okolors 0.9.0

Create color palettes from images by performing k-means clustering in the Oklab color space.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
//! Create a color palette from an image using k-means clustering in the Oklab color space.
//!
//! This library is a simple wrapper around the [`quantette`] crate but only exposes functionality
//! for generating color palettes. Additionally, this crate adds a few additional options not
//! present in [`quantette`]. It is intended that additional stylistic and opinionated options will
//! be added to this crate, whereas [`quantette`] is solely focused on accurate color quantization.
//!
//! # Features
//!
//! This crate has three features that are enabled by default:
//! - `threads`: adds the [`parallel`](Okolors::parallel) option.
//! - `image`: enables integration with the [`image`] crate.
//! - `std`: use Rust's standard library (this crate is `no_std` compatible if all features are disabled).
//!
//! # Examples
//!
//! To start, create an [`Okolors`] from a [`RgbImage`] (note that the `image` feature is needed):
//! ```
//! # fn main() -> Result<(), okolors::LengthOutOfRange> {
//! use okolors::Okolors;
//!
//! // let img = image::open("some image")?.into_rgb8();
//! let img = image::RgbImage::new(256, 256);
//! let palette_builder = Okolors::try_from(&img)?;
//! # Ok(())
//! # }
//! ```
//!
//! Instead of an [`RgbImage`], a slice of [`Srgb<u8>`] colors can be used instead:
//! ```
//! # use okolors::Okolors;
//! # use palette::Srgb;
//! let srgb = vec![Srgb::new(0, 0, 0)];
//! let palette_builder = Okolors::new(&srgb).unwrap();
//! ```
//!
//! If the default options aren't to your liking, you can tweak them:
//! ```
//! # use okolors::{LengthOutOfRange, Okolors};
//! # use image::RgbImage;
//! # fn main() -> Result<(), LengthOutOfRange> {
//! # let img = RgbImage::new(0, 0);
//! use okolors::{PaletteSize, KmeansOptions};
//!
//! let palette_builder = Okolors::try_from(&img)?
//!     .palette_size(PaletteSize::from_u8_clamped(16))
//!     .lightness_weight(0.5)
//!     .kmeans_options(KmeansOptions::new().sampling_factor(1.0))
//!     .sort_by_frequency(true)
//!     .parallel(true);
//! # Ok(())
//! # }
//! ```
//!
//! To finally generate the palette, use:
//! - [`Okolors::srgb8_palette`] for a [`Srgb<u8>`] palette
//! - [`Okolors::srgb_palette`] for a [`Srgb`] palette (components are `f32` instead of `u8`)
//! - [`Okolors::oklab_palette`] for an [`Oklab`] palette
//!
//! To clarify, the [`Oklab`] colorspace is used to quantize the colors in all cases.
//! The methods above just determine what colorspace you want the final colors converted into.
//!
//! All of the color types present in the public API for this crate
//! (like [`Srgb`] or [`Oklab`]) are from the [`palette`] crate.
//! You can check it out for more information. For example, its documentation
//! should provide you everything you need to know to [cast](palette::cast)
//! a `Vec<Srgb<u8>>` into a `Vec<[u8; 3]>` or vice versa.

#![warn(
    clippy::unwrap_used,
    clippy::expect_used,
    clippy::unreachable,
    clippy::panic,
    clippy::exit,
    clippy::unused_result_ok,
    clippy::print_stdout,
    clippy::print_stderr,
    missing_docs
)]

pub mod deps;

// We have tight integration/control over `quantette`, let's re-export the types directly.
pub use quantette::{LengthOutOfRange, MAX_PIXELS, PaletteSize, kmeans::KmeansOptions};

#[cfg(feature = "image")]
use image::RgbImage;
use palette::{IntoColor, Oklab, Srgb};
use quantette::{
    ImageRef,
    color_space::{oklab_to_srgb8, srgb8_to_oklab},
    dedup,
    kmeans::Kmeans,
    wu::{BinnerF32x3, WuF32x3},
};
#[cfg(feature = "threads")]
use {quantette::color_space::srgb8_to_oklab_par, rayon::prelude::*};

