Introduce pre-chain stage with area downsampling pass

[K1ngst0m]

User description

Why

The --app-width / --app-height CLI options currently only work with WSI proxy mode. Users need a way to control the source resolution fed into the RetroArch filter chain regardless of capture mode. A pre-chain downsampling stage enables resolution control for shader effects that benefit from lower-res input (CRT simulation, pixel art upscalers) without requiring WSI proxy overhead.

What Changes

• Add downsample.frag.slang shader in shaders/internal/ using area filtering for high-quality downsampling
• Introduce PreChain infrastructure in FilterChain to run internal passes before RetroArch shader passes
• Add DownsamplePass as the first pre-chain pass when --app-width/--app-height are specified
• Reuse existing FilterPass/Framebuffer infrastructure for pre-chain passes
• Change --app-width/--app-height semantics: these options now set the source resolution for the filter chain input (applies to all capture modes, not just WSI proxy)
• Store configured source resolution in Config::Capture for use by FilterChain

Impact

• Affected specs: render-pipeline
• Affected code:
  • shaders/internal/downsample.frag.slang (new)
  • src/render/chain/filter_chain.hpp - add pre-chain pass storage
  • src/render/chain/filter_chain.cpp - implement pre-chain recording
  • src/app/cli.cpp - update option descriptions
  • src/util/config.hpp - add source resolution fields

---

PR Type

Enhancement

---

Description

• Introduce extensible pre-chain pass infrastructure in FilterChain

• Add DownsamplePass with area filtering for resolution control

• Extend --app-width/--app-height to work in all capture modes

• Support single-dimension specification with aspect-ratio preservation

• Propagate source resolution through Config and RenderSettings

---

Diagram Walkthrough

flowchart LR
  CLI["CLI Options<br/>--app-width/height"]
  Config["Config::Render<br/>source_width/height"]
  RenderSettings["RenderSettings<br/>source_width/height"]
  VulkanBackend["VulkanBackend<br/>m_source_resolution"]
  FilterChain["FilterChain<br/>m_prechain_passes"]
  DownsamplePass["DownsamplePass<br/>Area Filter"]
  Output["Output to<br/>RetroArch Chain"]
  
  CLI -->|"parse"| Config
  Config -->|"propagate"| RenderSettings
  RenderSettings -->|"pass to create"| VulkanBackend
  VulkanBackend -->|"pass to create"| FilterChain
  FilterChain -->|"instantiate"| DownsamplePass
  DownsamplePass -->|"output becomes<br/>original_view"| Output

Loading

File Walkthrough

	Relevant files
Enhancement	11 files downsample_pass.hpp New DownsamplePass class header definition                              +61/-0    downsample_pass.cpp Implement area-filter downsampling pass logic                        +344/-0  filter_chain.hpp Add pre-chain pass and framebuffer vectors                              +18/-1    filter_chain.cpp Implement generic pre-chain recording infrastructure          +165/-18 vulkan_backend.hpp Add source_resolution field to RenderSettings                        +3/-0      vulkan_backend.cpp Propagate source resolution through backend initialization +8/-4      application.cpp Pass source resolution to RenderSettings                                  +2/-0      cli.cpp Update CLI option descriptions and remove validation          +4/-12    main.cpp Parse and propagate source resolution from CLI                      +6/-0      config.hpp Add source_width/height fields to Config::Render                  +2/-0      downsample.frag.slang New area-filter downsampling shader implementation              +73/-0
Formatting	1 files vulkan_debug.cpp Update debug log prefix from Vulkan to VVL                              +4/-4
Configuration changes	1 files CMakeLists.txt Add downsample_pass.cpp to build targets                                  +1/-0
Documentation	3 files proposal.md Document pre-chain and downsample feature proposal              +38/-0    spec.md Define requirements for pre-chain and downsample                  +93/-0    tasks.md Track implementation tasks for pre-chain feature                  +41/-0

---

Summary by CodeRabbit

• New Features

  • Pre-chain downsampling added to the render pipeline with area filtering for improved scaling quality and performance.
  • Configure source resolution using --app-width / --app-height; single-dimension input preserves aspect ratio and is supported at startup.

• Tests

  • CLI tests updated to allow single-dimension app width/height parsing.

