ctt 0.3.0

Compress images to GPU texture formats
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

//! Per-[`FormatKind`](crate::format_kind::FormatKind) encoders that write a
//! `Buffer<T>` back into a [`Surface`]'s byte vector.
//!
//! Input lanes are in *linear*, *premultiplied* space (for float pipelines);
//! unpremultiplication and sRGB re-encoding happen at the callers / here.

mod srgb;
pub use srgb::{store_bgr8_srgb_f32, store_bgra8_srgb_f32, store_srgb8_f32};

use half::f16;

use super::buffer::Buffer;

// ---- f32 pipeline stores ----

pub fn store_u8_unorm_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u8_unorm_f32");
    write_pixels(buf, channels, 1, |lanes, bytes| {
        for (&lane, byte) in lanes.iter().zip(bytes.iter_mut()) {
            *byte = (lane.clamp(0.0, 1.0) * 255.0).round() as u8;
        }
    })
}

pub fn store_i8_snorm_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i8_snorm_f32");
    write_pixels(buf, channels, 1, |lanes, bytes| {
        for (&lane, byte) in lanes.iter().zip(bytes.iter_mut()) {
            let v = (lane.clamp(-1.0, 1.0) * 127.0).round() as i8;
            *byte = v as u8;
        }
    })
}

pub fn store_bgra8_unorm_f32(buf: &Buffer<f32>) -> Vec<u8> {
    profiling::scope!("store_bgra8_unorm_f32");
    write_pixels(buf, 4, 1, |lanes, bytes| {
        let arr = <&mut [u8; 4]>::try_from(bytes).expect("4-byte pixel");
        arr[0] = (lanes[2].clamp(0.0, 1.0) * 255.0).round() as u8; // B
        arr[1] = (lanes[1].clamp(0.0, 1.0) * 255.0).round() as u8; // G
        arr[2] = (lanes[0].clamp(0.0, 1.0) * 255.0).round() as u8; // R
        arr[3] = (lanes[3].clamp(0.0, 1.0) * 255.0).round() as u8; // A
    })
}

pub fn store_bgr8_unorm_f32(buf: &Buffer<f32>) -> Vec<u8> {
    profiling::scope!("store_bgr8_unorm_f32");
    write_pixels(buf, 3, 1, |lanes, bytes| {
        let arr = <&mut [u8; 3]>::try_from(bytes).expect("3-byte pixel");
        arr[0] = (lanes[2].clamp(0.0, 1.0) * 255.0).round() as u8;
        arr[1] = (lanes[1].clamp(0.0, 1.0) * 255.0).round() as u8;
        arr[2] = (lanes[0].clamp(0.0, 1.0) * 255.0).round() as u8;
    })
}

pub fn store_u16_unorm_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u16_unorm_f32");
    write_pixels(buf, channels, 2, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<2>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            let v = (lane.clamp(0.0, 1.0) * 65535.0).round() as u16;
            *chunk = v.to_le_bytes();
        }
    })
}

pub fn store_i16_snorm_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i16_snorm_f32");
    write_pixels(buf, channels, 2, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<2>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            let v = (lane.clamp(-1.0, 1.0) * 32767.0).round() as i16;
            *chunk = v.to_le_bytes();
        }
    })
}

pub fn store_f16_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_f16_f32");
    write_pixels(buf, channels, 2, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<2>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = f16::from_f32(lane).to_le_bytes();
        }
    })
}

pub fn store_f32_f32(buf: &Buffer<f32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_f32_f32");
    write_pixels(buf, channels, 4, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<4>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = lane.to_le_bytes();
        }
    })
}

// ---- f64 pipeline stores ----

pub fn store_f32_f64(buf: &Buffer<f64>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_f32_f64");
    write_pixels_f64(buf, channels, 4, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<4>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = (lane as f32).to_le_bytes();
        }
    })
}

pub fn store_f64_f64(buf: &Buffer<f64>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_f64_f64");
    write_pixels_f64(buf, channels, 8, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<8>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = lane.to_le_bytes();
        }
    })
}