/// A builder struct to specify options for palette generation.
///
/// See the [crate] documentation for more information and examples.
#[derive(Debug, Clone)]
#[must_use]
pub struct Okolors<'a> {
    /// The colors to create a palette from.
    image: ImageRef<'a, Srgb<u8>>,
    /// The amount to scale down the lightness component by.
    lightness_weight: f32,
    /// The number of colors to have in the palette.
    palette_size: PaletteSize,
    /// The options to use for k-means quantization.
    kmeans_options: KmeansOptions,
    /// Whether or not to dedup the input colors as an optimization.
    dedup: Option<bool>,
    /// Whether to return the palette sorted by increasing frequency.
    sort_by_frequency: bool,
    #[cfg(feature = "threads")]
    /// Whether or not to use parallelism.
    parallel: bool,
}

#[cfg(feature = "image")]
impl<'a> TryFrom<&'a RgbImage> for Okolors<'a> {
    type Error = LengthOutOfRange;

    /// Try to create a [`Okolors`] from an [`RgbImage`].
    /// If the number of pixels in the image is above [`MAX_PIXELS`], then an error is returned.
    #[inline]
    fn try_from(image: &'a RgbImage) -> Result<Self, Self::Error> {
        let image = ImageRef::try_from(image)?;
        Ok(Self::from_image(image))
    }
}

impl<'a> From<ImageRef<'a, Srgb<u8>>> for Okolors<'a> {
    #[inline]
    fn from(image: ImageRef<'a, Srgb<u8>>) -> Self {
        Self::from_image(image)
    }
}

impl<'a> Okolors<'a> {
    #[inline]
    fn from_image(image: ImageRef<'a, Srgb<u8>>) -> Self {
        Self {
            image,
            lightness_weight: 0.325,
            palette_size: PaletteSize::from_u8_clamped(8),
            kmeans_options: KmeansOptions::new().sampling_factor(0.5),
            dedup: None,
            sort_by_frequency: false,
            #[cfg(feature = "threads")]
            parallel: false,
        }
    }

    /// Creates a new [`Okolors`] with default options.
    ///
    /// Returns `None` if the length of `colors` is above [`MAX_PIXELS`].
    ///
    /// Alternatively, use `Okolors::try_from` or `try_into` on an [`RgbImage`].
    #[inline]
    pub fn new(colors: &'a [Srgb<u8>]) -> Option<Self> {
        let len = colors.len().try_into().ok()?;
        let image = ImageRef::new(len, 1, colors).ok()?;
        Some(Self::from_image(image))
    }

    /// Sets the lightness weight used to scale down the lightness component of the colors.
    ///
    /// The brightness of colors has more influence on the perceived difference between colors.
    /// So, the generated the palette may contain colors that differ mainly in brightness only.
    /// The lightness weight is used scale down the lightness component of the colors,
    /// potentially bringing out more distinct hues in the final color palette.
    /// One downside to this is that colors near white and black may be merged into a shade of gray.
    ///
    /// The provided `lightness_weight` should be in the range `0.0..=1.0`,
    /// and it is clamped to this range otherwise.
    ///
    /// The default lightness weight is `0.325`.
    #[inline]
    pub fn lightness_weight(mut self, lightness_weight: f32) -> Self {
        self.lightness_weight = lightness_weight.clamp(f32::EPSILON, 1.0);
        self
    }

    /// Sets the palette size which determines the (maximum) number of colors to have in the palette.
    ///
    /// See the docs for [`PaletteSize`] for more information.
    ///
    /// The default palette size is `8`.
    ///
    /// # Examples
    ///
    /// ```
    /// # use okolors::{Okolors, PaletteSize};
    /// # fn main() -> Result<(), quantette::PaletteSizeFromIntError> {
    /// # let srgb = vec![palette::Srgb::new(0, 0, 0)];
    /// Okolors::new(&srgb)
    ///     .unwrap()
    ///     .palette_size(24u16.try_into()?)
    ///     .palette_size(PaletteSize::from_u8_clamped(16))
    ///     .palette_size(PaletteSize::MAX);
    /// # Ok(())
    /// # }
    /// ```
    #[inline]
    pub fn palette_size(mut self, palette_size: PaletteSize) -> Self {
        self.palette_size = palette_size;
        self
    }

