splatter 0.0.1

A gaussian splat render engine
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
263
264
265

use crate::{
    renderer::{DepthSorting, Renderer, Uniforms},
    utils::{transmute_slice, transmute_slice_mut},
};
use geometric_algebra::{ppga3d::Rotor, Signum};
use std::{
    convert::TryInto,
    fs::File,
    io::{BufRead, BufReader, Read, Seek, SeekFrom},
};

// Because of numerical precision issues in the sader we have to limit the excentricity of ellipsoids
const MAX_SIZE_VARIANCE: f32 = 1.5;

#[derive(Clone)]
#[repr(C)]
struct SerializedSplat {
    center: [f32; 3],
    n: [f32; 3],
    color: [f32; 3 * 16],
    alpha: f32,
    scale: [f32; 3],
    rotation: [f32; 4],
}

impl Default for SerializedSplat {
    fn default() -> Self {
        unsafe { std::mem::MaybeUninit::zeroed().assume_init() }
    }
}

pub(crate) type Splat = [f32; 60];

/// A point cloud of splats
pub struct Scene {
    pub(crate) compute_bind_groups: [wgpu::BindGroup; 4],
    pub(crate) render_bind_group: wgpu::BindGroup,
    pub(crate) splat_buffer: wgpu::Buffer,
    pub(crate) splat_positions: Vec<f32>,
    pub splat_count: usize,
}

impl Scene {
    /// Constructs a new [Scene] and allocates memory for it
    pub fn new(device: &wgpu::Device, renderer: &Renderer, mut splat_count: usize) -> Self {
        splat_count = splat_count.min(renderer.config.max_splat_count);
        let splat_buffer = device.create_buffer(&wgpu::BufferDescriptor {
            label: None,
            size: (splat_count * std::mem::size_of::<Splat>()) as u64,
            usage: wgpu::BufferUsages::STORAGE | wgpu::BufferUsages::COPY_DST,
            mapped_at_creation: false,
        });
        let uniform_bind_group_entry = wgpu::BindGroupEntry {
            binding: 0,
            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                buffer: &renderer.uniform_buffer,
                offset: 0,
                size: std::num::NonZeroU64::new(std::mem::size_of::<Uniforms>() as u64),
            }),
        };
        let sorting_bind_group_entry = wgpu::BindGroupEntry {
            binding: 2,
            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                buffer: &renderer.sorting_buffer,
                offset: 0,
                size: std::num::NonZeroU64::new(renderer.sorting_buffer_size as u64),
            }),
        };
        let splats_bind_group_entry = wgpu::BindGroupEntry {
            binding: 6,
            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                buffer: &splat_buffer,
                offset: 0,
                size: std::num::NonZeroU64::new((splat_count * std::mem::size_of::<Splat>()) as u64),
            }),
        };
        let compute_bind_groups: [wgpu::BindGroup; 4] = (0..4)
            .map(|pass_index| {
                device.create_bind_group(&wgpu::BindGroupDescriptor {
                    label: None,
                    layout: &renderer.compute_bind_group_layout,
                    entries: &[
                        uniform_bind_group_entry.clone(),
                        wgpu::BindGroupEntry {
                            binding: 1,
                            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                                buffer: &renderer.sorting_pass_buffers[pass_index],
                                offset: 0,
                                size: std::num::NonZeroU64::new(std::mem::size_of::<u32>() as u64),
                            }),
                        },
                        sorting_bind_group_entry.clone(),
                        wgpu::BindGroupEntry {
                            binding: 3,
                            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                                buffer: if pass_index & 1 == 0 {
                                    &renderer.entry_buffer_a
                                } else {
                                    &renderer.entry_buffer_b
                                },
                                offset: 0,
                                size: std::num::NonZeroU64::new((splat_count * std::mem::size_of::<(u32, u32)>()) as u64),
                            }),
                        },
                        wgpu::BindGroupEntry {
                            binding: 4,
                            resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                                buffer: if pass_index & 1 == 0 {
                                    &renderer.entry_buffer_b
                                } else {
                                    &renderer.entry_buffer_a
                                },
                                offset: 0,
                                size: std::num::NonZeroU64::new((splat_count * std::mem::size_of::<(u32, u32)>()) as u64),
                            }),
                        },
                        splats_bind_group_entry.clone(),
                    ],
                })
            })
            .collect::<Vec<wgpu::BindGroup>>()
            .try_into()
            .unwrap();
        let render_bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
            label: None,
            layout: &renderer.render_bind_group_layout,
            entries: &[
                uniform_bind_group_entry,
                wgpu::BindGroupEntry {
                    binding: 5,
                    resource: wgpu::BindingResource::Buffer(wgpu::BufferBinding {
                        buffer: if matches!(renderer.config.depth_sorting, DepthSorting::Cpu) || renderer.radix_digit_places & 1 == 0 {
                            &renderer.entry_buffer_a
                        } else {
                            &renderer.entry_buffer_b
                        },
                        offset: 0,
                        size: std::num::NonZeroU64::new((splat_count * std::mem::size_of::<(u32, u32)>()) as u64),
                    }),
                },
                splats_bind_group_entry,
            ],
        });
        Self {
            compute_bind_groups,
            render_bind_group,
            splat_buffer,
            splat_positions: if matches!(renderer.config.depth_sorting, DepthSorting::Cpu) {
                vec![0.0; splat_count * 3]
            } else {
                Vec::new()
            },
            splat_count,
        }
    }

