colconv 0.1.0

SIMD-dispatched color-conversion kernels covering the FFmpeg AVPixelFormat space, with a Sink-based API so consumers pick which derived outputs (RGB / Luma / HSV / custom) they want without paying for the ones they don't.
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

use super::super::*;

// ---- Tier 3 packed YUV 4:2:2 NEON scalar-equivalence ---------------
//
// 6 kernels (3 formats × {RGB, RGBA}) plus 3 luma extractors share a
// single deinterleave shape (`vld2q_u8` + `vuzp_u8`). The tests sweep
// every block-size boundary on the 16-pixel NEON main loop plus tail
// widths to exercise the scalar fallback dispatch.

/// Deterministic packed YUYV422-style buffer: `2 * width` bytes per
/// row, hash-like seed per byte position.
pub(super) fn packed_yuv422_buffer(width: usize, seed: usize) -> std::vec::Vec<u8> {
  (0..width * 2)
    .map(|i| ((i.wrapping_mul(seed).wrapping_add(seed * 3)) & 0xFF) as u8)
    .collect()
}

// ===== YUYV422 =====

fn check_yuyv422_rgb(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 3];
  let mut n = std::vec![0u8; width * 3];
  scalar::yuyv422_to_rgb_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    yuyv422_to_rgb_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON yuyv422→RGB diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

fn check_yuyv422_rgba(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 4];
  let mut n = std::vec![0u8; width * 4];
  scalar::yuyv422_to_rgba_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    yuyv422_to_rgba_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON yuyv422→RGBA diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_yuyv422_rgb_matches_scalar_all_matrices() {
  for m in [
    ColorMatrix::Bt601,
    ColorMatrix::Bt709,
    ColorMatrix::Bt2020Ncl,
    ColorMatrix::Smpte240m,
    ColorMatrix::Fcc,
    ColorMatrix::YCgCo,
  ] {
    for full in [true, false] {
      check_yuyv422_rgb(16, m, full);
      check_yuyv422_rgba(16, m, full);
    }
  }
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_yuyv422_matches_scalar_widths() {
  for w in [2usize, 4, 14, 16, 18, 30, 32, 34, 62, 64, 66, 1920, 1922] {
    check_yuyv422_rgb(w, ColorMatrix::Bt709, false);
    check_yuyv422_rgba(w, ColorMatrix::Bt709, true);
  }
}

// ===== UYVY422 =====

fn check_uyvy422_rgb(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 3];
  let mut n = std::vec![0u8; width * 3];
  scalar::uyvy422_to_rgb_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    uyvy422_to_rgb_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON uyvy422→RGB diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

fn check_uyvy422_rgba(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 4];
  let mut n = std::vec![0u8; width * 4];
  scalar::uyvy422_to_rgba_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    uyvy422_to_rgba_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON uyvy422→RGBA diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_uyvy422_rgb_matches_scalar_all_matrices() {
  for m in [
    ColorMatrix::Bt601,
    ColorMatrix::Bt709,
    ColorMatrix::Bt2020Ncl,
    ColorMatrix::Smpte240m,
    ColorMatrix::Fcc,
    ColorMatrix::YCgCo,
  ] {
    for full in [true, false] {
      check_uyvy422_rgb(16, m, full);
      check_uyvy422_rgba(16, m, full);
    }
  }
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_uyvy422_matches_scalar_widths() {
  for w in [2usize, 4, 14, 16, 18, 30, 32, 34, 62, 64, 66, 1920, 1922] {
    check_uyvy422_rgb(w, ColorMatrix::Bt709, false);
    check_uyvy422_rgba(w, ColorMatrix::Bt709, true);
  }
}

