finished tui

Cargo.toml

@@ -6,9 +6,26 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
+log = "0.4"
 image = "0.24"
 gif = "0.12"
-tiny-skia = "0.11"
+futures = "0.3"
+wgpu = {version="0.18", features=["webgl"]}
+bytemuck = {version="1.14", features=["derive"]}
+futures-intrusive = "0.5"
+pollster = "0.3"
+clap = { version = "4.4.16", features = ["derive"] }
+indicatif = "0.17"
+
+[dev-dependencies]
+flexi_logger = "0.27"
+
+[profile.dev]
+opt-level=1
+# codegen-units = 1
+# lto = "fat"
 
 [profile.release]
+opt-level=3
 codegen-units = 1
+lto = "fat"

src/draw.rs

@@ -0,0 +1,449 @@
+use wgpu::{
+  util::{BufferInitDescriptor, DeviceExt},
+  *,
+};
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct Vertex {
+  position: [f32; 2],
+  uv: [f32; 2],
+}
+
+impl Vertex {
+  pub fn desc() -> wgpu::VertexBufferLayout<'static> {
+    wgpu::VertexBufferLayout {
+      array_stride: std::mem::size_of::<Self>() as wgpu::BufferAddress,
+      step_mode: wgpu::VertexStepMode::Vertex,
+      attributes: &[
+        wgpu::VertexAttribute {
+          offset: 0,
+          shader_location: 0,
+          format: wgpu::VertexFormat::Float32x2,
+        },
+        wgpu::VertexAttribute {
+          offset: std::mem::size_of::<[f32; 2]>() as wgpu::BufferAddress,
+          shader_location: 1,
+          format: wgpu::VertexFormat::Float32x2,
+        },
+      ],
+    }
+  }
+}
+
+#[repr(C)]
+#[derive(Copy, Clone, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+pub struct Circle {
+  pub position: [f32; 2],
+  pub radius: f32,
+  pub color: [u8; 4],
+}
+
+impl Circle {
+  fn desc() -> wgpu::VertexBufferLayout<'static> {
+    use std::mem;
+    wgpu::VertexBufferLayout {
+      array_stride: mem::size_of::<Self>() as wgpu::BufferAddress,
+      step_mode: wgpu::VertexStepMode::Instance,
+      attributes: &[
+        wgpu::VertexAttribute {
+          offset: 0,
+          shader_location: 2,
+          format: wgpu::VertexFormat::Float32x2,
+        },
+        wgpu::VertexAttribute {
+          offset: mem::size_of::<[f32; 2]>() as wgpu::BufferAddress,
+          shader_location: 3,
+          format: wgpu::VertexFormat::Float32,
+        },
+        wgpu::VertexAttribute {
+          offset: mem::size_of::<[f32; 3]>() as wgpu::BufferAddress,
+          shader_location: 4,
+          format: wgpu::VertexFormat::Uint32,
+        },
+      ],
+    }
+  }
+}
+
+#[repr(C)]
+#[derive(Debug, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
+struct GpuUniforms {
+  width: f32,
+  height: f32,
+}
+
+const SQUARE: &[Vertex] = &[
+  Vertex {
+    position: [-1.0, 1.0],
+    uv: [0.0, 0.0],
+  },
+  Vertex {
+    position: [1.0, 1.0],
+    uv: [1.0, 0.0],
+  },
+  Vertex {
+    position: [1.0, -1.0],
+    uv: [1.0, 1.0],
+  },
+  Vertex {
+    position: [1.0, -1.0],
+    uv: [1.0, 1.0],
+  },
+  Vertex {
+    position: [-1.0, -1.0],
+    uv: [0.0, 1.0],
+  },
+  Vertex {
+    position: [-1.0, 1.0],
+    uv: [0.0, 0.0],
+  },
+];
+
+pub struct QuickDraw {
+  device: Device,
+  queue: Queue,
+  width: u32,
+  height: u32,
+  uniform_buffer: Buffer,
+  uniform_bind_group: BindGroup,
+  texture_desc: TextureDescriptor<'static>,
+  texture: Texture,
+  texture_view: TextureView,
+  instance_count: u64,
+  instance_buffer: Buffer,
+  output_buffer: Buffer,
+  vertex_buffer: Buffer,
+  pipeline: RenderPipeline,
+}
+
+impl QuickDraw {
+  pub async fn new(width: u32, height: u32, max_circles: u64) -> Self {
+    let instance = wgpu::Instance::new(wgpu::InstanceDescriptor {
+      backends: wgpu::Backends::all(),
+      ..Default::default()
+    });
