuple containing band names.

Return type:

tuple

This helps to get the bands of the input image:

fromPILimportImagewithImage.open("hopper.jpg")asim:print(im.getbands())# Returns ('R', 'G', 'B')

Image.getbbox(*,alpha_only:bool=True)→tuple[int,int,int,int]|None[source]¶

Calculates the bounding box of the non-zero regions in theimage.

Parameters:

alpha_only – Optional flag, defaulting toTrue.IfTrue and the image has an alpha channel, trim transparent pixels.Otherwise, trim pixels when all channels are zero.Keyword-only argument.

Returns:

The bounding box is returned as a 4-tuple defining theleft, upper, right, and lower pixel coordinate. SeeCoordinate system. If the image is completely empty, thismethod returns None.

This helps to get the bounding box coordinates of the input image:

fromPILimportImagewithImage.open("hopper.jpg")asim:print(im.getbbox())# Returns four coordinates in the format (left, upper, right, lower)

Image.getchannel(channel:int|str)→Image[source]¶

Returns an image containing a single channel of the source image.

Parameters:

channel – What channel to return. Could be index(0 for “R” channel of “RGB”) or channel name(“A” for alpha channel of “RGBA”).

Returns:

An image in “L” mode.

Added in version 4.3.0.

Image.getcolors(maxcolors:int=256)→list[tuple[int,tuple[int,...]]]|list[tuple[int,float]]|None[source]¶

Returns a list of colors used in this image.

The colors will be in the image’s mode. For example, an RGB image willreturn a tuple of (red, green, blue) color values, and a P image willreturn the index of the color in the palette.

Parameters:

maxcolors – Maximum number of colors. If this number isexceeded, this method returns None. The default limit is256 colors.

Returns:

An unsorted list of (count, pixel) values.

Image.getdata(band:int|None=None)→core.ImagingCore[source]¶

Returns the contents of this image as a sequence objectcontaining pixel values. The sequence object is flattened, sothat values for line one follow directly after the values ofline zero, and so on.

Note that the sequence object returned by this method is aninternal PIL data type, which only supports certain sequenceoperations. To convert it to an ordinary sequence (e.g. forprinting), uselist(im.getdata()).

Parameters:

band – What band to return. The default is to returnall bands. To return a single band, pass in the indexvalue (e.g. 0 to get the “R” band from an “RGB” image).

Returns:

A sequence-like object.

Image.getexif()→Exif[source]¶

Gets EXIF data from the image.

Returns:

anExif object.

Image.getextrema()→tuple[float,float]|tuple[tuple[int,int],...][source]¶

Gets the minimum and maximum pixel values for each band inthe image.

Returns:

For a single-band image, a 2-tuple containing theminimum and maximum pixel value. For a multi-band image,a tuple containing one 2-tuple for each band.

Image.getpalette(rawmode:str|None='RGB')→list[int]|None[source]¶

Returns the image palette as a list.

Parameters:

rawmode –

The mode in which to return the palette.None willreturn the palette in its current mode.

Added in version 9.1.0.

Returns:

A list of color values [r, g, b, …], or None if theimage has no palette.

Image.getpixel(xy:tuple[int,int]|list[int])→float|tuple[int,...]|None[source]¶

Returns the pixel value at a given position.

Parameters:

xy – The coordinate, given as (x, y). SeeCoordinate system.

Returns:

The pixel value. If the image is a multi-layer image,this method returns a tuple.

Image.getprojection()→tuple[list[int],list[int]][source]¶

Get projection to x and y axes

Returns:

Two sequences, indicating where there are non-zeropixels along the X-axis and the Y-axis, respectively.

Image.getxmp()→dict[str,Any][source]¶

Returns a dictionary containing the XMP tags.Requires defusedxml to be installed.

Returns:

XMP tags in a dictionary.

Image.histogram(mask:Image|None=None,extrema:tuple[float,float]|None=None)→list[int][source]¶

Returns a histogram for the image. The histogram is returned as alist of pixel counts, one for each pixel value in the sourceimage. Counts are grouped into 256 bins for each band, even ifthe image has more than 8 bits per band. If the image has morethan one band, the histograms for all bands are concatenated (forexample, the histogram for an “RGB” image contains 768 values).