    /// Loads the development test scene
    pub fn new_dev_test(device: &wgpu::Device, renderer: &Renderer, queue: &wgpu::Queue) -> Self {
        let mut scene = Self::new(device, renderer, 3);
        let splat_index_range = 0..3;
        let mut splat_data = vec![[0.0; 60]; splat_index_range.len()];
        for index in 0..splat_data.len() {
            splat_data[index][0] = 1.0;
            splat_data[index][1] = 0.0;
            splat_data[index][2] = 0.0;
            splat_data[index][3] = 0.0;
            splat_data[index][4] = if index == 0 { 1.0 } else { 0.0 };
            splat_data[index][5] = if index == 1 { 1.0 } else { 0.0 };
            splat_data[index][6] = if index == 2 { 1.0 } else { 0.0 };
            splat_data[index][8] = 0.001;
            splat_data[index][9] = 0.001;
            splat_data[index][10] = 0.2;
            splat_data[index][11] = 1.0;
            splat_data[index][12] = if index == 0 { 1.0 } else { 0.0 };
            splat_data[index][13] = if index == 1 { 1.0 } else { 0.0 };
            splat_data[index][14] = if index == 2 { 1.0 } else { 0.0 };
        }
        queue.write_buffer(
            &scene.splat_buffer,
            (splat_index_range.start * std::mem::size_of::<Splat>()) as u64,
            transmute_slice(&splat_data),
        );
        if !scene.splat_positions.is_empty() {
            for index in 0..splat_data.len() {
                let splat = &splat_data[index];
                scene.splat_positions[index * 3 + 0] = splat[4];
                scene.splat_positions[index * 3 + 1] = splat[5];
                scene.splat_positions[index * 3 + 2] = splat[6];
            }
        }
        scene
    }

    /// Parses the header of a splat file
    ///
    /// Returns the header length in bytes, the number of splats in the file and the file handle
    pub fn parse_file_header(file: File) -> (u16, usize, File) {
        let mut splat_count: usize = 0;
        let mut reader = BufReader::new(file);
        let mut line = String::new();
        loop {
            reader.read_line(&mut line).unwrap();
            if line == "end_header\n" {
                break;
            }
            if line.starts_with("element vertex ") {
                splat_count = line[15..line.len() - 1].parse().unwrap();
            }
            line.clear();
        }
        let file_header_size = reader.stream_position().unwrap() as u16;
        (file_header_size, splat_count, reader.into_inner())
    }

    /// Loads a chunk of the splat file or the entire file
    pub fn load_chunk(&mut self, queue: &wgpu::Queue, file: &mut File, file_header_size: u16, mut splat_index_range: std::ops::Range<usize>) {
        splat_index_range.end = splat_index_range.end.min(self.splat_count);
        let mut splat_data = vec![[0.0; 60]; splat_index_range.len()];
        let mut serialized_splats = vec![SerializedSplat::default(); splat_index_range.len()];
        file.seek(SeekFrom::Start(
            file_header_size as u64 + (splat_index_range.start * std::mem::size_of::<SerializedSplat>()) as u64,
        ))
        .unwrap();
        file.read_exact(transmute_slice_mut::<_, u8>(&mut serialized_splats)).unwrap();
        for index in 0..splat_data.len() {
            let serialized_splat = &serialized_splats[index];
            let rotor = <[f32; 4]>::from(Rotor::from(serialized_splat.rotation).signum());
            splat_data[index][0] = rotor[0];
            splat_data[index][1] = rotor[1];
            splat_data[index][2] = rotor[2];
            splat_data[index][3] = rotor[3];
            splat_data[index][4] = serialized_splat.center[0];
            splat_data[index][5] = serialized_splat.center[1];
            splat_data[index][6] = serialized_splat.center[2];
            let average = (serialized_splat.scale[0] + serialized_splat.scale[1] + serialized_splat.scale[2]) / 3.0;
            splat_data[index][8] = serialized_splat.scale[0]
                .max(average - MAX_SIZE_VARIANCE)
                .min(average + MAX_SIZE_VARIANCE)
                .exp();
            splat_data[index][9] = serialized_splat.scale[1]
                .max(average - MAX_SIZE_VARIANCE)
                .min(average + MAX_SIZE_VARIANCE)
                .exp();
            splat_data[index][10] = serialized_splat.scale[2]
                .max(average - MAX_SIZE_VARIANCE)
                .min(average + MAX_SIZE_VARIANCE)
                .exp();
            splat_data[index][11] = 1.0 / (1.0 + (-serialized_splat.alpha).exp());
            splat_data[index][12..].copy_from_slice(&serialized_splat.color[0..3 * 16]);
        }
        queue.write_buffer(
            &self.splat_buffer,
            (splat_index_range.start * std::mem::size_of::<Splat>()) as u64,
            transmute_slice(&splat_data),
        );
        if !self.splat_positions.is_empty() {
            for index in 0..splat_data.len() {
                let splat = &splat_data[index];
                self.splat_positions[index * 3 + 0] = splat[4];
                self.splat_positions[index * 3 + 1] = splat[5];
                self.splat_positions[index * 3 + 2] = splat[6];
            }
        }
    }
}