+
+    let adapter = match instance
+      .request_adapter(&wgpu::RequestAdapterOptions {
+        power_preference: wgpu::PowerPreference::HighPerformance,
+        compatible_surface: None,
+        ..Default::default()
+      })
+      .await
+    {
+      Some(a) => a,
+      None => {
+        eprintln!("Could not find WGPU adapter");
+        std::process::exit(1);
+      },
+    };
+    let (device, queue) =
+      match adapter.request_device(&Default::default(), None).await {
+        Ok(r) => r,
+        Err(e) => {
+          eprintln!("Could not find WGPU device:");
+          eprintln!("{}", e);
+          std::process::exit(1);
+        },
+      };
+
+    let uniforms = GpuUniforms {
+      width: width as f32,
+      height: height as f32,
+    };
+    let uniform_buffer = device.create_buffer_init(&BufferInitDescriptor {
+      label: Some("Uniform Buffer"),
+      contents: bytemuck::cast_slice(&[uniforms]),
+      usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+    });
+    let uniform_layout =
+      device.create_bind_group_layout(&BindGroupLayoutDescriptor {
+        entries: &[wgpu::BindGroupLayoutEntry {
+          binding: 0,
+          visibility: wgpu::ShaderStages::VERTEX,
+          ty: wgpu::BindingType::Buffer {
+            ty: wgpu::BufferBindingType::Uniform,
+            has_dynamic_offset: false,
+            min_binding_size: None,
+          },
+          count: None,
+        }],
+        label: Some("Uniform Bind Group Layout"),
+      });
+
+    let uniform_bind_group =
+      device.create_bind_group(&wgpu::BindGroupDescriptor {
+        label: Some("Uniform Bind Group"),
+        layout: &uniform_layout,
+        entries: &[BindGroupEntry {
+          binding: 0,
+          resource: uniform_buffer.as_entire_binding(),
+        }],
+      });
+
+    let texture_desc = wgpu::TextureDescriptor {
+      size: wgpu::Extent3d {
+        width,
+        height,
+        depth_or_array_layers: 1,
+      },
+      mip_level_count: 1,
+      sample_count: 1,
+      dimension: wgpu::TextureDimension::D2,
+      format: wgpu::TextureFormat::Rgba8Unorm,
+      view_formats: &[wgpu::TextureFormat::Rgba8Unorm],
+      usage: wgpu::TextureUsages::COPY_SRC
+        | wgpu::TextureUsages::RENDER_ATTACHMENT,
+      label: None,
+    };
+    let texture = device.create_texture(&texture_desc);
+    let texture_view = texture.create_view(&Default::default());
+
+    let instance_buffer = device.create_buffer(&BufferDescriptor {
+      label: Some("Instance Buffer"),
+      usage: BufferUsages::VERTEX | BufferUsages::COPY_DST,
+      mapped_at_creation: false,
+      size: max_circles * std::mem::size_of::<Circle>() as u64,
+    });
+
+    let u32_size = std::mem::size_of::<u32>() as u32;
+    let output_buffer_size = (u32_size * width * height) as wgpu::BufferAddress;
+    let output_buffer = device.create_buffer(&wgpu::BufferDescriptor {
+      size: output_buffer_size,
+      usage: wgpu::BufferUsages::COPY_DST
+        // this tells wpgu that we want to read this buffer from the cpu
+        | wgpu::BufferUsages::MAP_READ,
+      label: None,
+      mapped_at_creation: true,
+    });
+    let vertex_buffer = device.create_buffer_init(&BufferInitDescriptor {
+      label: Some("Vertex Buffer"),
+      usage: BufferUsages::VERTEX,
+      contents: bytemuck::cast_slice(SQUARE),
+    });
+    // Shaders
+    let shader = std::borrow::Cow::Borrowed(include_str!("shader.wgsl"));
+    let vert = device.create_shader_module(ShaderModuleDescriptor {
+      label: Some("Vertex Shader"),
