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
//! RGBA framebuffer — the rendering target for the entire widget tree.
//!
//! # Memory layout
//!
//! Pixels are stored in **bottom-up row order** (Y-up): row 0 is at the start
//! of the `Vec<u8>` and corresponds to the bottom edge of the image (Y = 0).
//! Row `height - 1` is at the end and corresponds to the top edge.
//!
//! This matches OpenGL's texture layout: `glTexImage2D` treats the first byte
//! as the bottom-left pixel, so the buffer can be uploaded directly without
//! any Y-flip at the GL boundary.
//!
//! For display targets that use a top-down layout (e.g. HTML Canvas
//! `putImageData`), use [`Framebuffer::pixels_flipped`] to obtain a copy with
//! rows reversed.
/// RGBA framebuffer with bottom-up (Y-up) row ordering.
pub struct Framebuffer {
pixels: Vec<u8>, // RGBA8, row 0 = bottom (Y = 0)
width: u32,
height: u32,
}
impl Framebuffer {
/// Create a new zeroed framebuffer.
pub fn new(width: u32, height: u32) -> Self {
Self {
pixels: vec![0u8; (width * height * 4) as usize],
width,
height,
}
}
pub fn width(&self) -> u32 {
self.width
}
pub fn height(&self) -> u32 {
self.height
}
/// Raw RGBA8 pixels in bottom-up row order. Row 0 = Y=0 = bottom of image.
/// Upload directly to OpenGL without modification.
pub fn pixels(&self) -> &[u8] {
&self.pixels
}
/// Mutable access to the raw pixel data.
pub fn pixels_mut(&mut self) -> &mut [u8] {
&mut self.pixels
}
/// Consume the framebuffer and return ownership of the underlying
/// `Vec<u8>`. Used by widget backbuffer caching to hand the bitmap
/// off to an `Arc` without copying.
pub fn into_pixels(self) -> Vec<u8> {
self.pixels
}
/// Resize the framebuffer (pixels are zeroed on resize).
pub fn resize(&mut self, width: u32, height: u32) {
if self.width != width || self.height != height {
self.width = width;
self.height = height;
self.pixels = vec![0u8; (width * height * 4) as usize];
}
}
/// Return a copy of the pixels with rows reversed (top-down / Y-down order).
///
/// Use this for HTML Canvas `putImageData`, which expects the first row to
/// be the top of the image.
pub fn pixels_flipped(&self) -> Vec<u8> {
let row_bytes = (self.width * 4) as usize;
let mut flipped = vec![0u8; self.pixels.len()];
for y in 0..self.height as usize {
let src = (self.height as usize - 1 - y) * row_bytes;
let dst = y * row_bytes;
flipped[dst..dst + row_bytes].copy_from_slice(&self.pixels[src..src + row_bytes]);
}
flipped
}
}
// ---------------------------------------------------------------------------
// Alpha conversion helpers
// ---------------------------------------------------------------------------
//
// The AGG rasterizer writes **premultiplied** RGBA into its framebuffer. The
// `DrawCtx::draw_image_rgba` API, in contrast, takes **straight-alpha** input
// (PNG/markdown images, screenshots, etc.). These helpers convert between
// the two representations at the boundary — Label un-premultiplies its AGG
// backbuffer before handing it to `draw_image_rgba`, and the software
// compositor premultiplies straight input before writing it back into an AGG
// framebuffer for blending.
/// Convert an RGBA8 buffer from **premultiplied** to **straight** alpha
/// in place. For each pixel, divides the colour channels by the alpha.
/// Zero-alpha pixels are zeroed out (there is no colour information to recover).
pub fn unpremultiply_rgba_inplace(data: &mut [u8]) {
for px in data.chunks_exact_mut(4) {
let a = px[3];
if a == 0 {
px[0] = 0;
px[1] = 0;
px[2] = 0;
} else if a < 255 {
let af = a as u32;
// Round-half-up division: (c * 255 + a/2) / a.
px[0] = (((px[0] as u32) * 255 + af / 2) / af).min(255) as u8;
px[1] = (((px[1] as u32) * 255 + af / 2) / af).min(255) as u8;
px[2] = (((px[2] as u32) * 255 + af / 2) / af).min(255) as u8;
}
}
}
/// Convert an RGBA8 buffer from **straight** to **premultiplied** alpha
/// in place. Each colour channel is multiplied by the alpha.
pub fn premultiply_rgba_inplace(data: &mut [u8]) {
for px in data.chunks_exact_mut(4) {
let a = px[3] as u32;
if a < 255 {
// Round-half-up: (c * a + 127) / 255.
px[0] = (((px[0] as u32) * a + 127) / 255) as u8;
px[1] = (((px[1] as u32) * a + 127) / 255) as u8;
px[2] = (((px[2] as u32) * a + 127) / 255) as u8;
}
}
}
#[cfg(test)]
mod alpha_tests {
use super::*;
/// Round-trip premul → unpremul on a half-alpha white pixel must recover
/// the original colour channels (within 1/255 rounding error).
#[test]
fn test_premul_roundtrip_half_alpha_white() {
// Straight white at 50 % opacity: (255, 255, 255, 128).
let mut px = [255, 255, 255, 128];
premultiply_rgba_inplace(&mut px);
// Premultiplied: (128, 128, 128, 128).
assert_eq!(px, [128, 128, 128, 128]);
unpremultiply_rgba_inplace(&mut px);
// Round-tripped: should recover (255, 255, 255, 128) exactly.
assert_eq!(px, [255, 255, 255, 128]);
}
/// Fully-opaque pixels are unchanged in both directions.
#[test]
fn test_premul_opaque_unchanged() {
let mut px = [60, 120, 240, 255];
premultiply_rgba_inplace(&mut px);
assert_eq!(px, [60, 120, 240, 255]);
unpremultiply_rgba_inplace(&mut px);
assert_eq!(px, [60, 120, 240, 255]);
}
/// Fully-transparent pixels zero out their colour when un-premultiplied.
#[test]
fn test_unpremul_zero_alpha_zeros_colour() {
let mut px = [10, 20, 30, 0];
unpremultiply_rgba_inplace(&mut px);
assert_eq!(px, [0, 0, 0, 0]);
}
}