    /// Sets the options to use for k-means quantization.
    ///
    /// See the docs for [`KmeansOptions`] for more information.
    ///
    /// The default k-means options is `KmeansOptions::new().sampling_factor(0.5)`.
    #[inline]
    pub fn kmeans_options(mut self, options: KmeansOptions) -> Self {
        self.kmeans_options = options;
        self
    }

    /// Sets whether or not to deduplicate colors in the input as an optimization.
    ///
    /// For large images it is recommended to deduplicate pixels for overall faster quantization.
    /// For smaller images, not performing deduplication can be faster. The optimal cutoff between a
    /// "large" and "small" image depends on:
    /// - the ratio between the number of unique pixels in the image and the total number of pixels in the image
    /// - hardware and enabled `target_feature`s
    ///
    /// Some experimentation may be needed based on your workflow and hardware.
    ///
    /// The default value is `None` (automatically choose whether or not to dedup).
    ///
    /// # Examples
    ///
    /// ```
    /// # use okolors::Okolors;
    /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
    /// # let srgb = vec![palette::Srgb::new(0, 0, 0)];
    /// Okolors::new(srgb.as_slice())
    ///     .unwrap()
    ///     .dedup(None) // automatically choose whether or not to dedup
    ///     .dedup(true) // always dedup
    ///     .dedup(false); // never dedup
    /// # Ok(())
    /// # }
    /// ```
    #[inline]
    pub fn dedup(mut self, dedup: impl Into<Option<bool>>) -> Self {
        self.dedup = dedup.into();
        self
    }

    /// Sort the returned palette by ascending frequency.
    ///
    /// Frequency refers to the number of pixel samples assigned to each palette color during
    /// k-means quantization.
    ///
    /// By default, the palette is not sorted.
    #[inline]
    pub fn sort_by_frequency(mut self, sort: bool) -> Self {
        self.sort_by_frequency = sort;
        self
    }

    /// Computes the color palette and returns it as [`Oklab`] colors.
    #[allow(clippy::missing_panics_doc)]
    #[must_use]
    pub fn oklab_palette(&self) -> Vec<Oklab> {
        fn finalize(
            kmeans: Kmeans<Oklab, f32, 3>,
            sort: bool,
            lightness_weight: f32,
        ) -> Vec<Oklab> {
            let mut palette = if sort {
                let (palette, counts) = kmeans.into_palette_and_counts();
                let mut pairs = palette.into_iter().zip(counts).collect::<Vec<_>>();
                pairs.sort_by_key(|&(_, n)| n);
                pairs.into_iter().map(|(c, _)| c).collect()
            } else {
                kmeans.into_palette().into_vec()
            };

            for color in &mut palette {
                color.l /= lightness_weight;
            }

            palette
        }

        let Self {
            image,
            palette_size,
            kmeans_options,
            dedup,
            lightness_weight,
            sort_by_frequency,
            #[cfg(feature = "threads")]
            parallel,
        } = *self;

        if image.is_empty() {
            return Vec::new();
        }

        let binner = {
            let mut ranges = BinnerF32x3::<16, 32, 16>::OKLAB_COMPONENT_RANGES_FROM_SRGB8;
            ranges[0].1 *= lightness_weight;
            BinnerF32x3::<16, 32, 16>::new(ranges)
        };