+      source: ShaderSource::Wgsl(shader.clone()),
+    });
+    let frag = device.create_shader_module(ShaderModuleDescriptor {
+      label: Some("Vertex Shader"),
+      source: ShaderSource::Wgsl(shader),
+    });
+    // Pipeline
+    let render_pipeline_layout =
+      device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+        label: Some("Render Pipeline Layout"),
+        bind_group_layouts: &[&uniform_layout],
+        push_constant_ranges: &[],
+      });
+    let pipeline =
+      device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+        multiview: None,
+        label: Some("Render Pipeline"),
+        layout: Some(&render_pipeline_layout),
+        vertex: wgpu::VertexState {
+          module: &vert,
+          entry_point: "vs_main",
+          buffers: &[Vertex::desc(), Circle::desc()],
+        },
+        fragment: Some(wgpu::FragmentState {
+          module: &frag,
+          entry_point: "fs_main",
+          targets: &[Some(wgpu::ColorTargetState {
+            format: texture_desc.format,
+            blend: Some(wgpu::BlendState::ALPHA_BLENDING),
+            write_mask: wgpu::ColorWrites::ALL,
+          })],
+        }),
+        primitive: wgpu::PrimitiveState {
+          topology: wgpu::PrimitiveTopology::TriangleList,
+          strip_index_format: None,
+          front_face: wgpu::FrontFace::Ccw,
+          cull_mode: None,
+          unclipped_depth: false,
+          conservative: false,
+          // Setting this to anything other than Fill requires
+          // Features::NON_FILL_POLYGON_MODE
+          polygon_mode: wgpu::PolygonMode::Fill,
+        },
+        depth_stencil: None,
+        multisample: wgpu::MultisampleState {
+          count: 1,
+          mask: !0,
+          alpha_to_coverage_enabled: false,
+        },
+      });
+
+    Self {
+      device,
+      queue,
+      width,
+      height,
+      uniform_buffer,
+      uniform_bind_group,
+      texture_desc,
+      texture,
+      texture_view,
+      instance_count: 0,
+      instance_buffer,
+      output_buffer,
+      vertex_buffer,
+      pipeline,
+    }
+  }
+
+  pub async fn resize(&mut self, width: u32, height: u32, max_circles: usize) {
+    let texture_desc = wgpu::TextureDescriptor {
+      size: wgpu::Extent3d {
+        width,
+        height,
+        depth_or_array_layers: 1,
+      },
+      mip_level_count: 1,
+      sample_count: 1,
+      dimension: wgpu::TextureDimension::D2,
+      format: wgpu::TextureFormat::Rgba8Unorm,
+      view_formats: &[wgpu::TextureFormat::Rgba8Unorm],
+      usage: wgpu::TextureUsages::COPY_SRC
+        | wgpu::TextureUsages::RENDER_ATTACHMENT,
+      label: None,
+    };
+    let texture = self.device.create_texture(&texture_desc);
+    let texture_view = texture.create_view(&Default::default());
+
+    let u32_size = std::mem::size_of::<u32>() as u32;
+    let output_buffer_size = (u32_size * width * height) as wgpu::BufferAddress;
+    let output_buffer = self.device.create_buffer(&wgpu::BufferDescriptor {
+      size: output_buffer_size,
+      usage: wgpu::BufferUsages::COPY_DST
+        // this tells wpgu that we want to read this buffer from the cpu
+        | wgpu::BufferUsages::MAP_READ,
+      label: None,
+      mapped_at_creation: true,
+    });
+    self.allocate(max_circles).await;
+    self.width = width;
+    self.height = height;
+    self.texture_desc = texture_desc;
+    self.texture = texture;
+    self.texture_view = texture_view;
+    self.output_buffer = output_buffer;
+  }
+
+  async fn draw_call(&mut self) {
+    let mut encoder = self
+      .device
