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
//! Post-processing transforms on Julia set images.
//!
//! These transforms are not meant to be exhaustive; rather, they represent some functionality
//! lacking from [`image`] and [`imageproc`] crates. You can absolutely use the transforms from
//! these crates together with (or instead of) transforms defined here.
//!
//! [`image`]: https://crates.io/crates/image
//! [`imageproc`]: https://crates.io/crates/imageproc
//!
//! # Examples
//!
//!
//! ```
//! # use image::GrayImage;
//! use julia_set::transform::{ApplyTransform, Negative, Smoothstep};
//!
//! let image: GrayImage = // ...
//! #    GrayImage::from_raw(10, 10, vec![0_u8; 100]).unwrap();
//! let transformed_image = image.apply(Negative).apply(Smoothstep).transform();
//! ```

use image::{ImageBuffer, Luma, Pixel};

use std::fmt;

/// Pixel-wise transform.
///
/// The function defined by the transform is applied separately to each pixel of the source image.
/// The transform may change the type of pixels (e.g., transform from grayscale to a color image),
/// or may leave it intact.
pub trait PixelTransform<Pix: Pixel> {
    /// Pixel type for the output image.
    type Output: Pixel + 'static;

    /// Performs transform on a single pixel.
    fn transform_pixel(&self, pixel: Pix) -> Self::Output;
}

/// No-op transform.
impl<Pix: Pixel + 'static> PixelTransform<Pix> for () {
    type Output = Pix;

    #[inline]
    fn transform_pixel(&self, pixel: Pix) -> Self::Output {
        pixel
    }
}

impl<Pix, O> PixelTransform<Pix> for Box<dyn PixelTransform<Pix, Output = O>>
where
    Pix: Pixel,
    O: Pixel + 'static,
{
    type Output = O;

    #[inline]
    fn transform_pixel(&self, pixel: Pix) -> Self::Output {
        (**self).transform_pixel(pixel)
    }
}

/// Composition of two transforms. The transforms are applied in the same order they are listed
/// in a tuple.
impl<F, G, Pix: Pixel> PixelTransform<Pix> for (F, G)
where
    F: PixelTransform<Pix>,
    G: PixelTransform<F::Output>,
{
    type Output = G::Output;

    #[inline]
    fn transform_pixel(&self, pixel: Pix) -> Self::Output {
        self.1.transform_pixel(self.0.transform_pixel(pixel))
    }
}

/// Transform that negates the value of each pixel of the input image.
#[derive(Debug, Clone, Copy, Default)]
pub struct Negative;

impl PixelTransform<Luma<u8>> for Negative {
    type Output = Luma<u8>;

    #[inline]
    fn transform_pixel(&self, pixel: Luma<u8>) -> Self::Output {
        Luma([u8::max_value() - pixel[0]])
    }
}

/// Transform that smooths the value of each pixel of the input image using cubic
/// Hermite interpolation (aka `smoothstep` from GLSL / OpenCL).
#[derive(Debug, Clone, Copy, Default)]
pub struct Smoothstep;

impl PixelTransform<Luma<u8>> for Smoothstep {
    type Output = Luma<u8>;

    #[inline]
    #[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
    fn transform_pixel(&self, pixel: Luma<u8>) -> Self::Output {
        let clamped_x = f32::from(pixel[0]) / 255.0;
        let output = clamped_x * clamped_x * (3.0 - 2.0 * clamped_x);
        Luma([(output * 255.0).round() as u8])
    }
}

/// Transform that colorizes the image using a palette.
#[derive(Clone)]
pub struct Palette<T> {
    pixels: [T; 256],
}

impl<T: fmt::Debug> fmt::Debug for Palette<T> {
    fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result {
        formatter
            .debug_struct("Palette")
            .field("pixels", &(&self.pixels as &[T]))
            .finish()
    }
}

impl<T> Palette<T>
where
    T: Pixel<Subpixel = u8> + 'static,
{
    /// Creates a palette based on the provided color stops. The provided `colors` are distributed
    /// at equal distances along the palette. Intermediate colors are obtained
    /// by linear interpolation between the two nearest color stops.
    ///
    /// # Panics
    ///
    /// - Panics if the number of `colors` is less than 2 or greater than 256.
    #[allow(
        clippy::cast_precision_loss,
        clippy::cast_possible_truncation,
        clippy::cast_sign_loss
    )]
    pub fn new(colors: &[T]) -> Self {
        assert!(colors.len() >= 2, "palette must contain at least 2 colors");
        assert!(
            colors.len() <= 256,
            "palette cannot contain more than 256 colors"
        );
        let len_scale = (colors.len() - 1) as f32;

        let mut pixels = [T::from_channels(0, 0, 0, 0); 256];
        for (i, pixel) in pixels.iter_mut().enumerate() {
            let float_i = i as f32 / 255.0 * len_scale;