        #[cfg(feature = "threads")]
        if parallel {
            let kmeans = if dedup.unwrap_or(image.num_pixels() >= 2048 * 2048) {
                let mut image = dedup::dedup_image_u8_3_counts_par(image)
                    .map(|palette| srgb8_to_oklab_par(&palette));
                image
                    .palette_mut()
                    .par_iter_mut()
                    .for_each(|c| c.l *= lightness_weight);
                #[allow(clippy::expect_used)]
                let centroids = WuF32x3::run_indexed_image_counts_par(&image, binner)
                    .expect("non-empty image")
                    .palette(palette_size);
                Kmeans::run_indexed_image_par(image.as_ref(), centroids, kmeans_options)
            } else {
                let mut image = image.map(srgb8_to_oklab_par);
                image
                    .as_mut_slice()
                    .par_iter_mut()
                    .for_each(|c| c.l *= lightness_weight);
                #[allow(clippy::expect_used)]
                let centroids = WuF32x3::run_image_par(image.as_ref(), binner)
                    .expect("non-empty image")
                    .palette(palette_size);
                Kmeans::run_image_par(image.as_ref(), centroids, kmeans_options)
            };
            return finalize(kmeans, sort_by_frequency, lightness_weight);
        }

        let kmeans = if dedup.unwrap_or(image.num_pixels() >= 2048 * 2048) {
            let mut image =
                dedup::dedup_image_u8_3_counts(image).map(|palette| srgb8_to_oklab(&palette));
            for c in image.palette_mut() {
                c.l *= lightness_weight;
            }
            #[allow(clippy::expect_used)]
            let centroids = WuF32x3::run_indexed_image_counts(&image, binner)
                .expect("non-empty image")
                .palette(palette_size);
            Kmeans::run_indexed_image(image.as_ref(), centroids, kmeans_options)
        } else {
            let mut image = image.map(srgb8_to_oklab);
            for c in image.as_mut_slice() {
                c.l *= lightness_weight;
            }
            #[allow(clippy::expect_used)]
            let centroids = WuF32x3::run_image(image.as_ref(), binner)
                .expect("non-empty image")
                .palette(palette_size);
            Kmeans::run_image(image.as_ref(), centroids, kmeans_options)
        };

        finalize(kmeans, sort_by_frequency, lightness_weight)
    }

    /// Computes the color palette and converts it to [`Srgb<u8>`] colors.
    #[must_use]
    #[inline]
    pub fn srgb8_palette(&self) -> Vec<Srgb<u8>> {
        let palette = self.oklab_palette();
        oklab_to_srgb8(&palette)
    }

    /// Computes the color palette and converts it to [`Srgb`] colors.
    #[must_use]
    #[inline]
    pub fn srgb_palette(&self) -> Vec<Srgb> {
        let palette = self.oklab_palette();
        palette.into_iter().map(IntoColor::into_color).collect()
    }
}

#[cfg(feature = "threads")]
impl Okolors<'_> {
    /// Sets whether or not to use multiple threads to compute the palette.
    ///
    /// The number of threads can be configured using a [`rayon::ThreadPool`].
    ///
    /// By default, single-threaded execution is used.
    #[inline]
    pub fn parallel(mut self, parallel: bool) -> Self {
        self.parallel = parallel;
        self
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use palette::{Srgb, cast::ComponentsInto as _};