+      .create_command_encoder(&wgpu::CommandEncoderDescriptor { label: None });
+    let render_pass_desc = wgpu::RenderPassDescriptor {
+      occlusion_query_set: None,
+      timestamp_writes: None,
+      label: Some("Render Pass"),
+      color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+        view: &self.texture_view,
+        resolve_target: None,
+        ops: wgpu::Operations {
+          load: wgpu::LoadOp::Clear(wgpu::Color {
+            r: 0.0,
+            g: 0.0,
+            b: 0.0,
+            a: 1.0,
+          }),
+          store: wgpu::StoreOp::Store,
+        },
+      })],
+      depth_stencil_attachment: None,
+    };
+    let mut render_pass = encoder.begin_render_pass(&render_pass_desc);
+
+    render_pass.set_pipeline(&self.pipeline);
+    render_pass.set_bind_group(0, &self.uniform_bind_group, &[]);
+    render_pass.set_vertex_buffer(0, self.vertex_buffer.slice(..));
+    render_pass.set_vertex_buffer(1, self.instance_buffer.slice(..));
+    render_pass.draw(0..6, 0..self.instance_count as u32);
+    drop(render_pass);
+
+    let u32_size = std::mem::size_of::<u32>() as u32;
+
+    encoder.copy_texture_to_buffer(
+      wgpu::ImageCopyTexture {
+        aspect: wgpu::TextureAspect::All,
+        texture: &self.texture,
+        mip_level: 0,
+        origin: wgpu::Origin3d::ZERO,
+      },
+      wgpu::ImageCopyBuffer {
+        buffer: &self.output_buffer,
+        layout: wgpu::ImageDataLayout {
+          offset: 0,
+          bytes_per_row: Some(u32_size * self.width),
+          rows_per_image: Some(self.height),
+        },
+      },
+      self.texture_desc.size,
+    );
+    self.output_buffer.unmap();
+    self.queue.submit(Some(encoder.finish()));
+  }
+
+  async fn read_output_buffer(&self) -> BufferView {
+    let buffer_slice = self.output_buffer.slice(..);
+
+    // NOTE: We have to create the mapping THEN device.poll()
+    // before await the future. Otherwise the application
+    // will freeze.
+    let (tx, rx) = futures_intrusive::channel::shared::oneshot_channel();
+    buffer_slice.map_async(wgpu::MapMode::Read, move |result| {
+      tx.send(result).unwrap();
+    });
+    // buffer_slice.map_async(wgpu::MapMode::Read, |_| {});
+    self.device.poll(wgpu::Maintain::Wait);
+    rx.receive().await.unwrap().unwrap();
+    buffer_slice.get_mapped_range()
+  }
+
+  // Allocates space for *size* circles, also clears the
+  // circles buffer
+  pub async fn allocate(&mut self, size: usize) {
+    let size = (size * std::mem::size_of::<Circle>()) as u64;
+    if size >= self.instance_buffer.size() {
+      self.instance_buffer = self.device.create_buffer(&BufferDescriptor {
+        label: Some("Instance Buffer"),
+        usage: BufferUsages::VERTEX | BufferUsages::COPY_DST,
+        mapped_at_creation: false,
+        size,
+      });
+    }
+  }
+
+  async fn write_circles(&mut self, circles: &[Circle]) {
+    self.allocate(circles.len()).await;
+    self.queue.write_buffer(
+      &self.instance_buffer,
+      0,
+      &bytemuck::cast_slice(&circles),
+    );
+    // self.queue.submit([]);
+    self.instance_count = circles.len() as u64;
+  }
+
+  async fn bytes(&self) -> Vec<u8> {
+    let mut buffer = vec![0; self.output_buffer.size() as usize];
+    {
+      let data = self.read_output_buffer().await;
+
+      for (i, d) in data.iter().enumerate() {
+        buffer[i] = *d;
+      }
+    }
+    buffer
+  }
+
+  pub async fn draw_circles(&mut self, circles: &[Circle]) -> Vec<u8> {
+    self.write_circles(circles).await;
+    self.draw_call().await;
+    self.bytes().await
+  }
+}

