Home

Make images look like pixel art

In this wiki you will find the complete guide to use MakeItPixel.

• Process (the detailed explanation of the MakeItPixel algorithm).
• Palettes (the complete explanation on how to use palettes with MakeItPixel).

Table of contents

• Dependencies
• Build
• Usage
  • CLI options
  • Configuration
    • Scaling
    • Color quantization
    • Palette

Dependencies

• SFML v2.5.1 (release)

Build

g++ src/* -Iinclude -lsfml-graphics -o makeitpixel

Usage

makeitpixel [-h] [-c FILE] [-x JSON] [-o DIR] FILES..

CLI options

Option	Description
-h, --help	Print this help message and exit.
-x, --config CONFIG	Set the CLI configuration as a JSON formatted string.
-c, --config-file PATH	Set the configuration file.
-o, --output-dir DIR	Set the output directory for the generated images.

Configuration

There are two levels of configuration:

1. File configuration. Overwrites default options.
2. CLI configuration. Overwrites file configuration and default options.

The configuration is specified in JSON format. Configuration parameters will be explained in detail below.

Scaling

The first step of the process is reducing the size of the image. The parameters involved are:

• width: Maximum width of the resultant image. Must be a number (default = 64).
• height: Maximum height of the resultant image. Must be a number (default = 64).
• select_pixel: Scaling down an image results on a loss of information, so this parameter tells what pixels to keep:

  Value	Effect
  "min"	Take the darker pixels. Results in darker images.
  "max"	Take the lighter pixels. Results in lighter images.
  "med"	Take the median pixel. Results in sharper images.
  "avg" (default)	Take the median pixel. Results in smoother images.

• normalize: To take advantage of the color processing to be performed later, may want to normalize the image (Make the lightest color white and the darkest, black).

  Value	Effect
  "no" (default)	Don't perform any normalization.
  "pre"	Normalize before scaling.
  "post"	Normalize after scaling.

Color quantization

The second step, optionally, is reducing the color space of the image. Certain strategies might require palette configuration.

• quantization: Strategy for choosing the new color of a pixel.

  Value	Effect
  "none" (default)	Skip this step and leave the image as it is after scaling.
  "bit1", "bit2", ..., "bit8"	Set the bits available to represent color for each RGB channel. "bit1" reduces the space to just 8 colors and "bit8" results in no change (equivalent to "none").
  "closest_rgb"	Choose the color from the palette that's closer in the RGB space. Useful for rich palettes.
  "closest_gray"	Choose the color from the palette with a closer gray value. Useful for sequential palettes.

• dithering: Object with parameters for dithering, listed below.

• dithering.method: Algorithm for dithering to apply during the quantization.

  Value	Effect
  "none" (default)	Quantize without dithering.
  "floydsteinberg"	Use the Floyd-Steinberg algorithm, based on error propagation.
  "ordered"	Use a matrix for an ordered dithering algorithm.

• dithering.matrix: For ordered dithering. Matrix to use: "Bayes2", "Bayes4" (default) or "Bayes8", that result in squared patch-like patterns; "Horizontal2" or "Horizontal4", that result in horizontal patterns; "Vertical2" or "Vertical4", that result in vertical patterns.

• dithering.threshold: Not implemented (default = 0).

• dithering.sparsity: For ordered dithering. Distance expected between the possible values of each RGB channel in the reduced color space.

  Value	Effect
  "auto" (default)	Compute most convenient sparsity for the quantization strategy and/or palette.
  Any number between 0 and 255	Specify manually the distance. A 0 value results in no dithering at all.

Palette

If you are using a quantization strategy that requires a palette, you can configure it with the following parameters:

• palette
  • A literal palette can be specified as an array of strings containing the hexadecimal values of the colors.
  • An auto-generated palette will be used if this value is an object. The parameters inside this object are described below.
• palette.main: A single color to use as reference. Must be a string with an hexadecimal value (default = "00ff00").
• palette.scheme: Color scheme to generate the base colors of the palette from the main one.

  Value	Base colors	Shifted hue degrees
  "mono"	1	0
  "analogous" (default)	3	30, 60
  "complementary"	2	180
  "split_complementary"	3	150, 210
  "triadic"	3	120, 240
  "rectangle"	4	60, 180, 240
  "square"	4	90, 180, 270

• palette.spectre: Defines how the palette is extended from the base colors, adding brighter and darker colors.

  Value	Effect
  "linear"	Only add darker values for the darker base color and the same for the brighter ones. Results in sequential palettes.
  "complete"	Add darker and brighter values for each base color. Result in richer palettes.

• palette.inter: Aproximate number of intermediate darker and brighter values.
• palette.disparity: Factor between 0 and 1 to get the darker and brighter values of the palette by interpolation. A value of 0 results in leaving only the base colors; a value of 1 includes black and white in the palette. Must be a number between 0 and 1 (default = 0.85).