• Chores

  • Render/config storage updated so configured source resolution is applied at launch.

• Style

  • Vulkan log prefix shortened for clearer diagnostics.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

Caution

Review failed

The pull request is closed.

📝 Walkthrough

Walkthrough

Adds an optional pre-chain downsampling stage to the render pipeline. A new DownsamplePass and area-filter shader run before the RetroArch passes when a source resolution is configured; FilterChain and backend/config/CLI are extended to create, record, and thread pre-chain passes and framebuffers into the main chain.

Changes

Cohort / File(s)	Summary
Shader & Pass Implementation shaders/internal/downsample.frag.slang, src/render/chain/downsample_pass.hpp, src/render/chain/downsample_pass.cpp, src/render/chain/CMakeLists.txt	New area-weighted downsample fragment shader and a Vulkan DownsamplePass with sampler/descriptor/pipeline creation, push-constants, per-frame descriptor updates and record path; build list updated.
FilterChain Pre-Chain Infrastructure src/render/chain/filter_chain.hpp, src/render/chain/filter_chain.cpp	Added m_source_resolution, pre-chain vectors (m_prechain_passes, m_prechain_framebuffers), changed create() signature to accept source resolution, added ensure_prechain_passes() and record_prechain(), integrated pre-chain execution and cleanup into recording flow.
Configuration & Backend Plumbing src/util/config.hpp, src/render/backend/vulkan_backend.hpp, src/render/backend/vulkan_backend.cpp	Added Config::Render::source_width/height and RenderSettings::source_width/height; VulkanBackend stores m_source_resolution and forwards it into FilterChain creation and shader-reload paths.
CLI & Application Integration src/app/cli.cpp, src/app/application.cpp, src/app/main.cpp, tests/app/test_cli.cpp	CLI/help updated to treat --app-width/--app-height as source resolution (single-dimension allowed); main applies CLI overrides into config.render; tests updated to accept single-dimension parsing.
Backend Debug & Logging src/render/backend/vulkan_debug.cpp	Log prefix changed from "[Vulkan]" to "[VVL]" in Vulkan debug callback messages.
Specs & Tasks openspec/changes/add-prechain-downsample/*	Proposal, spec, and tasks documenting pre-chain design, downsample behavior, CLI semantics (aspect-ratio fallback), and testing plan.

Sequence Diagram

sequenceDiagram
    participant App as Application
    participant Backend as VulkanBackend
    participant Chain as FilterChain
    participant Pass as DownsamplePass
    participant GPU as Vulkan GPU

    App->>App: Parse CLI (--app-width/--app-height)
    App->>Backend: init_filter_chain(source_resolution)
    Backend->>Chain: create(..., source_resolution)
    Chain->>Chain: store m_source_resolution
    alt source_resolution non-zero
        Chain->>Pass: create(DownsamplePass)
        Pass->>GPU: Allocate sampler & descriptors
        Pass->>GPU: Create pipeline layout & graphics pipeline
        Chain->>Chain: add pass to m_prechain_passes
    end

    rect rgba(100,150,200,0.5)
    Note over Chain,GPU: Per-frame rendering
    App->>Chain: record(cmd_buffer, frame_index)
    Chain->>Chain: ensure_prechain_passes(captured_extent)
    Chain->>Chain: record_prechain(original_view, original_extent, frame_index)
    Chain->>Pass: record(cmd, PassContext)
    Pass->>GPU: Bind descriptor with source texture
    Pass->>GPU: Push downsample constants
    Pass->>GPU: Draw area-filtered downsample
    Pass-->>Chain: Return downsampled image view/extent
    Chain->>Chain: Use prechain output as original_view for RetroArch passes
    Chain->>GPU: Record downstream passes with downsampled input
    end

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~50 minutes

Possibly related PRs

• Introduce pre-chain stage with area downsampling pass #60 — very closely related: introduces DownsamplePass, downsample shader and pre-chain members in FilterChain (near-identical changes).
• Use VK_KHR_present_wait for Frame Pacing #47 — overlaps VulkanBackend API and RenderSettings propagation that this change extends for source-resolution forwarding.
• feat(app): refactor CLI parsing #57 — touches CLI parsing and tests for --app-width/--app-height, related to the single-dimension allowance here.