src/drawing.rs

@@ -1,37 +0,0 @@
-use tiny_skia::*;
-
-use crate::sim::Circle;
-pub fn clear(pixmap: &mut Pixmap) {
-  let mut black = Paint::default();
-  black.set_color_rgba8(0, 0, 0, 255);
-  pixmap.fill_rect(
-    Rect::from_xywh(0.0, 0.0, pixmap.width() as f32, pixmap.height() as f32)
-      .unwrap(),
-    &black,
-    Transform::identity(),
-    None,
-  );
-}
-
-pub fn draw_circle(pixmap: &mut Pixmap, circle: &Circle) {
-  let x = circle.position.x;
-  let y = circle.position.y;
-  let radius = circle.radius;
-  let r = circle.color.0;
-  let g = circle.color.1;
-  let b = circle.color.2;
-  let mut paint = Paint::default();
-  paint.set_color_rgba8(r, g, b, 255);
-  let mut brush = PathBuilder::new();
-  brush.push_circle(x, y, radius);
-  brush.move_to(x, y);
-  brush.close();
-
-  pixmap.fill_path(
-    &brush.finish().unwrap(),
-    &paint,
-    FillRule::Winding,
-    tiny_skia::Transform::identity(),
-    None,
-  )
-}

src/lib.rs

@@ -1,58 +0,0 @@
-use std::io::Write;
-
-use image::{ImageBuffer, Rgb};
-
-mod drawing;
-mod helper;
-mod sim;
-
-fn write_frame<W: Write>(
-  sim: &sim::Simulation,
-  encoder: &mut gif::Encoder<&mut W>,
-) {
-  let mut pixbuf = tiny_skia::Pixmap::new(500, 500).unwrap();
-  drawing::clear(&mut pixbuf);
-  for index in 0..sim.circles.len() {
-    let circle = &sim.circles[index];
-    drawing::draw_circle(&mut pixbuf, circle);
-  }
-  let mut frame = gif::Frame::from_rgba(500, 500, pixbuf.data_mut());
-  frame.delay = 1;
-  encoder.write_frame(&frame).unwrap();
-}
-
-async fn write_frame_async<W: Write>(
-  sim: &sim::Simulation,
-  encoder: &mut gif::Encoder<&mut W>,
-) {
-  write_frame(sim, encoder);
-}
-
-pub fn generate(image: ImageBuffer<Rgb<u8>, Vec<u8>>) -> Vec<u8> {
-  let (mut sim, it) = sim::Simulation::simulate_image(500.0, 500.0, 8.0, image);
-
-  let mut buffer = Vec::<u8>::new();
-  let mut encoder = gif::Encoder::new(&mut buffer, 500, 500, &[]).unwrap();
-  // encoder.set_repeat(gif::Repeat::Infinite).unwrap();
-  while sim.clock < it {
-    write_frame(&sim, &mut encoder);
-    (0..5).for_each(|_| sim.step());
-  }
-  drop(encoder);
-  buffer
-}
-
-pub async fn generate_async(image: ImageBuffer<Rgb<u8>, Vec<u8>>) -> Vec<u8> {
-  let (mut sim, it) =
-    sim::Simulation::simulate_image_async(500.0, 500.0, 8.0, image).await;
-
-  let mut buffer = Vec::<u8>::new();
-  let mut encoder = gif::Encoder::new(&mut buffer, 500, 500, &[]).unwrap();
-  // encoder.set_repeat(gif::Repeat::Infinite).unwrap();
-  while sim.clock < it {
-    write_frame_async(&sim, &mut encoder).await;
-    (0..5).for_each(|_| sim.step());
-  }
-  drop(encoder);
-  buffer
-}

