Skip to content

Add kernelUnitLength support to feDiffuseLighting, feSpecularLighting, and feConvolveMatrix #1127

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
edco wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add kernelUnitLength support to feDiffuseLighting, feSpecularLighting, and feConvolveMatrix #1127

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .gitignore
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,3 +4,4 @@ target
.vscode
tools/build-*
**/diffs
.cargo
12 changes: 0 additions & 12 deletions Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

138 changes: 108 additions & 30 deletions crates/resvg/src/filter/convolve_matrix.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,23 +3,63 @@

use super::{ImageRefMut, f32_bound};
use rgb::RGBA8;
use usvg::ApproxEqUlps;
use usvg::filter::{ConvolveMatrix, EdgeMode};

#[inline]
fn pixel_at_f32(src: &ImageRefMut, fx: f32, fy: f32) -> (f32, f32, f32, f32) {
let fx = fx.clamp(0.0, (src.width - 1) as f32);
let fy = fy.clamp(0.0, (src.height - 1) as f32);
let x0 = fx.floor() as u32;
let y0 = fy.floor() as u32;
let x1 = (x0 + 1).min(src.width - 1);
let y1 = (y0 + 1).min(src.height - 1);
let dx = fx - x0 as f32;
let dy = fy - y0 as f32;

let p00 = src.pixel_at(x0, y0);
let p10 = src.pixel_at(x1, y0);
let p01 = src.pixel_at(x0, y1);
let p11 = src.pixel_at(x1, y1);

let interp = |c00: u8, c10: u8, c01: u8, c11: u8| -> f32 {
let top = c00 as f32 * (1.0 - dx) + c10 as f32 * dx;
let bottom = c01 as f32 * (1.0 - dx) + c11 as f32 * dx;
top * (1.0 - dy) + bottom * dy
};

(
interp(p00.r, p10.r, p01.r, p11.r),
interp(p00.g, p10.g, p01.g, p11.g),
interp(p00.b, p10.b, p01.b, p11.b),
interp(p00.a, p10.a, p01.a, p11.a),
)
}

/// Applies a convolve matrix.
///
/// Input image pixels should have a **premultiplied alpha** when `preserve_alpha=false`.
///
/// # Allocations
///
/// This method will allocate a copy of the `src` image as a back buffer.
pub fn apply(matrix: &ConvolveMatrix, src: ImageRefMut) {
pub fn apply(matrix: &ConvolveMatrix, ts: usvg::Transform, src: ImageRefMut) {
fn bound(min: i32, val: i32, max: i32) -> i32 {
core::cmp::max(min, core::cmp::min(max, val))
}

let width_max = src.width as i32 - 1;
let height_max = src.height as i32 - 1;

let (step_x, step_y) = if let Some((kx, ky)) = matrix.kernel_unit_length() {
let (sx, sy) = ts.get_scale();
(kx.get() * sx, ky.get() * sy)
} else {
(1.0, 1.0)
};

let use_unit_step = step_x.approx_eq_ulps(&1.0, 4) && step_y.approx_eq_ulps(&1.0, 4);

let mut buf = vec![RGBA8::default(); src.data.len()];
let mut buf = ImageRefMut::new(src.width, src.height, &mut buf);
let mut x = 0;
Expand All @@ -31,44 +71,82 @@ pub fn apply(matrix: &ConvolveMatrix, src: ImageRefMut) {
let mut new_a = 0.0;
for oy in 0..matrix.matrix().rows() {
for ox in 0..matrix.matrix().columns() {
let mut tx = x as i32 - matrix.matrix().target_x() as i32 + ox as i32;
let mut ty = y as i32 - matrix.matrix().target_y() as i32 + oy as i32;
let k = matrix.matrix().get(
matrix.matrix().columns() - ox - 1,
matrix.matrix().rows() - oy - 1,
);

if use_unit_step {
let mut tx = x as i32 - matrix.matrix().target_x() as i32 + ox as i32;
let mut ty = y as i32 - matrix.matrix().target_y() as i32 + oy as i32;

match matrix.edge_mode() {
EdgeMode::None => {
if tx < 0 || tx > width_max || ty < 0 || ty > height_max {
continue;
match matrix.edge_mode() {
EdgeMode::None => {
if tx < 0 || tx > width_max || ty < 0 || ty > height_max {
continue;
}
}
EdgeMode::Duplicate => {
tx = bound(0, tx, width_max);
ty = bound(0, ty, height_max);
}
EdgeMode::Wrap => {
while tx < 0 {
tx += src.width as i32;
}
tx %= src.width as i32;

while ty < 0 {
ty += src.height as i32;
}
ty %= src.height as i32;
}
}
EdgeMode::Duplicate => {
tx = bound(0, tx, width_max);
ty = bound(0, ty, height_max);

let p = src.pixel_at(tx as u32, ty as u32);
new_r += (p.r as f32) / 255.0 * k;
new_g += (p.g as f32) / 255.0 * k;
new_b += (p.b as f32) / 255.0 * k;

if !matrix.preserve_alpha() {
new_a += (p.a as f32) / 255.0 * k;
}
EdgeMode::Wrap => {
while tx < 0 {
tx += src.width as i32;
} else {
let mut fx =
x as f32 - matrix.matrix().target_x() as f32 * step_x + ox as f32 * step_x;
let mut fy =
y as f32 - matrix.matrix().target_y() as f32 * step_y + oy as f32 * step_y;

match matrix.edge_mode() {
EdgeMode::None => {
if fx < 0.0
|| fx > width_max as f32
|| fy < 0.0
|| fy > height_max as f32
{
continue;
}
}
tx %= src.width as i32;

while ty < 0 {
ty += src.height as i32;
EdgeMode::Duplicate => {
fx = fx.clamp(0.0, width_max as f32);
fy = fy.clamp(0.0, height_max as f32);
}
EdgeMode::Wrap => {
let w = src.width as f32;
let h = src.height as f32;
fx = fx.rem_euclid(w);
fy = fy.rem_euclid(h);
}
ty %= src.height as i32;
}
}

let k = matrix.matrix().get(
matrix.matrix().columns() - ox - 1,
matrix.matrix().rows() - oy - 1,
);
let (pr, pg, pb, pa) = pixel_at_f32(&src, fx, fy);
new_r += (pr / 255.0) * k;
new_g += (pg / 255.0) * k;
new_b += (pb / 255.0) * k;

let p = src.pixel_at(tx as u32, ty as u32);
new_r += (p.r as f32) / 255.0 * k;
new_g += (p.g as f32) / 255.0 * k;
new_b += (p.b as f32) / 255.0 * k;

if !matrix.preserve_alpha() {
new_a += (p.a as f32) / 255.0 * k;
if !matrix.preserve_alpha() {
new_a += (pa / 255.0) * k;
}
}
}
}
Expand Down
Loading
Loading