Poem

🐰 Hops through shaders, soft and spry,

I shrink each frame without a cry,
A pre-chain pass, a gentle art,
Pixels tumble, trimmed apart,
Thump-thump—rendering, neat and spry.

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 6.06% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately and concisely summarizes the main change: introducing a pre-chain stage with area downsampling capability to the render pipeline.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[qodo-code-review]

PR Compliance Guide 🔍

Below is a summary of compliance checks for this PR:

Security Compliance
⚪	Out-of-bounds indexing Description: update_descriptor() indexes m_descriptor_sets[frame_index] without validating that frame_index < m_num_sync_indices, which could cause out-of-bounds access and a crash (denial-of-service) if an unexpected/invalid frame index is ever passed to the pass. downsample_pass.cpp [63-78] Referred Code void DownsamplePass::update_descriptor(uint32_t frame_index, vk::ImageView source_view) { vk::DescriptorImageInfo image_info{}; image_info.sampler = *m_sampler; image_info.imageView = source_view; image_info.imageLayout = vk::ImageLayout::eShaderReadOnlyOptimal; vk::WriteDescriptorSet write{}; write.dstSet = m_descriptor_sets[frame_index]; write.dstBinding = 0; write.dstArrayElement = 0; write.descriptorCount = 1; write.descriptorType = vk::DescriptorType::eCombinedImageSampler; write.pImageInfo = &image_info; m_device.updateDescriptorSets(write, {}); }
Ticket Compliance
⚪	🎫 No ticket provided Create ticket/issue
Codebase Duplication Compliance
⚪	Codebase context is not defined Follow the guide to enable codebase context checks.
Custom Compliance
🟢	Generic: Comprehensive Audit Trails Objective: To create a detailed and reliable record of critical system actions for security analysis and compliance. Status: Passed Learn more about managing compliance generic rules or creating your own custom rules
	Generic: Meaningful Naming and Self-Documenting Code Objective: Ensure all identifiers clearly express their purpose and intent, making code self-documenting Status: Passed Learn more about managing compliance generic rules or creating your own custom rules
	Generic: Secure Error Handling Objective: To prevent the leakage of sensitive system information through error messages while providing sufficient detail for internal debugging. Status: Passed Learn more about managing compliance generic rules or creating your own custom rules
	Generic: Security-First Input Validation and Data Handling Objective: Ensure all data inputs are validated, sanitized, and handled securely to prevent vulnerabilities Status: Passed Learn more about managing compliance generic rules or creating your own custom rules
🔴	Generic: Robust Error Handling and Edge Case Management Objective: Ensure comprehensive error handling that provides meaningful context and graceful degradation Status: 🏷️ Zero division risk: Aspect-ratio computation divides by captured_extent.width/height without guarding against zero-sized captured frames, which can crash or produce invalid resolutions. Referred Code vk::Extent2D target_resolution = m_source_resolution; if (target_resolution.width == 0) { target_resolution.width = static_cast<uint32_t>(std::round(static_cast<float>(target_resolution.height) * static_cast<float>(captured_extent.width) / static_cast<float>(captured_extent.height))); target_resolution.width = std::max(1U, target_resolution.width); } else if (target_resolution.height == 0) { target_resolution.height = static_cast<uint32_t>(std::round(static_cast<float>(target_resolution.width) * static_cast<float>(captured_extent.height) / static_cast<float>(captured_extent.width))); target_resolution.height = std::max(1U, target_resolution.height); Learn more about managing compliance generic rules or creating your own custom rules
⚪	Generic: Secure Logging Practices Objective: To ensure logs are useful for debugging and auditing without exposing sensitive information like PII, PHI, or cardholder data. Status: 🏷️ Unstructured log format: New log lines are plain-text (not clearly structured/JSON), so it is unclear whether the logging backend enforces structured logging for auditing as required. Referred Code GOGGLES_LOG_INFO("FilterChain pre-chain enabled: {}x{}", source_resolution.width, source_resolution.height); } else if (source_resolution.width > 0 || source_resolution.height > 0) { GOGGLES_LOG_INFO("FilterChain pre-chain pending: width={}, height={}", source_resolution.width, source_resolution.height); } Learn more about managing compliance generic rules or creating your own custom rules
Update