src/main.rs

@@ -0,0 +1,176 @@
+#![feature(future_join)]
+use std::{cell::RefCell, future::join, io::Write, path::Path, sync::Arc};
+
+use draw::QuickDraw;
+use gif::Frame;
+use image::{ImageBuffer, Rgb};
+use indicatif::{ProgressBar, ProgressStyle};
+use log::debug;
+
+pub mod draw;
+pub mod helper;
+pub mod sim;
+pub mod tests;
+
+pub fn make_progress(msg: &'static str, max: u64) -> ProgressBar {
+  let progress = ProgressBar::new(max);
+  progress.set_style(ProgressStyle::with_template("{msg} :: {bar}").unwrap());
+  progress.set_message(msg);
+  progress
+}
+
+const WIDTH: f32 = 512.0;
+const HEIGHT: f32 = 512.0;
+
+pub async fn preprocess(
+  image: ImageBuffer<Rgb<u8>, Vec<u8>>,
+  radius: f32,
+) -> (sim::Simulation, usize, usize) {
+  let (sim, it, max_circles) =
+    sim::Simulation::simulate_image(WIDTH, HEIGHT, radius, image).await;
+  (sim, it, max_circles)
+}
+
+pub async fn simulate(
+  draw: &mut QuickDraw,
+  mut sim: Simulation,
+  it: usize,
+  step: usize,
+  max_circles: usize,
+) -> Vec<Frame<'static>> {
+  draw.resize(WIDTH as u32, HEIGHT as u32, max_circles).await;
+  let mut frames = vec![];
+  let progress =
+    make_progress("Simulating   ", ((it - sim.clock) / sim.substeps) as u64);
+  while sim.clock < it {
+    let circles = &sim
+      .circles
+      .iter()
+      .map(|c| draw::Circle {
+        position: [c.position.x, c.position.y],
+        radius: c.radius,
+        color: [c.color.0, c.color.1, c.color.2, 255],
+      })
+      .collect::<Vec<draw::Circle>>();
+    let bytes_future = draw.draw_circles(circles);
+    let steps = sim.steps(step);
+    let mut bytes = join!(bytes_future, steps).await.0;
+    let frame = gif::Frame::from_rgba(WIDTH as u16, HEIGHT as u16, &mut bytes);
+    frames.push(frame);
+    progress.inc(step as u64);
+  }
+  progress.finish();
+  frames
+}
+
+pub async fn encode(frames: Vec<Frame<'static>>, repeat: bool) -> Vec<u8> {
+  let mut buffer = Vec::<u8>::new();
+  let mut encoder =
+    gif::Encoder::new(&mut buffer, WIDTH as u16, HEIGHT as u16, &[]).unwrap();
+  let progress = make_progress("Encoding     ", frames.len() as u64);
+  for mut frame in frames {
+    frame.delay = 1;
+    encoder.write_frame(&frame).unwrap();
+    progress.inc(1);
+  }
+  progress.finish();
+  if repeat {
+    encoder.set_repeat(gif::Repeat::Infinite).unwrap();
+  }
+  drop(encoder);
+  buffer
+}
+
+use clap::Parser;
+use sim::Simulation;
+#[derive(Parser, Debug)]
+#[command(author, version, about, long_about = None)]
+struct Args {
+  /// Input file to process. All common image types are
+  /// supported, see the `image` crate docs for specific
+  /// compatibility
+  #[arg(short = 'i', value_hint = clap::ValueHint::DirPath)]
+  input: std::path::PathBuf,
+
+  /// Output file path ('./output.gif' by default)
+  #[arg(short = 'o', value_hint = clap::ValueHint::DirPath)]
+  output: Option<std::path::PathBuf>,
+
+  /// How many physics-steps between frames. Affects the
+  /// speed and length of the animation
+  #[arg(short = 's')]
+  step: Option<u32>,
+
+  /// Radius of the circle, smaller means more detailed but
+  /// slower to compute (8.0 by default, must be between
+  /// 1.0 and 50.0 inclusive)
+  #[arg(short = 'r')]
+  radius: Option<f32>,
+
+  /// Loop the GIF
+  #[arg(short = 'l', long = "loop")]
+  looping: bool,
+}
+
+fn open_image<P: AsRef<Path>>(path: P) -> ImageBuffer<Rgb<u8>, Vec<u8>> {
+  let image = match image::io::Reader::open(path) {
+    Ok(i) => i,
+    Err(e) => {
+      eprintln!("Error opening file for processing: ");
+      eprintln!("{}", e);
+      std::process::exit(1);
+    },
+  };
+
+  let decoded = match image.decode() {
+    Ok(d) => d,
+    Err(e) => {
+      eprintln!("Error processing file: ");
+      eprintln!("{}", e);
+      std::process::exit(1);
+    },
+  };
+  decoded.to_rgb8()
+}
+
+fn main() {
+  let args = Args::parse();
+
+  let output = args.output.unwrap_or("output.gif".into());
+  let step = args.step.unwrap_or(20) as usize;
+  if step == 0 {
+    eprintln!("Physics step cannot be 0");
+    std::process::exit(1);
+  }
+  let radius = args.radius.unwrap_or(8.0);
+  if radius < 1.0 || radius > 50.0 {
+    eprintln!("Invalid radius {}", radius);
+    eprintln!("Must be between 1.0 and 50.0 inclusive");
+    std::process::exit(1);
+  }
+
+  let image = open_image(args.input);
+  let (sim, it, max) = pollster::block_on(preprocess(image, radius));
+  let mut draw = pollster::block_on(QuickDraw::new(512, 512, 1000));
+  let frames = pollster::block_on(simulate(&mut draw, sim, it, step, max));
+  let gif = pollster::block_on(encode(frames, args.looping));
+  let mut file = match std::fs::File::create(output.clone()) {
+    Ok(f) => f,
+    Err(e) => {
+      eprintln!("Error opening file for writing:");
+      eprintln!("{}", e);
+      std::process::exit(1);
+    },
+  };
+
+  match file.write(&gif) {
+    Ok(_) => {
+      println!("Successfully created file at '{}'", output.display());
+    },
+    Err(e) => {
+      eprintln!("Error writing output file: ");
+      eprintln!("{}", e);
+      std::process::exit(1);
+    },
+  };
+}

