Skip to content

How Slicing Works

This section provides a conceptual overview of the core logic behind image-slicer.

The Core Engine: pyvips

image-slicer uses pyvips, a Python binding for the libvips image processing library. libvips is designed for performance and can process very large images without loading them entirely into memory. It works by streaming the image from disk, processing it in small chunks, and writing the output to disk. This is what allows image-slicer to be so fast and memory-efficient.

Slicing Logic

When you request to slice an image, here's the process image-slicer follows:

  1. Calculate Tile Dimensions:

    • If you provide cols and rows, the tile width and height are calculated by dividing the image dimensions by the number of columns and rows.
    • If you provide number_of_tiles, the library finds all the factor pairs of that number (e.g., for 12, it finds (1, 12), (2, 6), (3, 4), etc.). It then chooses the pair with the smallest difference (e.g., (3, 4)) to create a grid that is as close to a square as possible.

  2. Handle Imperfect Divisions:

    • It's rare for an image's dimensions to be perfectly divisible by the number of tiles. For example, an 800-pixel wide image sliced into 3 columns would result in tiles that are 266.66 pixels wide.
    • image-slicer handles this by using math.ceil() to round the tile dimension up to the nearest whole pixel (e.g., 267 pixels).
    • When cropping the tiles, it ensures that the crop area never goes beyond the image boundaries. This means that tiles on the right and bottom edges might be slightly smaller than the other tiles. This is known as creating "partial" tiles.

  3. Iterate and Crop:

    • The library iterates through the calculated grid, from top to bottom and left to right.
    • For each position, it uses pyvips to crop a section of the source image corresponding to the tile's dimensions.
    • The cropped tile is then saved to the specified output directory with the formatted filename.