// ===== YVYU422 =====

fn check_yvyu422_rgb(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 3];
  let mut n = std::vec![0u8; width * 3];
  scalar::yvyu422_to_rgb_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    yvyu422_to_rgb_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON yvyu422→RGB diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

fn check_yvyu422_rgba(width: usize, matrix: ColorMatrix, full_range: bool) {
  let p = packed_yuv422_buffer(width, 37);
  let mut s = std::vec![0u8; width * 4];
  let mut n = std::vec![0u8; width * 4];
  scalar::yvyu422_to_rgba_row(&p, &mut s, width, matrix, full_range);
  unsafe {
    yvyu422_to_rgba_row(&p, &mut n, width, matrix, full_range);
  }
  assert_eq!(
    s, n,
    "NEON yvyu422→RGBA diverges (width={width}, matrix={matrix:?}, full_range={full_range})"
  );
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_yvyu422_rgb_matches_scalar_all_matrices() {
  for m in [
    ColorMatrix::Bt601,
    ColorMatrix::Bt709,
    ColorMatrix::Bt2020Ncl,
    ColorMatrix::Smpte240m,
    ColorMatrix::Fcc,
    ColorMatrix::YCgCo,
  ] {
    for full in [true, false] {
      check_yvyu422_rgb(16, m, full);
      check_yvyu422_rgba(16, m, full);
    }
  }
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_yvyu422_matches_scalar_widths() {
  for w in [2usize, 4, 14, 16, 18, 30, 32, 34, 62, 64, 66, 1920, 1922] {
    check_yvyu422_rgb(w, ColorMatrix::Bt709, false);
    check_yvyu422_rgba(w, ColorMatrix::Bt709, true);
  }
}

// ===== Luma extraction (3 formats; YUYV / UYVY / YVYU) =====

fn check_luma(width: usize, format: char) {
  let p = packed_yuv422_buffer(width, 53);
  let mut s = std::vec![0u8; width];
  let mut n = std::vec![0u8; width];
  match format {
    'y' => unsafe {
      scalar::yuyv422_to_luma_row(&p, &mut s, width);
      yuyv422_to_luma_row(&p, &mut n, width);
    },
    'u' => unsafe {
      scalar::uyvy422_to_luma_row(&p, &mut s, width);
      uyvy422_to_luma_row(&p, &mut n, width);
    },
    'v' => unsafe {
      scalar::yvyu422_to_luma_row(&p, &mut s, width);
      yvyu422_to_luma_row(&p, &mut n, width);
    },
    _ => unreachable!(),
  }
  assert_eq!(s, n, "NEON {format}xyz luma diverges (width={width})");
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_packed_yuv422_luma_matches_scalar_widths() {
  for w in [2usize, 4, 14, 16, 18, 30, 32, 34, 62, 64, 66, 1920, 1922] {
    check_luma(w, 'y');
    check_luma(w, 'u');
    check_luma(w, 'v');
  }
}

// ===== u16 luma extraction (3 formats; YUYV / UYVY / YVYU) =====

fn check_luma_u16(width: usize, format: char) {
  let p = packed_yuv422_buffer(width, 71);
  let mut s = std::vec![0u16; width];
  let mut n = std::vec![0u16; width];
  match format {
    'y' => unsafe {
      scalar::yuyv422_to_luma_u16_row(&p, &mut s, width);
      yuyv422_to_luma_u16_row(&p, &mut n, width);
    },
    'u' => unsafe {
      scalar::uyvy422_to_luma_u16_row(&p, &mut s, width);
      uyvy422_to_luma_u16_row(&p, &mut n, width);
    },
    'v' => unsafe {
      scalar::yvyu422_to_luma_u16_row(&p, &mut s, width);
      yvyu422_to_luma_u16_row(&p, &mut n, width);
    },
    _ => unreachable!(),
  }
  assert_eq!(s, n, "NEON {format}xyz luma_u16 diverges (width={width})");
}

#[test]
#[cfg_attr(miri, ignore = "NEON SIMD intrinsics unsupported by Miri")]
fn neon_packed_yuv422_luma_u16_matches_scalar_widths() {
  // Sweep covers the 16-px NEON main loop boundary plus tails (15 / 17 / 31 / 33).
  for w in [
    1usize, 2, 4, 8, 14, 15, 16, 17, 18, 30, 31, 32, 33, 34, 62, 63, 64, 66, 1920, 1922,
  ] {
    check_luma_u16(w, 'y');
    check_luma_u16(w, 'u');
    check_luma_u16(w, 'v');
  }
}