            let mut prev_color_idx = float_i as usize; // floors as expected
            if prev_color_idx == colors.len() - 1 {
                prev_color_idx -= 1;
            }
            debug_assert!(prev_color_idx + 1 < colors.len());

            let prev_color = colors[prev_color_idx].channels();
            let next_color = colors[prev_color_idx + 1].channels();
            let blend_factor = float_i - prev_color_idx as f32;
            debug_assert!((0.0..=1.0).contains(&blend_factor));

            let mut blended_channels = [0_u8; 4];
            let channel_count = T::CHANNEL_COUNT as usize;
            for (ch, blended_channel) in blended_channels[..channel_count].iter_mut().enumerate() {
                let blended = f32::from(prev_color[ch]) * (1.0 - blend_factor)
                    + f32::from(next_color[ch]) * blend_factor;
                *blended_channel = blended.round() as u8;
            }
            *pixel = *T::from_slice(&blended_channels[..channel_count]);
        }

        Self { pixels }
    }
}

impl<Pix: Pixel + 'static> PixelTransform<Luma<u8>> for Palette<Pix> {
    type Output = Pix;

    #[inline]
    fn transform_pixel(&self, pixel: Luma<u8>) -> Self::Output {
        self.pixels[pixel[0] as usize]
    }
}

/// Trait allowing to lazily apply one or more [`PixelTransform`]s to an image.
///
/// This trait is implemented for [`ImageBuffer`]s and for [`ImageAndTransform`]s, allowing
/// to chain transforms.
pub trait ApplyTransform<Pix: Pixel, F> {
    /// Combined transform after applying `transform`.
    type CombinedTransform: PixelTransform<Pix>;
    /// Appends `transform` to the list of transforms to be performed on the source image.
    fn apply(self, transform: F) -> ImageAndTransform<Pix, Self::CombinedTransform>;
}

/// Source image together with one or more [`PixelTransform`]s. Transforms are applied lazily,
/// once [`Self::transform()`] is called.
#[derive(Debug)]
pub struct ImageAndTransform<Pix, F>
where
    Pix: Pixel,
{
    source_image: ImageBuffer<Pix, Vec<Pix::Subpixel>>,
    transform: F,
}

impl<Pix, F> ImageAndTransform<Pix, F>
where
    Pix: Pixel + Copy + 'static,
    F: PixelTransform<Pix>,
    <F::Output as Pixel>::Subpixel: 'static,
{
    /// Applies transforms accumulated in this object to the source image.
    pub fn transform(&self) -> ImageBuffer<F::Output, Vec<<F::Output as Pixel>::Subpixel>> {
        let mut output = ImageBuffer::new(self.source_image.width(), self.source_image.height());

        let output_iter = self
            .source_image
            .enumerate_pixels()
            .map(|(x, y, pixel)| (x, y, self.transform.transform_pixel(*pixel)));
        for (x, y, out_pixel) in output_iter {
            output[(x, y)] = out_pixel;
        }
        output
    }
}

impl<Pix, F> ApplyTransform<Pix, F> for ImageBuffer<Pix, Vec<Pix::Subpixel>>
where
    Pix: Pixel,
    F: PixelTransform<Pix>,
{
    type CombinedTransform = F;

    fn apply(self, transform: F) -> ImageAndTransform<Pix, F> {
        ImageAndTransform {
            source_image: self,
            transform,
        }
    }
}

impl<Pix, F, G> ApplyTransform<Pix, G> for ImageAndTransform<Pix, F>
where
    Pix: Pixel,
    F: PixelTransform<Pix>,
    G: PixelTransform<F::Output>,
{
    type CombinedTransform = (F, G);

    fn apply(self, transform: G) -> ImageAndTransform<Pix, (F, G)> {
        ImageAndTransform {
            source_image: self.source_image,
            transform: (self.transform, transform),
        }
    }
}

#[cfg(test)]
#[allow(
    clippy::cast_possible_truncation,
    clippy::cast_precision_loss,
    clippy::cast_sign_loss
)]
mod tests {
    use super::*;
    use image::{GrayImage, Rgb};

    #[test]
    fn simple_transform() {
        let image = GrayImage::from_fn(100, 100, |x, y| Luma::from([(x + y) as u8]));
        let image = image.apply(Negative).apply(Smoothstep).transform();
        for (x, y, pix) in image.enumerate_pixels() {
            let negated = (255 - x - y) as f32 / 255.0;
            let smoothed = negated * negated * (3.0 - 2.0 * negated);
            let expected_pixel = (smoothed * 255.0).round() as u8;
            assert_eq!(pix[0], expected_pixel);
        }
    }

    #[test]
    fn palette_basics() {
        let palette = Palette::new(&[Rgb([0, 255, 0]), Rgb([255, 255, 255])]);
        for (i, &pixel) in palette.pixels.iter().enumerate() {
            assert_eq!(pixel, Rgb([i as u8, 255, i as u8]));
        }
    }
}