Dithering Mask generator
Implementation of Mask generation - "Blue-noise dithered sampling", Georgiev and Fajardo, 2016
Note that for 1D-large mask, it is better to use Jeremy Dupuy's implementation: https://github.com/jdupuy/BlueNoiseDitherMaskTiles His implementation uses cluster-and-void and is GPU accelerated.
Here this code can compute arbitrary mask dimension on CPU. The code use cluster-and-void for 1D approaches. For higher dimension, a simple anneling method with auto-tuning temperature is used.
Example to generate 32x32 masks of different dimensions:
cargo run --release -- 64 1 mask_d1
cargo run --release -- 64 2 mask_d2
cargo run --release -- 64 3 mask_d3
Example output:
| Dim 1 | Dim 2 | Dim 3 |
|---|---|---|
 |
 |
 |
The command also output .mask file. This file contains an header flag: MASK\n {size} {size} {dimension}\n-1.0\n then the floating point value of the mask. This simple format is easy to integrate inside your own renderer.
let header = format!("MASK\n{} {} {}\n-1.0\n", size, size, dimension);
file.write_all(header.as_bytes()).unwrap();
for d in 0..dimension {
for y in 0..size {
for x in 0..size {
let p = img[d as usize][get_index((x, y))];
file.write_f32::<LittleEndian>(p.abs()).unwrap();
}
}
}