Compliance status legend

🟢 - Fully Compliant
🟡 - Partial Compliant
🔴 - Not Compliant
⚪ - Requires Further Human Verification
🏷️ - Compliance label

[qodo-code-review]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Improve downsampling quality for large ratios Modify the downsampling shader to use a strided loop instead of truncating the sampling area for large downscaling ratios. This improves image quality by ensuring the entire source pixel area is considered. shaders/internal/downsample.frag.slang [38-65] // Limit sample count for performance (max 8x8 = 64 samples) int max_samples = 8; -int2 sample_range = min(i_max - i_min, int2(max_samples, max_samples)); -i_max = i_min + sample_range; +int2 full_range = i_max - i_min; +int2 step = max(1, full_range / max_samples); +float2 step_weight_multiplier = float2(step); float4 accum = float4(0.0, 0.0, 0.0, 0.0); float total_weight = 0.0; // Area filter: weight each source pixel by its coverage -for (int y = i_min.y; y < i_max.y; y++) { - for (int x = i_min.x; x < i_max.x; x++) { +for (int y = i_min.y; y < i_max.y; y += step.y) { + for (int x = i_min.x; x < i_max.x; x += step.x) { // Calculate this pixel's contribution (overlap with the box) float2 pixel_min = float2(x, y); - ... + float2 pixel_max = float2(x + step.x, y + step.y); + + float2 overlap_min = max(pixel_min, box_min); + float2 overlap_max = min(pixel_max, box_max); + float2 overlap_size = max(overlap_max - overlap_min, float2(0.0, 0.0)); + float weight = overlap_size.x * overlap_size.y; if (weight > 0.0) { // Sample at pixel center float2 sample_uv = (float2(x, y) + 0.5) * pc.source_size.zw; accum += source_texture.Sample(sample_uv) * weight; total_weight += weight; } } } [To ensure code accuracy, apply this suggestion manually] Suggestion importance[1-10]: 8 __ Why: The suggestion correctly identifies a flaw in the downsampling shader's logic that causes incorrect rendering for large downscaling ratios by truncating the sample area, and proposes a valid fix using a strided loop that improves output quality.	Medium
	Prevent division by zero during initialization In ensure_prechain_passes, add checks to prevent division-by-zero when captured_extent.width or captured_extent.height is zero during aspect ratio calculation. src/render/chain/filter_chain.cpp [738-779] auto FilterChain::ensure_prechain_passes(vk::Extent2D captured_extent) -> Result<void> { if (!m_prechain_passes.empty() && !m_prechain_framebuffers.empty()) { return {}; } if (m_source_resolution.width == 0 && m_source_resolution.height == 0) { return {}; } vk::Extent2D target_resolution = m_source_resolution; if (target_resolution.width == 0) { + if (captured_extent.height == 0) { + return {}; // Avoid division by zero + } target_resolution.width = static_cast<uint32_t>(std::round(static_cast<float>(target_resolution.height) * static_cast<float>(captured_extent.width) / static_cast<float>(captured_extent.height))); target_resolution.width = std::max(1U, target_resolution.width); } else if (target_resolution.height == 0) { + if (captured_extent.width == 0) { + return {}; // Avoid division by zero + } target_resolution.height = static_cast<uint32_t>(std::round(static_cast<float>(target_resolution.width) * static_cast<float>(captured_extent.height) / static_cast<float>(captured_extent.width))); target_resolution.height = std::max(1U, target_resolution.height); } m_source_resolution = target_resolution; DownsamplePassConfig downsample_config{ .target_format = vk::Format::eR8G8B8A8Unorm, .num_sync_indices = m_num_sync_indices, .shader_dir = m_shader_dir, }; m_prechain_passes.push_back( GOGGLES_TRY(DownsamplePass::create(m_vk_ctx, *m_shader_runtime, downsample_config))); m_prechain_framebuffers.push_back(GOGGLES_TRY(Framebuffer::create( m_vk_ctx.device, m_vk_ctx.physical_device, vk::Format::eR8G8B8A8Unorm, target_resolution))); GOGGLES_LOG_INFO("FilterChain pre-chain initialized (aspect-ratio): {}x{} (from {}x{})", target_resolution.width, target_resolution.height, captured_extent.width, captured_extent.height); return {}; } Apply / Chat Suggestion importance[1-10]: 8 __ Why: This suggestion correctly identifies a potential division-by-zero crash in ensure_prechain_passes when calculating aspect ratio with a zero-dimension captured frame and provides a necessary check to prevent it.	Medium
