test(M00): add camera pose sweep

This commit is contained in:
2026-07-12 01:46:02 +04:00
parent 8d4641a43b
commit b48195d08d
5 changed files with 255 additions and 2 deletions
@@ -0,0 +1,78 @@
# M00-QAR-CAMERA-POSE-SWEEP-001 — Empirical camera pose sweep
<!-- OPENWC_CLAIM:M00-QAR-CAMERA-POSE-SWEEP-001:sindo-main-codex:2026-07-14 -->
## Ownership
- Target: M00
- Program: QAR
- Owner/Agent ID: sindo-main-codex
- Branch: `work/sindo-main-codex/m00-camera-pose-sweep`
- Lease expires UTC: 2026-07-14
- Integrator: milestone integrator
## Outcome
Recover reproducible checkpoint framing by sweeping bounded yaw/pitch offsets and ranking paired-image error.
## Non-goals
- Changing runtime player camera defaults.
- Claiming an exact original-client camera pose from perceptual score alone.
- Adding computer-vision registration or image warping.
## Paths
- Exclusive: camera-pose CLI and sweep orchestration
- Shared/hotspots: renderer baseline documentation
- Generated/ignored: local candidate PNGs and JSON reports
## Contracts and data
- Public API/events: additive camera yaw/pitch offset and single-pass capture options; new PowerShell sweep command
- Schema/format version: capture report schema remains version 1 with additive camera-offset fields
- Migration/compatibility: existing capture commands retain zero offsets and two passes
- Consumers: M00 paired-fidelity workflow
## Dependencies
- Requires: build 12340 reference image, capture tool, paired-image comparator
- Blocks: reproducible manual-reference framing
- External state: original screenshots remain outside Git
## Verification
- Commands: capture dry-run offset regression, comparator self-test, bounded local sweep when a display/reference is available, repository gates
- Fixtures: one named M00 checkpoint
- Fidelity evidence: ranked yaw/pitch candidates with explicit human-approval requirement
- Performance budget: offline diagnostic; single-pass mode avoids redundant warm capture
## Documentation deliverables
- Inline public API docs: capture and sweep usage
- Module specification: camera-pose data flow, verification and source map
- Data-flow diagram: updated pose-sweep flow
- Sequence/state/dependency diagrams: unchanged
- Source map/status updates: baseline workflow and findings
## Simplicity and naming
- Important names introduced: `camera_yaw_offset_degrees`, `camera_pitch_offset_degrees`, `single_pass`
- Simplest considered solution: bounded grid over existing capture/comparator tools
- Rejected complexity/abstractions: feature matching, optimizer framework, image transforms
- Unavoidable complexity and justification: original-client camera angles are not exposed by tested APIs
- Measured optimization evidence: single-pass mode halves captures per candidate
## Status
- State: ready-for-review
- Done: additive capture offsets, single-pass calibration mode, bounded grid runner, ranked JSON contract, grid-plan regression and documentation
- Next: run a coarse-to-fine real sweep with the private build 12340 reference directory, then obtain human framing approval
- Blocked by:
## Handoff
- Commit: this work-package commit
- Results: dry-run reported yaw 12.50/pitch -7.50; comparator self-test passed; a 3x2 plan produced six unique candidates; full seven-checkpoint M00 dry-run passed with default zero offsets
- Remaining risks: no private reference directory was available in this worktree, so real perceptual ranking and human approval remain external evidence
- Documentation updated: `docs/CAMERA_POSE_SWEEP.md`; renderer module verification, risk and source map
+39
View File
@@ -0,0 +1,39 @@
# Renderer Camera Pose Sweep
This offline M00 diagnostic recovers reproducible checkpoint framing when an original-client screenshot has no recorded camera yaw or pitch. It ranks a bounded grid; it does not change renderer or player-camera defaults and does not prove the exact build 12340 camera pose without human approval.
## Camera contract
`capture_render_checkpoints.gd` accepts additive `--camera-yaw-offset` and `--camera-pitch-offset` values in degrees. Yaw rotates the no-roll checkpoint basis around Godot world `Vector3.UP`; pitch then rotates around the camera-local right axis. Zero offsets preserve the manifest target exactly. `--single-pass` captures only `cold_process` and is intended for pose calibration, not performance baselines.
## Workflow
```powershell
.\tools\sweep_render_checkpoint_camera_pose.ps1 `
-ReferenceDirectory 'D:\private-fixtures\wow-3.3.5a-checkpoints' `
-Checkpoint elwynn_adt_boundary `
-YawOffsets -15,-10,-5,0,5,10,15 `
-PitchOffsets -10,-5,0,5,10 `
-WaitSeconds 2
```
Each candidate receives its own output directory and comparison report. `ranking.json` sorts candidates by mean perceptual error and then changed-pixel ratio. Use `-PlanOnly` to validate the Cartesian grid and output paths without rendering. Run a coarse grid first, then a finer grid around the best candidate.