A bilevel image (mode “1”) is treated as a grayscale (“L”) imageby this method.

If a mask is provided, the method returns a histogram for thoseparts of the image where the mask image is non-zero. The maskimage must have the same size as the image, and be either abi-level image (mode “1”) or a grayscale image (“L”).

Parameters:

• mask – An optional mask.

• extrema – An optional tuple of manually-specified extrema.

Returns:

A list containing pixel counts.

Image.paste(im:Image|str|float|tuple[float,...],box:Image|tuple[int,int,int,int]|tuple[int,int]|None=None,mask:Image|None=None)→None[source]¶

Pastes another image into this image. The box argument is eithera 2-tuple giving the upper left corner, a 4-tuple defining theleft, upper, right, and lower pixel coordinate, or None (same as(0, 0)). SeeCoordinate system. If a 4-tuple is given, the sizeof the pasted image must match the size of the region.

If the modes don’t match, the pasted image is converted to the mode ofthis image (see theconvert() method fordetails).

Instead of an image, the source can be a integer or tuplecontaining pixel values. The method then fills the regionwith the given color. When creating RGB images, you canalso use color strings as supported by the ImageColor module. SeeColors for more information.

If a mask is given, this method updates only the regionsindicated by the mask. You can use either “1”, “L”, “LA”, “RGBA”or “RGBa” images (if present, the alpha band is used as mask).Where the mask is 255, the given image is copied as is. Wherethe mask is 0, the current value is preserved. Intermediatevalues will mix the two images together, including their alphachannels if they have them.

Seealpha_composite() if you want tocombine images with respect to their alpha channels.

Parameters:

• im – Source image or pixel value (integer, float or tuple).

• box –

  An optional 4-tuple giving the region to paste into.If a 2-tuple is used instead, it’s treated as the upper leftcorner. If omitted or None, the source is pasted into theupper left corner.

  If an image is given as the second argument and there is nothird, the box defaults to (0, 0), and the second argumentis interpreted as a mask image.

• mask – An optional mask image.

Image.point(lut:Sequence[float]|NumpyArray|Callable[[int],float]|Callable[[ImagePointTransform],ImagePointTransform|float]|ImagePointHandler,mode:str|None=None)→Image[source]¶

Maps this image through a lookup table or function.

Parameters:

• lut –

  A lookup table, containing 256 (or 65536 ifself.mode==”I” and mode == “L”) values per band in theimage. A function can be used instead, it should take asingle argument. The function is called once for eachpossible pixel value, and the resulting table is applied toall bands of the image.

  It may also be anImagePointHandlerobject:

  classExample(Image.ImagePointHandler):defpoint(self,im:Image)->Image:# Return result

• mode – Output mode (default is same as input). This can only be used ifthe source image has mode “L” or “P”, and the output has mode “1” or thesource image mode is “I” and the output mode is “L”.

Returns:

AnImage object.

Image.putalpha(alpha:Image|int)→None[source]¶

Adds or replaces the alpha layer in this image. If the imagedoes not have an alpha layer, it’s converted to “LA” or “RGBA”.The new layer must be either “L” or “1”.

Parameters:

alpha – The new alpha layer. This can either be an “L” or “1”image having the same size as this image, or an integer.

Image.putdata(data:Sequence[float]|Sequence[Sequence[int]]|core.ImagingCore|NumpyArray,scale:float=1.0,offset:float=0.0)→None[source]¶

Copies pixel data from a flattened sequence object into the image. Thevalues should start at the upper left corner (0, 0), continue to theend of the line, followed directly by the first value of the secondline, and so on. Data will be read until either the image or thesequence ends. The scale and offset values are used to adjust thesequence values:pixel = value*scale + offset.

Parameters:

• data – A flattened sequence object. SeeColors for moreinformation about values.

• scale – An optional scale value. The default is 1.0.

