a-sixel 0.4.0

A small sixel + palette selection + dithering library.
Documentation

a-sixel

A-Sixel library for encoding sixel images.

Basic Usage

use a_sixel::BitMergeSixelEncoderBest;
use image::RgbImage;

let img = RgbImage::new(100, 100);
println!("{}", <BitMergeSixelEncoderBest>::encode(&img));

Choosing an Encoder

  • I want good quality:
    • Use BitMergeSixelEncoderBest or KMeansSixelEncoder.
  • I'm time constrained:
    • Use BitMergeSixelEncoderLow, BitSixelEncoder, or OctreeSixelEncoder.
  • I'm really time constrained and can sacrifice a little quality:
    • Use BitSixelEncoder<NoDither>.

For a more detailed breakdown, here's the encoders by average speed and quality against the test images (speed figures will vary) at 256 colors with Sierra dithering:

Algorithm MSE DSSIM PHash Distance Mean ΔE Max ΔE ΔE >2.3 ΔE >5.0 Execution Time (ms)
adu 15.07 0.0053 8.66 1.79 12.88 31.6% 4.4% 1416
bit 35.82 0.0132 31.14 3.16 11.03 64.5% 15.1% 426
bit-no-dither 31.78 0.0214 39.10 3.09 10.23 64.0% 13.4% 274
bit-merge-low 10.36 0.0037 13.55 1.89 10.01 31.0% 2.2% 932
bit-merge 10.29 0.0037 13.52 1.83 10.26 30.6% 2.2% 1496
bit-merge-better 10.31 0.0037 13.45 1.85 10.21 30.6% 2.2% 1275
bit-merge-best 10.67 0.0038 13.97 1.95 9.98 32.4% 2.2% 785
focal 14.62 0.0056 19.97 2.30 9.16 45.3% 3.3% 2428
k-means 10.07 0.0036 13.07 1.80 10.17 29.1% 2.2% 2996
k-medians 17.67 0.0068 21.07 2.61 10.17 53.6% 5.1% 7305
median-cut 19.63 0.0059 16.45 2.24 10.36 42.2% 5.9% 692
octree 54.48 0.0148 26.03 3.89 12.49 78.6% 25.4% 682
wu 17.89 0.0068 21.03 2.34 10.24 46.3% 5.1% 1853

Note: Execution time includes the time taken to compute error statistics - this is non-trivial. For example, exclusive of error statistics computation, bit-no-dither takes <100ms on average. Performance figures will vary based on machine, etc. They are only useful for comparing algorithms against each other within this dataset.

Here's the encoders at 16 colors with Sierra dithering:

Algorithm MSE DSSIM PHash Distance Mean ΔE Max ΔE ΔE >2.3 ΔE >5.0 Execution Time (ms)
adu 116.85 0.036 39.83 4.02 18.39 66% 33% 332
bit 178.47 0.049 59.79 5.53 16.61 89% 51% 307
bit-merge 95.17 0.030 41.52 3.95 16.16 67% 31% 712
focal 118.57 0.035 48.59 4.36 16.88 72% 34% 2150
k-means 99.36 0.031 43.10 3.99 16.41 67% 31% 637
k-medians 166.88 0.050 60.59 5.48 16.77 88% 52% 5447
median-cut 164.52 0.037 45.28 4.68 16.72 74% 42% 374
octree 459.37 0.085 75.07 7.69 18.89 98% 74% 446
wu 125.84 0.039 50.52 4.48 16.70 75% 39% 906

This doesn't tell the full story, as sometimes a low MSE or DSSIM can be achieved while losing some highlight colors in the image. Take flowers.png for example:

This preserves the grey shades that make up the image well, but completely loses the blue of the flowers at the base of the trees.

  • MSE: 3.23
  • DSSIM: 0.0020
  • PHash Distance: 10
  • Mean ΔE: 1.42
  • Max ΔE: 11.49
  • ΔE >2.3: 14%
  • ΔE >5.0: 0%
  • MSE: 9.15
  • DSSIM: 0.0091
  • PHash Distance: 16
  • Mean ΔE: 2.10
  • Max ΔE: 9.23
  • ΔE >2.3: 38%
  • ΔE >5.0: 2%

This sacrifices some differentiation between shades of grey, but preserves the blue of the flowers.

License: MIT OR Apache-2.0