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

//! 8-bit YUV 4:4:4 → RGB / RGBA dispatchers (`yuv_444_to_rgb_row`,
//! `yuv_444_to_rgba_row`). Extracted from the parent
//! `dispatch::yuv444` module per source format for organization.

#[cfg(any(
  target_arch = "aarch64",
  target_arch = "x86_64",
  target_arch = "wasm32"
))]
use crate::row::arch;
#[cfg(target_arch = "aarch64")]
use crate::row::neon_available;
#[cfg(target_arch = "wasm32")]
use crate::row::simd128_available;
#[cfg(target_arch = "x86_64")]
use crate::row::{avx2_available, avx512_available, sse41_available};
use crate::{
  ColorMatrix,
  row::{rgb_row_bytes, rgba_row_bytes, scalar},
};

/// Converts one row of YUV 4:4:4 planar to packed RGB. Dispatches
/// to the best available SIMD backend for the current target.
///
/// Same numerical contract as [`yuv_420_to_rgb_row`]; the difference
/// is 4:4:4 chroma — one U / V pair per Y pixel, full-width chroma
/// planes, no chroma upsampling, no width parity constraint. See
/// `scalar::yuv_444_to_rgb_row` for the reference implementation.
///
/// `use_simd = false` forces the scalar reference path.
#[cfg_attr(not(tarpaulin), inline(always))]
#[allow(clippy::too_many_arguments)]
pub fn yuv_444_to_rgb_row(
  y: &[u8],
  u: &[u8],
  v: &[u8],
  rgb_out: &mut [u8],
  width: usize,
  matrix: ColorMatrix,
  full_range: bool,
  use_simd: bool,
) {
  let rgb_min = rgb_row_bytes(width);
  assert!(y.len() >= width, "y row too short");
  assert!(u.len() >= width, "u row too short");
  assert!(v.len() >= width, "v row too short");
  assert!(rgb_out.len() >= rgb_min, "rgb_out row too short");

  if use_simd {
    cfg_select! {
      target_arch = "aarch64" => {
        if neon_available() {
          // SAFETY: `neon_available()` verified NEON is present.
          unsafe {
            arch::neon::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
          }
          return;
        }
      },
      target_arch = "x86_64" => {
        if avx512_available() {
          // SAFETY: AVX-512BW verified.
          unsafe {
            arch::x86_avx512::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
          }
          return;
        }
        if avx2_available() {
          // SAFETY: AVX2 verified.
          unsafe {
            arch::x86_avx2::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
          }
          return;
        }
        if sse41_available() {
          // SAFETY: SSE4.1 verified.
          unsafe {
            arch::x86_sse41::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
          }
          return;
        }
      },
      target_arch = "wasm32" => {
        if simd128_available() {
          // SAFETY: simd128 verified at compile time.
          unsafe {
            arch::wasm_simd128::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
          }
          return;
        }
      },
      _ => {}
    }
  }

  scalar::yuv_444_to_rgb_row(y, u, v, rgb_out, width, matrix, full_range);
}

/// Converts one row of YUV 4:4:4 planar to packed **RGBA** (8-bit).
/// Same numerical contract as [`yuv_444_to_rgb_row`]; the only
/// differences are the per-pixel stride (4 vs 3) and the alpha byte
/// (`0xFF`, opaque, for every pixel). `rgba_out.len() >= 4 * width`.
/// `use_simd = false` forces scalar.
#[cfg_attr(not(tarpaulin), inline(always))]
#[allow(clippy::too_many_arguments)]
pub fn yuv_444_to_rgba_row(
  y: &[u8],
  u: &[u8],
  v: &[u8],
  rgba_out: &mut [u8],
  width: usize,
  matrix: ColorMatrix,
  full_range: bool,
  use_simd: bool,
) {
  let rgba_min = rgba_row_bytes(width);
  assert!(y.len() >= width, "y row too short");
  assert!(u.len() >= width, "u row too short");
  assert!(v.len() >= width, "v row too short");
  assert!(rgba_out.len() >= rgba_min, "rgba_out row too short");

  if use_simd {
    cfg_select! {
      target_arch = "aarch64" => {
        if neon_available() {
          unsafe {
            arch::neon::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
          }
          return;
        }
      },
      target_arch = "x86_64" => {
        if avx512_available() {
          unsafe {
            arch::x86_avx512::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
          }
          return;
        }
        if avx2_available() {
          unsafe {
            arch::x86_avx2::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
          }
          return;
        }
        if sse41_available() {
          unsafe {
            arch::x86_sse41::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
          }
          return;
        }
      },
      target_arch = "wasm32" => {
        if simd128_available() {
          unsafe {
            arch::wasm_simd128::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
          }
          return;
        }
      },
      _ => {}
    }
  }

  scalar::yuv_444_to_rgba_row(y, u, v, rgba_out, width, matrix, full_range);
}