• offset – An optional offset value. The default is 0.0.

Image.putpalette(data:ImagePalette.ImagePalette|bytes|Sequence[int],rawmode:str='RGB')→None[source]¶

Attaches a palette to this image. The image must be a “P”, “PA”, “L”or “LA” image.

The palette sequence must contain at most 256 colors, made up of oneinteger value for each channel in the raw mode.For example, if the raw mode is “RGB”, then it can contain at most 768values, made up of red, green and blue values for the corresponding pixelindex in the 256 colors.If the raw mode is “RGBA”, then it can contain at most 1024 values,containing red, green, blue and alpha values.

Alternatively, an 8-bit string may be used instead of an integer sequence.

Parameters:

• data – A palette sequence (either a list or a string).

• rawmode – The raw mode of the palette. Either “RGB”, “RGBA”, or a modethat can be transformed to “RGB” or “RGBA” (e.g. “R”, “BGR;15”, “RGBA;L”).

Image.putpixel(xy:tuple[int,int],value:float|tuple[int,...]|list[int])→None[source]¶

Modifies the pixel at the given position. The color is given asa single numerical value for single-band images, and a tuple formulti-band images. In addition to this, RGB and RGBA tuples areaccepted for P and PA images. SeeColors for more information.

Note that this method is relatively slow. For more extensive changes,usepaste() or theImageDrawmodule instead.

See:

• paste()

• putdata()

• ImageDraw

Parameters:

• xy – The pixel coordinate, given as (x, y). SeeCoordinate system.

• value – The pixel value.

Image.quantize(colors:int=256,method:int|None=None,kmeans:int=0,palette:Image|None=None,dither:Dither=Dither.FLOYDSTEINBERG)→Image[source]¶

Convert the image to ‘P’ mode with the specified numberof colors.

Parameters:

• colors – The desired number of colors, <= 256

• method –

  Quantize.MEDIANCUT (median cut),Quantize.MAXCOVERAGE (maximum coverage),Quantize.FASTOCTREE (fast octree),Quantize.LIBIMAGEQUANT (libimagequant; check supportusingPIL.features.check_feature() withfeature="libimagequant").

  By default,Quantize.MEDIANCUT will be used.

  The exception to this is RGBA images.Quantize.MEDIANCUTandQuantize.MAXCOVERAGE do not support RGBA images, soQuantize.FASTOCTREE is used by default instead.

• kmeans – Integer greater than or equal to zero.

• palette – Quantize to the palette of givenPIL.Image.Image.

• dither – Dithering method, used when converting frommode “RGB” to “P” or from “RGB” or “L” to “1”.Available methods areDither.NONE orDither.FLOYDSTEINBERG(default).

Returns:

A new image

Image.reduce(factor:int|tuple[int,int],box:tuple[int,int,int,int]|None=None)→Image[source]¶

Returns a copy of the image reducedfactor times.If the size of the image is not dividable byfactor,the resulting size will be rounded up.

Parameters:

• factor – A greater than 0 integer or tuple of two integersfor width and height separately.

• box – An optional 4-tuple of ints providingthe source image region to be reduced.The values must be within(0,0,width,height) rectangle.If omitted orNone, the entire source is used.

Image.remap_palette(dest_map:list[int],source_palette:bytes|bytearray|None=None)→Image[source]¶

Rewrites the image to reorder the palette.

Parameters:

• dest_map – A list of indexes into the original palette.e.g.[1,0] would swap a two item palette, andlist(range(256))is the identity transform.

• source_palette – Bytes or None.

Returns:

AnImage object.

Image.resize(size:tuple[int,int]|list[int]|NumpyArray,resample:int|None=None,box:tuple[float,float,float,float]|None=None,reducing_gap:float|None=None)→Image[source]¶

Returns a resized copy of this image.

Parameters:

• size – The requested size in pixels, as a tuple or array:(width, height).