    #[rustfmt::skip]
    fn test_colors() -> Vec<Srgb<u8>> {
        vec![
            92, 88, 169, 165, 149, 73, 71, 40, 98, 83, 27, 26, 60, 128, 246, 159, 239, 169, 96, 30, 166, 176, 222, 97, 90, 70, 180, 179, 50, 228, 181, 40, 254, 145, 9, 78, 245, 233, 56, 84, 53, 206, 200, 144, 18, 254, 153, 172, 223, 72, 106, 137, 14, 80, 239, 226, 123, 194, 101, 45, 76, 50, 123, 191, 174, 48, 111, 113, 179, 128, 130, 102, 126, 243, 217, 64, 200, 191, 229, 251, 214, 70, 3, 67, 144, 244, 134, 135, 56, 56, 32, 221, 192, 216, 13, 56, 44, 181, 97, 110, 206, 127, 119, 110, 175, 195, 190, 120, 38, 123, 177, 226, 54, 223, 196, 60, 106, 167, 18, 123, 227, 127, 16, 204, 120, 122, 75, 30, 230, 142, 31, 217, 182, 59, 187, 239, 108, 11, 85, 49, 145, 24, 23, 185, 43, 69, 179, 66, 140, 107, 226, 19, 91, 101, 220, 155, 253, 116, 238, 117, 110, 200, 0, 193, 15, 153, 4, 67, 15, 187, 210, 42, 179, 90, 84, 90, 172, 128, 20, 92, 6, 170, 137, 172, 90, 43, 22, 234, 31, 212, 91, 47, 185, 27, 142, 187, 223, 24, 113, 208, 177, 70, 85, 100, 120, 77, 38, 204, 203, 121, 212, 36, 92, 244, 70, 138, 212, 17, 75, 240, 39, 75, 186, 221, 115, 240, 170, 63, 74, 224, 227, 120, 211, 88, 232, 2, 147, 156, 32, 28, 127, 238, 114, 31, 29, 255, 173, 73, 182, 101, 9, 111, 42, 139, 70, 15, 233, 172, 133, 223, 175, 178, 90, 98, 195, 53, 125, 208, 3, 253, 237, 181, 133, 4, 199, 253, 221, 58, 124, 99, 126, 239, 253, 151, 224, 24, 73, 6, 86, 161, 76, 151, 255, 10, 23, 65, 32, 210, 248, 81, 38, 48, 9, 108, 199, 14, 138, 170, 35, 55, 108, 6, 88, 188, 118, 195, 26, 8, 130, 175, 179, 227, 196, 145, 207, 102, 19, 216, 220, 45, 202, 154, 21, 156, 170, 212, 94, 215, 116, 193, 90, 180, 171, 47, 21, 77, 23, 123, 44, 78, 143, 42, 2, 74, 255, 133, 118, 33, 155, 111, 202, 64, 45, 63, 63, 214, 96, 139, 135, 219, 245, 179, 192, 127, 60, 35, 241, 104, 248, 230, 223, 6, 62, 218, 80, 205, 120, 184, 35, 154, 32, 105, 114, 8, 41, 111, 62, 6, 25, 95, 92, 58, 93, 240, 248, 169, 200, 218, 52, 158, 14, 14, 55, 250, 127, 136, 121, 102, 27, 182, 227, 205, 181, 76, 120, 153, 207, 142, 125, 169, 141, 124, 248, 169, 158, 190, 9, 3, 167, 93, 100, 181, 73, 103, 242, 153, 223, 185, 64, 151, 120, 186, 169, 4, 95, 137, 88, 82, 119, 205, 53, 97, 74, 55, 163, 232, 225, 60, 124, 248, 135, 218, 11, 99, 85, 108, 37, 178, 126, 222, 178, 80, 48, 13, 176, 179, 111, 243, 161, 245, 57, 252, 241, 40, 10, 190, 17, 17, 249, 108, 253, 70, 237, 119, 123, 113, 103, 9, 45, 60, 51, 160, 48, 191, 237, 109, 32, 124, 31, 32, 145, 68, 99, 149, 15, 91, 186, 160, 185, 96, 217, 110, 125, 187, 124, 68, 118, 32, 105, 9, 88, 87, 75, 48, 142, 194, 190, 54, 40, 20, 194, 46, 29, 75, 253, 48, 216, 170, 184, 118, 169, 51, 41, 244, 153, 147, 59, 133, 237, 163, 69, 114, 5, 61, 88, 110, 154, 83, 77, 107, 102, 152, 56, 33, 46, 158, 66, 129, 180, 36, 189, 131, 94, 