oxihuman-mesh 0.2.1

Mesh processing, topology, and geometry algorithms for OxiHuman
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

//! Relaxation-based vertex smoothing.
#![allow(dead_code)]

/// Configuration for mesh relaxation.
#[allow(dead_code)]
pub struct RelaxConfig {
    pub iterations: usize,
    pub lambda: f32,
    pub boundary_only: bool,
}

impl Default for RelaxConfig {
    fn default() -> Self {
        RelaxConfig { iterations: 5, lambda: 0.5, boundary_only: false }
    }
}

/// Relax a mesh by averaging vertex positions with neighbors.
#[allow(dead_code)]
pub fn relax_mesh(positions: &[[f32;3]], indices: &[u32], config: &RelaxConfig) -> Vec<[f32;3]> {
    relax_n_steps(positions, indices, config.iterations, config.lambda)
}

/// Perform one relaxation step.
#[allow(dead_code)]
pub fn relax_step(positions: &[[f32;3]], adjacency: &[Vec<usize>], lambda: f32) -> Vec<[f32;3]> {
    let n = positions.len();
    let mut result = positions.to_vec();
    for i in 0..n {
        let neighbors = &adjacency[i];
        if neighbors.is_empty() { continue; }
        let cnt = neighbors.len() as f32;
        let sum: [f32;3] = neighbors.iter().fold([0.0;3], |acc, &j| {
            [acc[0]+positions[j][0], acc[1]+positions[j][1], acc[2]+positions[j][2]]
        });
        result[i][0] = positions[i][0] + lambda * (sum[0]/cnt - positions[i][0]);
        result[i][1] = positions[i][1] + lambda * (sum[1]/cnt - positions[i][1]);
        result[i][2] = positions[i][2] + lambda * (sum[2]/cnt - positions[i][2]);
    }
    result
}

/// Build vertex adjacency list from triangle indices.
#[allow(dead_code)]
pub fn build_vertex_adjacency2(n: usize, indices: &[u32]) -> Vec<Vec<usize>> {
    let mut adj: Vec<std::collections::BTreeSet<usize>> = vec![std::collections::BTreeSet::new(); n];
    let tris = indices.len() / 3;
    for t in 0..tris {
        let i0 = indices[t*3] as usize;
        let i1 = indices[t*3+1] as usize;
        let i2 = indices[t*3+2] as usize;
        if i0 < n && i1 < n && i2 < n {
            adj[i0].insert(i1); adj[i0].insert(i2);
            adj[i1].insert(i0); adj[i1].insert(i2);
            adj[i2].insert(i0); adj[i2].insert(i1);
        }
    }
    adj.into_iter().map(|s| s.into_iter().collect()).collect()
}

/// Compute a weighted Laplacian position for a vertex.
#[allow(dead_code)]
pub fn weighted_laplacian(
    vertex: usize,
    positions: &[[f32;3]],
    neighbors: &[usize],
    weights: &[f32],
) -> [f32;3] {
    if neighbors.is_empty() { return positions[vertex]; }
    let mut sum = [0.0f32;3];
    let mut wsum = 0.0f32;
    for (&n, &w) in neighbors.iter().zip(weights.iter()) {
        if n < positions.len() {
            sum[0] += positions[n][0] * w;
            sum[1] += positions[n][1] * w;
            sum[2] += positions[n][2] * w;
            wsum += w;
        }
    }
    if wsum < 1e-10 { return positions[vertex]; }
    [sum[0]/wsum, sum[1]/wsum, sum[2]/wsum]
}