• resample – An optional resampling filter. This can beone ofResampling.NEAREST,Resampling.BOX,Resampling.BILINEAR,Resampling.HAMMING,Resampling.BICUBIC orResampling.LANCZOS.If the image has mode “1” or “P”, it is always set toResampling.NEAREST. Otherwise, the default filter isResampling.BICUBIC. See:Filters.

• box – An optional 4-tuple of floats providingthe source image region to be scaled.The values must be within (0, 0, width, height) rectangle.If omitted or None, the entire source is used.

• reducing_gap – Apply optimization by resizing the imagein two steps. First, reducing the image by integer timesusingreduce().Second, resizing using regular resampling. The last stepchanges size no less than byreducing_gap times.reducing_gap may be None (no first step is performed)or should be greater than 1.0. The biggerreducing_gap,the closer the result to the fair resampling.The smallerreducing_gap, the faster resizing.Withreducing_gap greater or equal to 3.0, the result isindistinguishable from fair resampling in most cases.The default value is None (no optimization).

Returns:

AnImage object.

This resizes the given image from(width,height) to(width/2,height/2):

fromPILimportImagewithImage.open("hopper.jpg")asim:# Provide the target width and height of the image(width,height)=(im.width//2,im.height//2)im_resized=im.resize((width,height))

Image.rotate(angle:float,resample:Resampling=Resampling.NEAREST,expand:int|bool=False,center:tuple[float,float]|None=None,translate:tuple[int,int]|None=None,fillcolor:float|tuple[float,...]|str|None=None)→Image[source]¶

Returns a rotated copy of this image. This method returns acopy of this image, rotated the given number of degrees counterclockwise around its centre.

Parameters:

• angle – In degrees counter clockwise.

• resample – An optional resampling filter. This can beone ofResampling.NEAREST (use nearest neighbour),Resampling.BILINEAR (linear interpolation in a 2x2environment), orResampling.BICUBIC (cubic splineinterpolation in a 4x4 environment). If omitted, or if the image hasmode “1” or “P”, it is set toResampling.NEAREST.SeeFilters.

• expand – Optional expansion flag. If true, expands the outputimage to make it large enough to hold the entire rotated image.If false or omitted, make the output image the same size as theinput image. Note that the expand flag assumes rotation aroundthe center and no translation.

• center – Optional center of rotation (a 2-tuple). Origin isthe upper left corner. Default is the center of the image.

• translate – An optional post-rotate translation (a 2-tuple).

• fillcolor – An optional color for area outside the rotated image.

Returns:

AnImage object.

This rotates the input image bytheta degrees counter clockwise:

fromPILimportImagewithImage.open("hopper.jpg")asim:# Rotate the image by 60 degrees counter clockwisetheta=60# Angle is in degrees counter clockwiseim_rotated=im.rotate(angle=theta)

Image.save(fp:StrOrBytesPath|IO[bytes],format:str|None=None,**params:Any)→None[source]¶

Saves this image under the given filename. If no format isspecified, the format to use is determined from the filenameextension, if possible.

Keyword options can be used to provide additional instructionsto the writer. If a writer doesn’t recognise an option, it issilently ignored. The available options are described in theimage format documentation for each writer.

You can use a file object instead of a filename. In this case,you must always specify the format. The file object mustimplement theseek,tell, andwritemethods, and be opened in binary mode.

Parameters:

• fp – A filename (string), os.PathLike object or file object.

• format – Optional format override. If omitted, theformat to use is determined from the filename extension.If a file object was used instead of a filename, thisparameter should always be used.

• params –

  Extra parameters to the image writer. These can also beset on the image itself throughencoderinfo. This is useful whensaving multiple images:

  # Saving XMP data to a single imagefromPILimportImagered=Image.new("RGB",(1,1),"#f00")red.save("out.mpo",xmp=b"test")# Saving XMP data to the second frame of an imagefromPILimportImageblack=Image.new("RGB",(1,1))red=Image.new("RGB",(1,1),"#f00")red.encoderinfo={"xmp":b"test"}black.save("out.mpo",save_all=True,append_images=[red])

Returns:

None

Raises:

• ValueError – If the output format could not be determinedfrom the file name. Use the format option to solve this.