src/shader.wgsl

@@ -0,0 +1,58 @@
+struct Uniforms {
+  width: f32, height: f32,
+}
+
+@group(0) @binding(0) var<uniform> uniforms: Uniforms;
+
+struct VertexInput {
+  @location(0) position: vec2<f32>,
+  @location(1) uv: vec2<f32>,
+  @location(2) offset: vec2<f32>,
+  @location(3) radius: f32,
+  @location(4) color: u32,
+}
+
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) uv: vec2<f32>,
+    @location(1) color: vec4<f32>,
+}
+
+fn get_color(input: u32) -> vec4<f32> {
+    let r = f32(input & 255u) / 255.0;
+    let g = f32((input & (255u << 8u)) >> 8u) / 255.0;
+    let b = f32((input & (255u << 16u)) >> 16u) / 255.0;
+    let a = f32((input & (255u << 24u)) >> 24u) / 255.0;
+    return vec4<f32>(r, g, b, a);
+}
+
+@vertex
+fn vs_main(
+    model: VertexInput
+) -> VertexOutput {
+    var out: VertexOutput;
+    var x = (model.position.x * model.radius) - uniforms.width / 2.0;
+    x += model.offset.x;
+    var y = (model.position.y * model.radius) - uniforms.height / 2.0;
+    y += model.offset.y;
+    var norm_x = x / (uniforms.width / 2.0);
+    var norm_y = -y / (uniforms.height / 2.0);
+    out.clip_position = vec4<f32>(norm_x, norm_y, 0.0, 1.0);
+    out.uv = model.uv;
+    out.color = get_color(model.color);
+    return out;
+}
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    var x = in.uv.x - 0.5;
+    var y = in.uv.y - 0.5;
+    var d = sqrt(x * x + y * y) * 2.0;
+    if d > 1.1 {
+      discard;
+    } else if d > 1.0 {
+        return vec4<f32>(in.color.xyz, 1.1 - d);
+    } else {
+        return in.color;
+    }
+}

src/sim.rs

@@ -1,6 +1,6 @@
 use std::hash::{Hash, Hasher};
 
-use crate::helper::*;
+use crate::{helper::*, make_progress};
 use image::{ImageBuffer, Rgb};
 
 pub struct Circle {
@@ -16,7 +16,7 @@ pub struct Simulation {
   pub colors: Vec<Color>,
   pub max_circles: usize,
   pub clock: usize,
-  substeps: usize,
+  pub substeps: usize,
   rand_seed: usize,
   timescale: f32,
   circle_radius: f32,
@@ -27,6 +27,8 @@ pub struct Simulation {
 }
 
 impl Simulation {
+  pub const POST_PROCESS: usize = 120;
+
   pub fn new(
     width: f32,
     height: f32,
@@ -35,12 +37,12 @@ impl Simulation {
   ) -> Self {
     let area = width * height;
     let circle_area = circle_radius.powi(2) * std::f32::consts::PI;
-    let approx_max = ((area / circle_area).round() * 1.0) as usize;
+    let approx_max = ((area / circle_area).round() * 0.9) as usize;
     Self {
       circles: Vec::with_capacity(approx_max),
       max_circles: approx_max,
       timescale: 1.0 / 60.0,
-      substeps: 4,
+      substeps: 8,
       colors: vec![Color(255, 255, 255); approx_max],
       clock: rand_seed,
       rand_seed,
@@ -48,12 +50,15 @@ impl Simulation {
       radius_variance: circle_radius * 0.1,
       area_size: (width, height),
       gravity: height,
-      response_mod: 0.4,
+      response_mod: 0.9,
     }
   }
 
   #[inline]
-  fn assign_colors_from_image(&mut self, img: ImageBuffer<Rgb<u8>, Vec<u8>>) {
+  async fn assign_colors_from_image(
+    &mut self,
+    img: ImageBuffer<Rgb<u8>, Vec<u8>>,
+  ) {
     let (width, height) = (img.width() as f32 - 1.0, img.height() as f32 - 1.0);
     for (pos, index) in self.circles.iter().map(|c| (c.position, c.index)) {
       let img_x =
@@ -68,53 +73,37 @@ impl Simulation {
   }
 