233, 145, 10, 17, 168, 110, 47, 85, 84, 124, 120, 116, 168, 215, 178, 188, 159, 198, 245, 139, 42, 255, 56, 252, 77, 191, 14, 217, 108, 180, 44, 112, 245, 122, 189, 242, 41, 78, 218, 155, 241, 116, 244, 82, 192, 128, 86, 138, 175, 245, 51, 210, 2, 123, 114, 95, 113, 201, 86, 112, 68, 23, 18, 31, 146, 47, 222, 247, 251, 120, 59, 243, 239, 201, 115, 166, 126, 189, 227, 238, 204, 127, 11, 5, 52, 196, 63, 46, 149, 184, 150, 122, 143, 215, 1, 115, 164, 99, 238, 166, 1, 21, 132, 88, 117, 104, 88, 216, 106, 132, 233, 31, 41, 160, 153, 89, 180, 66, 151, 201, 50, 164, 208, 15, 160, 43, 43, 205, 204, 148, 178, 102, 188, 72, 46, 251, 40, 137, 184, 252, 241, 224, 101, 17, 77, 157, 83, 96, 93, 211, 83, 209, 73, 112, 195, 74, 91, 54, 41, 168, 129, 87, 81, 149, 63, 173, 37, 36, 112, 184, 36, 28, 8, 129, 153, 124, 27, 134, 50, 134, 24, 205, 118, 68, 60, 59, 214, 39, 78, 18, 243, 225, 206, 19, 125, 90, 210, 225, 249, 254, 210, 125, 34, 224, 203, 42, 126, 3, 126, 107, 10, 121, 113, 207, 234, 248, 44, 31, 158, 223, 128, 47, 147, 61, 0, 63, 44, 84, 252, 39, 69, 75, 190, 129, 116, 40, 198, 230, 137, 53, 130, 106, 68, 194, 233, 58, 197, 160, 130, 205, 169, 243, 118, 175, 252, 67, 251, 81, 20, 108, 22, 247, 161, 8, 50, 37, 224, 251, 154, 197, 172, 93, 113, 46, 206, 148, 47, 119, 102, 140, 82, 128, 115, 144, 250, 77, 110, 152, 182, 160, 44, 131, 202, 202, 130, 6, 187, 152, 19, 208, 179, 185, 78, 9, 124, 55, 17, 76, 70, 38, 94, 105, 152, 240, 85, 248, 246, 66, 39, 226, 25, 112, 7, 112, 123, 107, 179, 237, 82, 212, 236, 167, 236, 231, 103, 123, 133, 224, 169, 146, 235, 91, 49, 231, 69, 160, 6, 254, 137, 35, 207, 148, 223, 206, 250, 143, 189, 47, 132, 122, 163, 94, 39, 176, 108, 66, 163, 206, 247, 201, 87, 229, 10, 124, 164, 235, 6, 76, 188, 117, 127, 67, 217, 124, 24, 184, 114, 206, 145, 197, 108, 219, 3, 225, 89, 154, 25, 112, 11, 56, 213, 39, 21, 231, 5, 180, 33, 52, 155, 249, 221, 100, 39, 79, 150, 55, 196, 150, 1, 135, 185, 227, 26, 218, 153, 80, 17, 167, 90, 227, 172, 229, 129, 113, 31, 200, 121, 23, 223, 155, 73, 242, 66, 79, 220, 29, 78, 179, 185, 170, 111, 87, 245, 196, 208, 94, 108, 25, 31, 92, 212, 68, 252, 164, 6, 63, 214, 206, 226, 6, 172, 175, 170, 141, 236, 152, 139, 48, 178, 195, 137, 95, 47, 204, 47, 38, 78, 203, 24, 240, 73, 83, 191, 145, 29, 20, 161, 27, 142, 205, 65, 205, 169, 138, 148, 45, 124, 59, 114, 108, 6, 70, 100, 229, 220, 191, 68, 105, 3, 159, 13, 13, 195, 103, 203, 41, 242, 137, 153, 202, 121, 135, 252, 167, 154, 57, 234, 75, 163, 219, 195, 72, 100, 5, 67, 223, 207, 195, 0, 189, 48, 63, 101, 91, 80, 37, 151, 18, 5, 41, 109, 77, 222, 164, 85, 53, 189, 164, 66, 217, 195, 183, 41, 220, 251, 1, 76, 133, 72, 114, 24, 110, 73, 23, 105, 220, 61, 248, 148, 52, 130, 134, 84, 252, 181, 80, 180, 152, 166, 116, 248, 23, 16, 51, 227, 195, 249, 