	Fix pre-chain barrier access flags In record_prechain, fix the pre-barrier by setting srcAccessMask to empty when transitioning the image layout from eUndefined to avoid an invalid access flag. src/render/chain/filter_chain.cpp [363-367] vk::ImageMemoryBarrier pre_barrier{}; -pre_barrier.srcAccessMask = vk::AccessFlagBits::eShaderRead; +pre_barrier.srcAccessMask = {}; // no prior access on undefined layout pre_barrier.dstAccessMask = vk::AccessFlagBits::eColorAttachmentWrite; pre_barrier.oldLayout = vk::ImageLayout::eUndefined; pre_barrier.newLayout = vk::ImageLayout::eColorAttachmentOptimal; Apply / Chat Suggestion importance[1-10]: 7 __ Why: The suggestion correctly identifies that the srcAccessMask for an image layout transition from eUndefined should be empty, not eShaderRead. This is a valid Vulkan validation layer fix that improves correctness.	Medium
General	Ensure last source extent is correct In the record function, update m_last_source_extent with effective_original_extent after the pre-chain processing. This ensures it accurately reflects the source dimensions for the main filter chain. src/render/chain/filter_chain.cpp [413-564] void FilterChain::record(vk::CommandBuffer cmd, vk::Image original_image, vk::ImageView original_view, vk::Extent2D original_extent, vk::ImageView swapchain_view, vk::Extent2D viewport_extent, uint32_t frame_index, ScaleMode scale_mode, uint32_t integer_scale) { GOGGLES_PROFILE_FUNCTION(); m_last_scale_mode = scale_mode; m_last_integer_scale = integer_scale; - m_last_source_extent = original_extent; GOGGLES_MUST(ensure_prechain_passes(original_extent)); auto [effective_original_view, effective_original_extent] = record_prechain(cmd, original_view, original_extent, frame_index); + + m_last_source_extent = effective_original_extent; GOGGLES_MUST(ensure_frame_history(effective_original_extent)); if (m_passes.empty() || m_bypass_enabled.load(std::memory_order_relaxed)) { ... if (m_frame_history.is_initialized()) { auto history_image = !m_prechain_framebuffers.empty() ? m_prechain_framebuffers.back()->image() : original_image; m_frame_history.push(cmd, history_image, effective_original_extent); } copy_feedback_framebuffers(cmd); m_frame_count++; } [To ensure code accuracy, apply this suggestion manually] Suggestion importance[1-10]: 6 __ Why: The suggestion correctly points out that m_last_source_extent should be updated with the post-pre-chain effective_original_extent for logical consistency, preventing potential bugs if other code relies on this member variable.	Low
	Clarify new CLI semantics Update the CLI help footer to clarify that --app-width and --app-height can be used individually to control the source resolution for the new pre-chain downsampling. src/app/cli.cpp [124-130] app.footer(R"(Usage: goggles --detach goggles [options] -- <app> [app_args...] Notes: - Default mode (no --detach) launches the target app with capture + input forwarding enabled. - - '--' is required before <app> to avoid app args (e.g. '--config') being parsed as Goggles options.)"); + - '--' is required before <app> to avoid app args (e.g. '--config') being parsed as Goggles options. + - `--app-width`/`--app-height` now control source resolution and may be specified alone)"); [To ensure code accuracy, apply this suggestion manually] Suggestion importance[1-10]: 4 __ Why: The suggestion correctly identifies that the CLI help text is missing information about the new behavior of --app-width/--app-height. Updating the documentation improves user experience and discoverability.	Low
	Log only complete resolutions In run_app, modify the logging statement to only print the source resolution when both width and height are non-zero, avoiding misleading logs like "Source resolution: 640x0". src/app/main.cpp [375-380] if (cli_opts.app_width != 0 || cli_opts.app_height != 0) { config.render.source_width = cli_opts.app_width; config.render.source_height = cli_opts.app_height; - GOGGLES_LOG_INFO("Source resolution: {}x{}", config.render.source_width, - config.render.source_height); + if (config.render.source_width != 0 && config.render.source_height != 0) { + GOGGLES_LOG_INFO("Source resolution: {}x{}", config.render.source_width, + config.render.source_height); + } } Apply / Chat Suggestion importance[1-10]: 3 __ Why: The suggestion correctly points out that logging a resolution with a zero dimension can be misleading. While a minor issue, conditionalizing the log message improves clarity.	Low
Update