// ---- u32 pipeline stores ----

pub fn store_u8_uint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u8_uint_u32");
    write_pixels_u32(buf, channels, 1, |lanes, bytes| {
        for (&lane, byte) in lanes.iter().zip(bytes.iter_mut()) {
            *byte = lane.min(u8::MAX as u32) as u8;
        }
    })
}

pub fn store_i8_sint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i8_sint_u32");
    write_pixels_u32(buf, channels, 1, |lanes, bytes| {
        for (&lane, byte) in lanes.iter().zip(bytes.iter_mut()) {
            let v = (lane as i32).clamp(i8::MIN as i32, i8::MAX as i32) as i8;
            *byte = v as u8;
        }
    })
}

pub fn store_u16_uint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u16_uint_u32");
    write_pixels_u32(buf, channels, 2, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<2>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            let v = lane.min(u16::MAX as u32) as u16;
            *chunk = v.to_le_bytes();
        }
    })
}

pub fn store_i16_sint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i16_sint_u32");
    write_pixels_u32(buf, channels, 2, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<2>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            let v = (lane as i32).clamp(i16::MIN as i32, i16::MAX as i32) as i16;
            *chunk = v.to_le_bytes();
        }
    })
}

pub fn store_u32_uint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u32_uint_u32");
    write_pixels_u32(buf, channels, 4, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<4>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = lane.to_le_bytes();
        }
    })
}

pub fn store_i32_sint_u32(buf: &Buffer<u32>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i32_sint_u32");
    write_pixels_u32(buf, channels, 4, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<4>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            // bit-cast u32 → i32 → le_bytes (same bytes either way).
            *chunk = lane.to_le_bytes();
        }
    })
}

// ---- u64 pipeline stores ----

pub fn store_u64_uint_u64(buf: &Buffer<u64>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_u64_uint_u64");
    write_pixels_u64(buf, channels, 8, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<8>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = lane.to_le_bytes();
        }
    })
}

pub fn store_i64_sint_u64(buf: &Buffer<u64>, channels: usize) -> Vec<u8> {
    profiling::scope!("store_i64_sint_u64");
    write_pixels_u64(buf, channels, 8, |lanes, bytes| {
        let (chunks, _) = bytes.as_chunks_mut::<8>();
        for (&lane, chunk) in lanes.iter().zip(chunks) {
            *chunk = lane.to_le_bytes();
        }
    })
}

// ---- Helpers ----

fn write_pixels(
    buf: &Buffer<f32>,
    channels: usize,
    channel_bytes: usize,
    mut encode: impl FnMut(&[f32; 4], &mut [u8]),
) -> Vec<u8> {
    let pixel_bytes = channels * channel_bytes;
    let mut out = vec![0u8; buf.pixels.len() * pixel_bytes];
    for (pixel, bytes) in buf.pixels.iter().zip(out.chunks_exact_mut(pixel_bytes)) {
        encode(pixel, bytes);
    }
    out
}

fn write_pixels_f64(
    buf: &Buffer<f64>,
    channels: usize,
    channel_bytes: usize,
    mut encode: impl FnMut(&[f64; 4], &mut [u8]),
) -> Vec<u8> {
    let pixel_bytes = channels * channel_bytes;
    let mut out = vec![0u8; buf.pixels.len() * pixel_bytes];
    for (pixel, bytes) in buf.pixels.iter().zip(out.chunks_exact_mut(pixel_bytes)) {
        encode(pixel, bytes);
    }
    out
}

fn write_pixels_u32(
    buf: &Buffer<u32>,
    channels: usize,
    channel_bytes: usize,
    mut encode: impl FnMut(&[u32; 4], &mut [u8]),
) -> Vec<u8> {
    let pixel_bytes = channels * channel_bytes;
    let mut out = vec![0u8; buf.pixels.len() * pixel_bytes];
    for (pixel, bytes) in buf.pixels.iter().zip(out.chunks_exact_mut(pixel_bytes)) {
        encode(pixel, bytes);
    }
    out
}

fn write_pixels_u64(
    buf: &Buffer<u64>,
    channels: usize,
    channel_bytes: usize,
    mut encode: impl FnMut(&[u64; 4], &mut [u8]),
) -> Vec<u8> {
    let pixel_bytes = channels * channel_bytes;
    let mut out = vec![0u8; buf.pixels.len() * pixel_bytes];
    for (pixel, bytes) in buf.pixels.iter().zip(out.chunks_exact_mut(pixel_bytes)) {
        encode(pixel, bytes);
    }
    out
}