178, 163, 178, 70, 226, 79, 113, 6, 61, 122, 56, 99, 40, 22, 37, 58, 58, 150, 13, 63, 35, 59, 115, 139, 195, 222, 162, 160, 186, 57, 118, 5, 104, 79, 235, 174, 84, 123, 79, 221, 25, 149, 110, 116, 16, 215, 43, 153, 87, 86, 20, 174, 42, 238, 248, 66, 23, 25, 31, 112, 17, 83, 14, 112, 24, 37, 126, 66, 64, 6, 47, 207, 32, 184, 1, 237, 52, 79, 135, 204, 219, 180, 208, 18, 106, 87, 70, 50, 20, 168, 205, 89, 245, 38, 207, 136, 192, 179, 142, 82, 248, 42, 104, 39, 155, 214, 253, 34, 69, 14, 200, 1, 244, 210, 59, 193, 90, 161, 137, 48, 46, 172, 160, 75, 175, 194, 212, 76, 215, 236, 98, 144, 255, 206, 143, 141, 53, 238, 64, 6, 22, 164, 17, 244, 88, 73, 47, 197, 130, 24, 13, 234, 207, 117, 122, 244, 175, 250, 238, 244, 149, 206, 63, 112, 202, 102, 201, 224, 100, 79, 195, 126, 120, 237, 55, 67, 90, 246, 239, 248, 222, 137, 216, 194, 107, 213, 8, 182, 98, 69, 216, 200, 218, 252, 135, 201, 253, 194, 17, 70, 89, 250, 185, 50, 72, 127, 224, 225, 112, 67, 82, 20, 196, 114, 118, 85, 86, 119, 71, 31, 220, 197, 92, 5, 246, 208, 110, 141, 182, 174, 119, 56, 74, 238, 234, 216, 7, 11, 251, 158, 0, 43, 192, 4, 227, 62, 135, 161, 164, 221, 82, 99, 89, 216, 194, 79, 236, 14, 127, 255, 191, 142, 143, 229, 23, 206, 20, 61, 73, 165, 47, 235, 205, 47, 215, 247, 106, 239, 172, 116, 150, 140, 191, 53, 146, 68, 114, 173, 62, 204, 232, 115, 130, 35, 154, 22, 248, 180, 85, 178, 151, 236, 28, 46, 23, 212, 219, 82, 214, 239, 34, 102, 18, 58, 153, 241, 28, 133, 58, 226, 176, 255, 177, 208, 21, 118, 200, 119, 38, 14, 164, 74, 174, 112, 32, 191
        ].components_into()
    }

    #[test]
    fn zero_lightness_weight() {
        let colors = test_colors();
        let palette = Okolors::new(&colors)
            .unwrap()
            .lightness_weight(0.0)
            .oklab_palette();

        assert!(!palette.into_iter().any(|oklab| oklab.l.is_nan()));
    }

    #[test]
    fn not_enough_colors() {
        let colors = test_colors();
        let k = 100;
        let palette = Okolors::new(&colors[..k])
            .unwrap()
            .palette_size(PaletteSize::MAX)
            .oklab_palette();

        assert!(palette.len() <= k);
    }

    #[test]
    fn no_samples() {
        let colors = test_colors();
        let palette = Okolors::new(colors.as_slice())
            .unwrap()
            .kmeans_options(KmeansOptions::new().sampling_factor(0.0))
            .oklab_palette();

        assert_eq!(palette.len(), 8);
    }

    #[test]
    #[cfg(feature = "threads")]
    fn zero_batch_size() {
        let colors = test_colors();
        let palette = Okolors::new(colors.as_slice())
            .unwrap()
            .kmeans_options(KmeansOptions::new().batch_size(0))
            .parallel(true)
            .oklab_palette();

        assert_eq!(palette.len(), 8);
    }
}