Concepts

The Python Imaging Library handles raster images; that is, rectangles ofpixel data.

Bands

An image can consist of one or more bands of data. The Python Imaging Libraryallows you to store several bands in a single image, provided they all have thesame dimensions and depth. For example, a PNG image might have ‘R’, ‘G’, ‘B’,and ‘A’ bands for the red, green, blue, and alpha transparency values. Manyoperations act on each band separately, e.g., histograms. It is often useful tothink of each pixel as having one value per band.

To get the number and names of bands in an image, use thegetbands() method.

Modes

The mode of an image defines the type and depth of a pixel in the image.Each pixel uses the full range of the bit depth. So a 1-bit pixel has a rangeof 0-1, an 8-bit pixel has a range of 0-255 and so on. The current releasesupports the following standard modes:

• 1 (1-bit pixels, black and white, stored with one pixel per byte)


• L (8-bit pixels, black and white)


• P (8-bit pixels, mapped to any other mode using a color palette)


• RGB (3x8-bit pixels, true color)


• RGBA (4x8-bit pixels, true color with transparency mask)


• CMYK (4x8-bit pixels, color separation)


• YCbCr (3x8-bit pixels, color video format)
  
  
  • Note that this refers to the JPEG, and not the ITU-R BT.2020, standard
  
  
  
  


• LAB (3x8-bit pixels, the Lab color space)


• HSV (3x8-bit pixels, Hue, Saturation, Value color space)


• I (32-bit signed integer pixels)


• F (32-bit floating point pixels)

Pillow also provides limited support for a few special modes, including:

• LA (L with alpha)


• PA (P with alpha)


• RGBX (true color with padding)


• RGBa (true color with premultiplied alpha)


• La (L with premultiplied alpha)


• I;16 (16-bit unsigned integer pixels)


• I;16L (16-bit little endian unsigned integer pixels)


• I;16B (16-bit big endian unsigned integer pixels)


• I;16N (16-bit native endian unsigned integer pixels)


• BGR;15 (15-bit reversed true colour)


• BGR;16 (16-bit reversed true colour)


• BGR;24 (24-bit reversed true colour)


• BGR;32 (32-bit reversed true colour)

However, Pillow doesn’t support user-defined modes; if you need to handle bandcombinations that are not listed above, use a sequence of Image objects.

You can read the mode of an image through the modeattribute. This is a string containing one of the above values.

Size

You can read the image size through the sizeattribute. This is a 2-tuple, containing the horizontal and vertical size inpixels.

Coordinate System

The Python Imaging Library uses a Cartesian pixel coordinate system, with (0,0)in the upper left corner. Note that the coordinates refer to the implied pixelcorners; the centre of a pixel addressed as (0, 0) actually lies at (0.5, 0.5).

Coordinates are usually passed to the library as 2-tuples (x, y). Rectanglesare represented as 4-tuples, with the upper left corner given first. Forexample, a rectangle covering all of an 800x600 pixel image is written as (0,0, 800, 600).

Palette

The palette mode (P) uses a color palette to define the actual color foreach pixel.

Info

You can attach auxiliary information to an image using theinfo attribute. This is a dictionary object.

How such information is handled when loading and saving image files is up tothe file format handler (see the chapter on Image file formats). Mosthandlers add properties to the info attribute whenloading an image, but ignore it when saving images.

Orientation

A common element of the info attribute for JPG andTIFF images is the EXIF orientation tag. This is an instruction for how theimage data should be oriented. For example, it may instruct an image to berotated by 90 degrees, or to be mirrored. To apply this information to animage, exif_transpose() can be used.

Filters

For geometry operations that may map multiple input pixels to a single outputpixel, the Python Imaging Library provides different resampling filters.

• NEAREST
• Pick one nearest pixel from the input image. Ignore all other input pixels.
• BOX
• Each pixel of source image contributes to one pixel of thedestination image with identical weights.For upscaling is equivalent of NEAREST.This filter can only be used with the resize()and thumbnail() methods.

New in version 3.4.0.

• BILINEAR
• For resize calculate the output pixel value using linear interpolationon all pixels that may contribute to the output value.For other transformations linear interpolation over a 2x2 environmentin the input image is used.
• HAMMING
• Produces a sharper image than BILINEAR, doesn’t have dislocationson local level like with BOX.This filter can only be used with the resize()and thumbnail() methods.

New in version 3.4.0.

• BICUBIC
• For resize calculate the output pixel value using cubic interpolationon all pixels that may contribute to the output value.For other transformations cubic interpolation over a 4x4 environmentin the input image is used.
• LANCZOS
• Calculate the output pixel value using a high-quality Lanczos filter (atruncated sinc) on all pixels that may contribute to the output value.This filter can only be used with the resize()and thumbnail() methods.

New in version 1.1.3.

Filters comparison table