[coderabbitai]

Actionable comments posted: 2

🤖 Fix all issues with AI agents

In `@openspec/changes/add-prechain-downsample/specs/render-pipeline/spec.md`:
- Around line 80-86: The scenario example's computed width is off by one
compared to the implementation: update the spec to match the code's rounding
behavior (std::round) by changing the expected width from 854 to 853 (since
round(480 * 1920 / 1080) = round(853.333...) = 853), or explicitly note that
implementations using std::round will produce 853; reference the implementation
in filter_chain.cpp and the use of std::round when making this change.

In `@src/render/chain/downsample_pass.cpp`:
- Around line 85-93: The code computes inverses for source and target dimensions
in DownsamplePushConstants
(pc.source_inv_width/pc.source_inv_height/pc.target_inv_width/pc.target_inv_height)
without checking for zero; add a defensive guard around ctx.source_extent and
ctx.output_extent width/height: if a dimension is zero (or extremely close to
zero) set the corresponding inverse to 0.0f (or 1.0f/epsilon) and/or
early-return/skip the downsample dispatch, otherwise compute 1.0f/width as
before; update all uses of pc.* to assume these safe values.

🧹 Nitpick comments (3)

shaders/internal/downsample.frag.slang (1)

38-41: Consider centering the sample window for extreme downsampling ratios.

When the source region exceeds 8×8 pixels, the current approach truncates from the end (i_max = i_min + sample_range), which may skip high-overlap pixels at the upper boundary. For extreme downsampling ratios, centering the 8×8 window on the box midpoint could improve quality:

     int max_samples = 8;
     int2 sample_range = min(i_max - i_min, int2(max_samples, max_samples));
-    i_max = i_min + sample_range;
+    int2 excess = (i_max - i_min) - sample_range;
+    i_min = i_min + excess / 2;
+    i_max = i_min + sample_range;

This is a minor quality improvement for edge cases with large scale factors.

src/app/main.cpp (1)

375-380: Consider clarifying the log message for single-dimension specification.

When only one dimension is specified (e.g., --app-width 640), the log will show Source resolution: 640x0, which may confuse users. Consider logging only the specified dimension or indicating that the other will be computed:

     if (cli_opts.app_width != 0 || cli_opts.app_height != 0) {
         config.render.source_width = cli_opts.app_width;
         config.render.source_height = cli_opts.app_height;
-        GOGGLES_LOG_INFO("Source resolution: {}x{}", config.render.source_width,
-                         config.render.source_height);
+        if (config.render.source_width != 0 && config.render.source_height != 0) {
+            GOGGLES_LOG_INFO("Source resolution: {}x{}", config.render.source_width,
+                             config.render.source_height);
+        } else if (config.render.source_width != 0) {
+            GOGGLES_LOG_INFO("Source width: {} (height from aspect ratio)",
+                             config.render.source_width);
+        } else {
+            GOGGLES_LOG_INFO("Source height: {} (width from aspect ratio)",
+                             config.render.source_height);
+        }
     }

This is a minor UX improvement for log clarity.

src/util/config.hpp (1)

67-68: Consider adding documentation comments for the new fields.

The other fields in Config::Render are self-explanatory, but the semantics of source_width/source_height (pre-chain downsampling target, 0 = disabled) could benefit from brief documentation:

         uint32_t integer_scale = 0;
+        uint32_t source_width = 0;  ///< Target width for pre-chain downsampling (0 = disabled)
+        uint32_t source_height = 0; ///< Target height for pre-chain downsampling (0 = disabled)
-        uint32_t source_width = 0;
-        uint32_t source_height = 0;