/// Relax only boundary vertices.
#[allow(dead_code)]
pub fn relax_boundary_only(
    positions: &[[f32;3]],
    indices: &[u32],
    lambda: f32,
    iterations: usize,
) -> Vec<[f32;3]> {
    let n = positions.len();
    let adj = build_vertex_adjacency2(n, indices);
    // detect boundary (vertices with fewer connections)
    let avg_degree = if !adj.is_empty() { adj.iter().map(|a| a.len()).sum::<usize>() / adj.len() } else { 0 };
    let boundary: Vec<bool> = adj.iter().map(|a| a.len() < avg_degree).collect();
    let mut pos = positions.to_vec();
    for _ in 0..iterations {
        let prev = pos.clone();
        for i in 0..n {
            if !boundary[i] { continue; }
            if adj[i].is_empty() { continue; }
            let cnt = adj[i].len() as f32;
            let sum = adj[i].iter().fold([0.0f32;3], |acc, &j| {
                [acc[0]+prev[j][0], acc[1]+prev[j][1], acc[2]+prev[j][2]]
            });
            pos[i][0] = prev[i][0] + lambda * (sum[0]/cnt - prev[i][0]);
            pos[i][1] = prev[i][1] + lambda * (sum[1]/cnt - prev[i][1]);
            pos[i][2] = prev[i][2] + lambda * (sum[2]/cnt - prev[i][2]);
        }
    }
    pos
}

/// Relax a mesh for N steps.
#[allow(dead_code)]
pub fn relax_n_steps(
    positions: &[[f32;3]],
    indices: &[u32],
    n_steps: usize,
    lambda: f32,
) -> Vec<[f32;3]> {
    let n = positions.len();
    let adj = build_vertex_adjacency2(n, indices);
    let mut pos = positions.to_vec();
    for _ in 0..n_steps {
        pos = relax_step(&pos, &adj, lambda);
    }
    pos
}

#[cfg(test)]
mod tests {
    use super::*;

    fn square_mesh() -> (Vec<[f32;3]>, Vec<u32>) {
        let pos = vec![[0.0f32,0.0,0.0],[1.0,0.0,0.0],[0.0,1.0,0.0],[1.0,1.0,0.0]];
        let idx = vec![0u32,1,2, 1,3,2];
        (pos, idx)
    }

    #[test]
    fn test_relax_mesh_count() {
        let (pos, idx) = square_mesh();
        let cfg = RelaxConfig::default();
        let r = relax_mesh(&pos, &idx, &cfg);
        assert_eq!(r.len(), 4);
    }

    #[test]
    fn test_build_vertex_adjacency2_count() {
        let (pos, idx) = square_mesh();
        let adj = build_vertex_adjacency2(pos.len(), &idx);
        assert_eq!(adj.len(), 4);
    }

    #[test]
    fn test_relax_step_moves_vertices() {
        let (pos, idx) = square_mesh();
        let adj = build_vertex_adjacency2(pos.len(), &idx);
        let r = relax_step(&pos, &adj, 0.5);
        assert_eq!(r.len(), 4);
    }

    #[test]
    fn test_weighted_laplacian() {
        let pos = vec![[0.0f32,0.0,0.0],[1.0,0.0,0.0],[0.0,1.0,0.0]];
        let neighbors = vec![1usize, 2];
        let weights = vec![1.0f32, 1.0];
        let r = weighted_laplacian(0, &pos, &neighbors, &weights);
        assert!((r[0] - 0.5).abs() < 1e-4);
    }

    #[test]
    fn test_relax_boundary_only_count() {
        let (pos, idx) = square_mesh();
        let r = relax_boundary_only(&pos, &idx, 0.5, 2);
        assert_eq!(r.len(), 4);
    }

    #[test]
    fn test_relax_n_steps_count() {
        let (pos, idx) = square_mesh();
        let r = relax_n_steps(&pos, &idx, 3, 0.5);
        assert_eq!(r.len(), 4);
    }

    #[test]
    fn test_relax_config_default() {
        let cfg = RelaxConfig::default();
        assert_eq!(cfg.iterations, 5);
        assert!(!cfg.boundary_only);
    }

    #[test]
    fn test_adjacency_has_neighbors() {
        let (pos, idx) = square_mesh();
        let adj = build_vertex_adjacency2(pos.len(), &idx);
        assert!(adj.iter().any(|a| !a.is_empty()));
    }

    #[test]
    fn test_relax_zero_lambda_no_change() {
        let (pos, idx) = square_mesh();
        let r = relax_n_steps(&pos, &idx, 3, 0.0);
        for (a, b) in pos.iter().zip(r.iter()) {
            assert!((a[0]-b[0]).abs() < 1e-5);
        }
    }
}