• OSError – If the file could not be written. The filemay have been created, and may contain partial data.

Image.seek(frame:int)→None[source]¶

Seeks to the given frame in this sequence file. If you seekbeyond the end of the sequence, the method raises anEOFError exception. When a sequence file is opened, thelibrary automatically seeks to frame 0.

Seetell().

If defined,n_frames refers to thenumber of available frames.

Parameters:

frame – Frame number, starting at 0.

Raises:

EOFError – If the call attempts to seek beyond the endof the sequence.

Image.show(title:str|None=None)→None[source]¶

Displays this image. This method is mainly intended for debugging purposes.

This method callsPIL.ImageShow.show() internally. You can usePIL.ImageShow.register() to override its default behaviour.

The image is first saved to a temporary file. By default, it will be inPNG format.

On Unix, the image is then opened using thexdg-open,display,gm,eog orxv utility, depending on which one can be found.

On macOS, the image is opened with the native Preview application.

On Windows, the image is opened with the standard PNG display utility.

Parameters:

title – Optional title to use for the image window, where possible.

Image.split()→tuple[Image,...][source]¶

Split this image into individual bands. This method returns atuple of individual image bands from an image. For example,splitting an “RGB” image creates three new images eachcontaining a copy of one of the original bands (red, green,blue).

If you need only one band,getchannel()method can be more convenient and faster.

Returns:

A tuple containing bands.

Image.tell()→int[source]¶

Returns the current frame number. Seeseek().

If defined,n_frames refers to thenumber of available frames.

Returns:

Frame number, starting with 0.

Image.thumbnail(size:tuple[float,float],resample:Resampling=Resampling.BICUBIC,reducing_gap:float|None=2.0)→None[source]¶

Make this image into a thumbnail. This method modifies theimage to contain a thumbnail version of itself, no larger thanthe given size. This method calculates an appropriate thumbnailsize to preserve the aspect of the image, calls thedraft() method to configure the file reader(where applicable), and finally resizes the image.

Note that this function modifies theImageobject in place. If you need to use the full resolution image as well,apply this method to acopy() of the originalimage.

Parameters:

• size – The requested size in pixels, as a 2-tuple:(width, height).

• resample – Optional resampling filter. This can be oneofResampling.NEAREST,Resampling.BOX,Resampling.BILINEAR,Resampling.HAMMING,Resampling.BICUBIC orResampling.LANCZOS.If omitted, it defaults toResampling.BICUBIC.(wasResampling.NEAREST prior to version 2.5.0).See:Filters.

• reducing_gap – Apply optimization by resizing the imagein two steps. First, reducing the image by integer timesusingreduce() ordraft() for JPEG images.Second, resizing using regular resampling. The last stepchanges size no less than byreducing_gap times.reducing_gap may be None (no first step is performed)or should be greater than 1.0. The biggerreducing_gap,the closer the result to the fair resampling.The smallerreducing_gap, the faster resizing.Withreducing_gap greater or equal to 3.0, the result isindistinguishable from fair resampling in most cases.The default value is 2.0 (very close to fair resamplingwhile still being faster in many cases).

Returns:

None

Image.tobitmap(name:str='image')→bytes[source]¶

Returns the image converted to an X11 bitmap.

Note

This method only works for mode “1” images.

Parameters:

name – The name prefix to use for the bitmap variables.

Returns:

A string containing an X11 bitmap.

Raises:

ValueError – If the mode is not “1”

Image.tobytes(encoder_name:str='raw',*args:Any)→bytes[source]¶

Return image as a bytes object.

Warning

This method returns raw image data derived from Pillow’s internalstorage. For compressed image data (e.g. PNG, JPEG) usesave(), with a BytesIO parameter for in-memory data.

Parameters:

• encoder_name –

  What encoder to use.

  The default is to use the standard “raw” encoder.To see how this packs pixel data into the returnedbytes, seelibImaging/Pack.c.

  A list of C encoders can be seen under codecssection of the function array in_imaging.c. Python encoders are registeredwithin the relevant plugins.

