Build data to rayon

[doubleailes]

PR Type

Enhancement

---

Description

• Refactored rendering logic for improved parallelism and efficiency

  • Collects rays per sample, processes in parallel, and accumulates color per pixel
  • Progress bar now tracks per-ray progress instead of per-row

• Added set_mut_pixel method to buffer for color accumulation

• Simplified and updated sample scene RON file structure

  • Flatter camera/object/material definitions and new frame setting

• Minor code formatting and clarity improvements

---

Changes walkthrough 📝

	Relevant files
Enhancement	buffer.rs Add pixel color accumulation method to buffer                        crates/crust-render/src/buffer.rs Added set_mut_pixel method for in-place pixel color accumulation Documented new method for clarity +15/-0    tracer.rs Refactor rendering for parallelism and color accumulation crates/crust-render/src/tracer.rs Refactored rendering loop to collect all rays and process in parallel Progress bar now tracks per-ray instead of per-row Uses new set_mut_pixel for accumulating color per pixel Improved readability and efficiency of ray color calculation +32/-28  teapot.ron Simplify and update sample scene RON structure                      samples/teapot.ron Simplified RON structure for camera, objects, and materials Added smooth: true to teapot object Added new frame setting to settings Flattened vector/matrix definitions for clarity +18/-48
Formatting	instance.rs Code formatting for normal transformation                                crates/crust-render/src/instance.rs Reformatted code for clarity in normal transformation logic No functional changes, only formatting +5/-1      main.rs Minor formatting in render mode selection                                crates/crust-render/src/main.rs Minor formatting: consistent spacing in conditional +1/-1

---

Need help?

Type /help how to ... in the comments thread for any questions about Qodo Merge usage.

Check out the documentation for more information.

[qodo-code-review]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Memory Usage The refactored rendering code collects all rays upfront in a vector before processing, which could lead to high memory usage for large scenes or high sample counts. Consider if a chunked approach might be more memory-efficient. let mut d: Vec<(usize, usize, Ray)> = Vec::with_capacity(pixels_count); for j in (0..self.settings.height).rev() { for i in 0..self.settings.width { for sample in 0..self.settings.samples_per_pixel { let (dx, dy) = if (sample as usize) < cmj_samples.len() { cmj_samples[sample as usize] } else { (utils::random(), utils::random()) }; let u = (i as f32 + dx) / (self.settings.width - 1) as f32; let v = (j as f32 + dy) / (self.settings.height - 1) as f32; let ray = self.camera.get_ray(u, v); d.push((i, j, ray)); } } } Progress Bar The progress bar is declared but not initialized with a value. The code references 'bar' on line 40 but the initialization is missing, which could cause a compilation error. let bar = ProgressBar::new(pixels_count as u64); bar.set_style(

[qodo-code-review]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Reduce memory usage The memory usage is excessive because all rays are stored in memory before processing. For high-resolution renders with many samples, this could lead to out-of-memory errors. Consider processing rays in batches or chunks to maintain parallelism while reducing memory overhead. crates/crust-render/src/tracer.rs [47-64] -let mut d: Vec<(usize, usize, Ray)> = Vec::with_capacity(pixels_count); +let chunk_size = 1024; // Process rays in manageable chunks +let mut buffer = Buffer::new(self.settings.width, self.settings.height); + for j in (0..self.settings.height).rev() { - for i in 0..self.settings.width { - for sample in 0..self.settings.samples_per_pixel { - let (dx, dy) = if (sample as usize) < cmj_samples.len() { - cmj_samples[sample as usize] - } else { - (utils::random(), utils::random()) - }; - - let u = (i as f32 + dx) / (self.settings.width - 1) as f32; - let v = (j as f32 + dy) / (self.settings.height - 1) as f32; - - let ray = self.camera.get_ray(u, v); - d.push((i, j, ray)); + for i_chunk in (0..self.settings.width).step_by(chunk_size) { + let mut rays = Vec::with_capacity(chunk_size * self.settings.samples_per_pixel as usize); + + let i_end = (i_chunk + chunk_size).min(self.settings.width); + for i in i_chunk..i_end { + for sample in 0..self.settings.samples_per_pixel { + let (dx, dy) = if (sample as usize) < cmj_samples.len() { + cmj_samples[sample as usize] + } else { + (utils::random(), utils::random()) + }; + + let u = (i as f32 + dx) / (self.settings.width - 1) as f32; + let v = (j as f32 + dy) / (self.settings.height - 1) as f32; + + let ray = self.camera.get_ray(u, v); + rays.push((i, j, ray)); + } + } + + // Process this chunk in parallel + let colors: Vec<(usize, usize, Vec3A)> = rays + .into_par_iter() + .map(|(i, j, r)| { + let col = ray_color( + &r, + &self.world, + &self.lights, + self.settings.max_depth as i32, + ); + bar.inc(1); + (i, j, col) + }) + .collect(); + + // Accumulate colors for this chunk + for (i, j, col) in colors { + buffer.set_mut_pixel(i, j, col / self.settings.samples_per_pixel as f32); } } } +bar.finish(); +return buffer; + Apply this suggestion Suggestion importance[1-10]: 8 __ Why: This suggestion addresses a significant memory optimization issue. Processing all rays at once could lead to excessive memory usage for high-resolution renders. The chunked approach maintains parallelism while significantly reducing peak memory consumption, which could prevent out-of-memory errors in production.	Medium
More