   #[inline]
-  pub fn simulate_image(
+  pub async fn simulate_image(
     width: f32,
     height: f32,
     circle_radius: f32,
     img: ImageBuffer<Rgb<u8>, Vec<u8>>,
-  ) -> (Self, usize) {
+  ) -> (Self, usize, usize) {
     let image_hash = ({
       let mut s = std::hash::DefaultHasher::new();
       img.hash(&mut s);
       s.finish()
     } % 1204) as usize;
     let mut sim = Simulation::new(width, height, circle_radius, image_hash);
+    let progress = make_progress(
+      "Preprocessing",
+      (sim.max_circles + Self::POST_PROCESS) as u64,
+    );
     while sim.circles() < sim.max_circles {
-      sim.step();
+      sim.step().await;
+      progress.inc(2);
     }
-    (0..120).for_each(|_| sim.step());
+    for _ in 0..Self::POST_PROCESS {
+      sim.step().await;
+      progress.inc(1);
+    }
+    progress.finish();
     let total_iterations = sim.clock;
-    sim.assign_colors_from_image(img);
+    let max_circles = sim.circles.len();
+    sim.assign_colors_from_image(img).await;
     sim.circles.clear();
     sim.clock = sim.rand_seed;
-    (sim, total_iterations)
-  }
-
-  #[inline]
-  pub async fn simulate_image_async(
-    width: f32,
-    height: f32,
-    circle_radius: f32,
-    img: ImageBuffer<Rgb<u8>, Vec<u8>>,
-  ) -> (Self, usize) {
-    let image_hash = ({
-      let mut s = std::hash::DefaultHasher::new();
-      img.hash(&mut s);
-      s.finish()
-    } % 1204) as usize;
-    let mut sim = Simulation::new(width, height, circle_radius, image_hash);
-    while sim.circles() < sim.max_circles {
-      sim.step_async().await;
-    }
-    for _ in 0..120 {
-      sim.step_async().await
-    }
-    let total_iterations = sim.clock;
-    sim.assign_colors_from_image(img);
-    sim.circles.clear();
-    sim.clock = sim.rand_seed;
-    (sim, total_iterations)
+    (sim, total_iterations, max_circles)
   }
 
   #[inline]
@@ -155,7 +144,7 @@ impl Simulation {
 
   // Insertion sort
   #[inline]
-  fn sort(&mut self) {
+  async fn sort(&mut self) {
     if self.circles.len() == 1 {
       return;
     }
@@ -170,7 +159,7 @@ impl Simulation {
   }
 
   #[inline]
-  fn integrate(&mut self) {
+  async fn integrate(&mut self) {
     let delta = self.timescale * (1.0 / self.substeps as f32);
     let gravity = Vector2::new(0.0, self.gravity) * delta.powi(2);
     self.circles.iter_mut().for_each(|circle| {
@@ -181,7 +170,7 @@ impl Simulation {
   }
 
   #[inline]
-  fn collide(&mut self) {
+  async fn collide(&mut self) {
     for i in 0..self.circles.len() {
       // Apply gravity
       for j in i..self.circles.len() {
@@ -229,7 +218,7 @@ impl Simulation {
   }
 
   #[inline]
-  pub fn step(&mut self) {
+  pub async fn step(&mut self) {
     if self.circles.len() < self.max_circles {
       self.launch();
     }
@@ -237,42 +226,22 @@ impl Simulation {
       self.launch2();
     }
 
-    (0..self.substeps).for_each(|_| {
+    for _ in 0..self.substeps {
       self.constrain_rect();
-      self.sort();
-      self.collide();
-      self.integrate();
+      self.sort().await;
+      self.collide().await;
+      self.integrate().await;
       self.clock += 1;
-    });
+    }
   }
 
   #[inline]
-  pub async fn step_async(&mut self) {
-    if self.circles.len() < self.max_circles {
-      self.launch();
+  pub async fn steps(&mut self, steps: usize) {
+    for _ in 0..steps {
+      self.step().await;
     }
-    if self.circles.len() < self.max_circles {
-      self.launch2();
   }
 
-    (0..self.substeps).for_each(|_| {
-      self.constrain_rect();
-      self.sort();
-      self.collide();
-      self.integrate();
-      self.clock += 1;
-    });
-  }
-
-  // #[inline]
-  // pub fn draw(&self, d: &mut RaylibDrawHandle) {
-  //   self
-  //     .circles
-  //     .iter()
-  //     .map(|c| (c.position, c.radius, c.color))
-  //     .for_each(|(pos, radius, color)| d.draw_circle_v(pos,
-  // radius, color)); }
-
   pub fn circles(&self) -> usize {
     self.circles.len()
   }

src/tests.rs

@@ -0,0 +1,19 @@
+#[cfg(test)]
+#[test]
+fn bench_async() {
+  use std::io::Write;
+  flexi_logger::Logger::try_with_str("debug").unwrap();
+
+  let image = image::io::Reader::open(
+    "resources/mona
+  lisa.png",
+  )
+  .unwrap()
+  .decode()
+  .unwrap()
+  .to_rgb8();
+  // let gif = pollster::block_on(crate::generate_async(image, None));
+  // let mut file =
+  // std::fs::File::create("test.gif").unwrap();
+  // file.write(&gif).unwrap();
+}