• args – Extra arguments to the encoder.

Returns:

Abytes object.

Image.transform(size:tuple[int,int],method:Transform|ImageTransformHandler|SupportsGetData,data:Sequence[Any]|None=None,resample:int=Resampling.NEAREST,fill:int=1,fillcolor:float|tuple[float,...]|str|None=None)→Image[source]¶

Transforms this image. This method creates a new image with thegiven size, and the same mode as the original, and copies datato the new image using the given transform.

Parameters:

• size – The output size in pixels, as a 2-tuple:(width, height).

• method –

  The transformation method. This is one ofTransform.EXTENT (cut out a rectangular subregion),Transform.AFFINE (affine transform),Transform.PERSPECTIVE (perspective transform),Transform.QUAD (map a quadrilateral to a rectangle), orTransform.MESH (map a number of source quadrilateralsin one operation).

  It may also be anImageTransformHandlerobject:

  classExample(Image.ImageTransformHandler):deftransform(self,size,data,resample,fill=1):# Return result

  Implementations ofImageTransformHandlerfor some of theTransform methods are providedinImageTransform.

  It may also be an object with amethod.getdata methodthat returns a tuple supplying newmethod anddata values:

  classExample:defgetdata(self):method=Image.Transform.EXTENTdata=(0,0,100,100)returnmethod,data

• data – Extra data to the transformation method.

• resample – Optional resampling filter. It can be one ofResampling.NEAREST (use nearest neighbour),Resampling.BILINEAR (linear interpolation in a 2x2environment), orResampling.BICUBIC (cubic splineinterpolation in a 4x4 environment). If omitted, or if the imagehas mode “1” or “P”, it is set toResampling.NEAREST.See:Filters.

• fill – Ifmethod is anImageTransformHandler object, this is one ofthe arguments passed to it. Otherwise, it is unused.

• fillcolor – Optional fill color for the area outside thetransform in the output image.

Returns:

AnImage object.

Image.transpose(method:Transpose)→Image[source]¶

Transpose image (flip or rotate in 90 degree steps)

Parameters:

method – One ofTranspose.FLIP_LEFT_RIGHT,Transpose.FLIP_TOP_BOTTOM,Transpose.ROTATE_90,Transpose.ROTATE_180,Transpose.ROTATE_270,Transpose.TRANSPOSE orTranspose.TRANSVERSE.

Returns:

Returns a flipped or rotated copy of this image.

This flips the input image by using theTranspose.FLIP_LEFT_RIGHTmethod.

fromPILimportImagewithImage.open("hopper.jpg")asim:# Flip the image from left to rightim_flipped=im.transpose(method=Image.Transpose.FLIP_LEFT_RIGHT)# To flip the image from top to bottom,# use the method "Image.Transpose.FLIP_TOP_BOTTOM"

Image.verify()→None[source]¶

Verifies the contents of a file. For data read from a file, thismethod attempts to determine if the file is broken, withoutactually decoding the image data. If this method finds anyproblems, it raises suitable exceptions. If you need to loadthe image after using this method, you must reopen the imagefile.

Image.load()→core.PixelAccess|None[source]¶

Allocates storage for the image and loads the pixel data. Innormal cases, you don’t need to call this method, since theImage class automatically loads an opened image when it isaccessed for the first time.

If the file associated with the image was opened by Pillow, then thismethod will close it. The exception to this is if the image hasmultiple frames, in which case the file will be left open for seekoperations. SeeFile handling in Pillow for more information.

Returns:

An image access object.

Return type:

PixelAccess

Image.close()→None[source]¶

This operation will destroy the image core and release its memory.The image data will be unusable afterward.

This function is required to close images that have multiple frames orhave not had their file read and closed by theload() method. SeeFile handling in Pillow formore information.

Image attributes¶

Instances of theImage class have the following attributes:

Image.filename:str¶

The filename or path of the source file. Only images created with thefactory functionopen have a filename attribute. If the input is afile like object, the filename attribute is set to an empty string.

Image.format:str|None¶

The file format of the source file. For images created by the libraryitself (via a factory function, or by running a method on an existingimage), this attribute is set toNone.

Image.mode:str¶

Image mode. This is a string specifying the pixel format used by the image.Typical values are “1”, “L”, “RGB”, or “CMYK.” SeeModes for a full list.

Image.size:tuple[int]¶

Image size, in pixels. The size is given as a 2-tuple (width, height).

Image.width:int¶

Image width, in pixels.

Image.height:int¶

Image height, in pixels.

Image.palette:PIL.ImagePalette.ImagePalette|None¶

Colour palette table, if any. If mode is “P” or “PA”, this should be aninstance of theImagePalette class.Otherwise, it should be set toNone.

Image.info:dict¶

A dictionary holding data associated with the image. This dictionary isused by file handlers to pass on various non-image information read fromthe file. See documentation for the various file handlers for details.

Most methods ignore the dictionary when returning new images; since thekeys are not standardized, it’s not possible for a method to know if theoperation affects the dictionary. If you need the information later on,keep a reference to the info dictionary returned from the open method.

Unless noted elsewhere, this dictionary does not affect saving files.

Image.is_animated:bool¶

True if this image has more than one frame, orFalse otherwise.

This attribute is only defined by image plugins that support animated images.Plugins may leave this attribute undefined if they don’t support loadinganimated images, even if the given format supports animated images.

Given that this attribute is not present for all images usegetattr(image,"is_animated",False) to check if Pillow is aware of multipleframes in an image regardless of its format.

See also

n_frames,seek() andtell()

Image.n_frames:int¶

The number of frames in this image.

This attribute is only defined by image plugins that support animated images.Plugins may leave this attribute undefined if they don’t support loadinganimated images, even if the given format supports animated images.

Given that this attribute is not present for all images usegetattr(image,"n_frames",1) to check the number of frames that Pillow isaware of in an image regardless of its format.

See also

is_animated,seek() andtell()

Image.has_transparency_data¶

Determine if an image has transparency data, whether in the form of analpha channel, a palette with an alpha channel, or a “transparency” keyin the info dictionary.

Note the image might still appear solid, if all of the values shownwithin are opaque.

Returns:

A boolean.

Classes¶

classPIL.Image.Exif[source]¶

Bases:MutableMapping

This class provides read and write access to EXIF image data:

fromPILimportImageim=Image.open("exif.png")exif=im.getexif()# Returns an instance of this class

Information can be read and written, iterated over or deleted:

print(exif[274])# 1exif[274]=2fork,vinexif.items():print("Tag",k,"Value",v)# Tag 274 Value 2delexif[274]

To access information beyond IFD0,get_ifd()returns a dictionary:

fromPILimportExifTagsim=Image.open("exif_gps.jpg")exif=im.getexif()gps_ifd=exif.get_ifd(ExifTags.IFD.GPSInfo)print(gps_ifd)

Other IFDs includeExifTags.IFD.Exif,ExifTags.IFD.MakerNote,ExifTags.IFD.Interop andExifTags.IFD.IFD1.

ExifTags also has enum classes to provide names for data:

print(exif[ExifTags.Base.Software])# PILprint(gps_ifd[ExifTags.GPS.GPSDateStamp])# 1999:99:99 99:99:99

bigtiff=False¶

endian:str|None=None¶

get_ifd(tag:int)→dict[int,Any][source]¶

hide_offsets()→None[source]¶

load(data:bytes)→None[source]¶

load_from_fp(fp:IO[bytes],offset:int|None=None)→None[source]¶

tobytes(offset:int=8)→bytes[source]¶

classPIL.Image.ImagePointHandler[source]¶

Used as a mixin by point transforms(for use withpoint())

classPIL.Image.ImagePointTransform(scale:float,offset:float)[source]¶

Used withpoint() for single band images with more than8 bits, this represents an affine transformation, where the value is multiplied byscale andoffset is added.

classPIL.Image.ImageTransformHandler[source]¶

Used as a mixin by geometry transforms(for use withtransform())