```mermaid
flowchart LR
Y[Yaw offsets in degrees] --> G[Bounded Cartesian grid]
P[Pitch offsets in degrees] --> G
G --> C[Single-pass checkpoint captures]
R[Original-client reference] --> D[Perceptual comparator]
C --> D
D --> J[ranking.json]
J --> H[Human framing approval]
```
## Interpretation and recovery
The smallest error is a candidate, not automatic fidelity approval. Geometry, materials, lighting and FOV can move the perceptual optimum away from the true original-client pose. Inspect the best images manually and retain the original reference outside Git. A missing matching reference or candidate is a hard error; correct the checkpoint filter or filenames and rerun. Generated PNG and JSON outputs are disposable local evidence.
## Verification
The headless capture dry-run must report requested yaw and pitch offsets. Comparator `--self-test` covers paired metrics. Sweep `-PlanOnly` covers deterministic grid expansion and output naming without requiring a display. A real ranked sweep requires a display and a private build 12340 reference image.
+4 -1
View File
@@ -195,7 +195,7 @@ Exact exported settings and cache versions remain documented in [`../../RENDER.m
## Verification ## Verification
- Unit/contract tests: material mapping, unique-ID dedupe, placement probes, baseline manifest, five-point coordinate calibration, synthetic perceptual checkpoint diff. - Unit/contract tests: material mapping, unique-ID dedupe, placement probes, baseline manifest, five-point coordinate calibration, synthetic perceptual checkpoint diff, camera-pose grid plan.
- Integration/E2E: Eastern Kingdoms/Kalimdor streaming scenes and seven cold/warm checkpoints. - Integration/E2E: Eastern Kingdoms/Kalimdor streaming scenes and seven cold/warm checkpoints.
- Fidelity evidence: пять локальных build 12340 reference JPG откалибровали terrain/ADT/M2/WMO/liquid viewpoints; automated paired-image metrics exist, но synthetic animation/dusk и полный human approval ещё не закрыты. - Fidelity evidence: пять локальных build 12340 reference JPG откалибровали terrain/ADT/M2/WMO/liquid viewpoints; automated paired-image metrics exist, но synthetic animation/dusk и полный human approval ещё не закрыты.
- Performance budgets: M00 report records cold/warm p95 and max hitch; no final acceptance threshold yet. - Performance budgets: M00 report records cold/warm p95 and max hitch; no final acceptance threshold yet.
@@ -229,6 +229,7 @@ Exact exported settings and cache versions remain documented in [`../../RENDER.m
- Terrain-height probe исключил under-terrain состояние для всех пяти точек; waterfall exact-XZ miss классифицирован как TriangleMesh seam/edge и подтверждён nearby sample в 2 units. - Terrain-height probe исключил under-terrain состояние для всех пяти точек; waterfall exact-XZ miss классифицирован как TriangleMesh seam/edge и подтверждён nearby sample в 2 units.
- Camera-occluder probe не нашёл camera containment в пяти точках; paired mismatch локализован прежде всего в manual look direction/target/FOV calibration, с явным ограничением по RID-only geometry. - Camera-occluder probe не нашёл camera containment в пяти точках; paired mismatch локализован прежде всего в manual look direction/target/FOV calibration, с явным ограничением по RID-only geometry.
- Empirical FOV sweep выявил, что checkpoint camera должна явно вызывать `make_current()`; после исправления projection ranking остаётся inconclusive из-за неизвестного manual yaw/pitch/framing reference. - Empirical FOV sweep выявил, что checkpoint camera должна явно вызывать `make_current()`; после исправления projection ranking остаётся inconclusive из-за неизвестного manual yaw/pitch/framing reference.
- Camera-pose sweep can now rank bounded yaw/pitch grids without changing manifest defaults; perceptual ranking remains diagnostic and requires human framing approval.
- D3D12 descriptor issues remain; the capture-path anonymous `RefCounted` shutdown leak is regression-covered by a clean verbose dry-run, while other RID/resource diagnostics still require independent evidence. - D3D12 descriptor issues remain; the capture-path anonymous `RefCounted` shutdown leak is regression-covered by a clean verbose dry-run, while other RID/resource diagnostics still require independent evidence.
- M2/WMO/material/particle/ribbon/portal parity incomplete. - M2/WMO/material/particle/ribbon/portal parity incomplete.
- Public API is mostly exported configuration rather than stable contracts. - Public API is mostly exported configuration rather than stable contracts.
@@ -248,6 +249,7 @@ Exact exported settings and cache versions remain documented in [`../../RENDER.m
| `tools/run_render_baseline.ps1` | Unified M00 baseline runner | | `tools/run_render_baseline.ps1` | Unified M00 baseline runner |
| `src/tools/compare_render_checkpoints.gd` | Offline JPG/PNG paired-image perceptual metrics and JSON pass/fail report | | `src/tools/compare_render_checkpoints.gd` | Offline JPG/PNG paired-image perceptual metrics and JSON pass/fail report |
| `src/tools/capture_render_checkpoints.gd` | Deterministic no-roll checkpoint camera, performance and visual capture | | `src/tools/capture_render_checkpoints.gd` | Deterministic no-roll checkpoint camera, performance and visual capture |
| `tools/sweep_render_checkpoint_camera_pose.ps1` | Offline yaw/pitch capture grid and paired-error ranking |
| `src/tools/verify_render_coordinate_calibration.gd` | Build 12340 camera-coordinate golden point round-trip diagnostic | | `src/tools/verify_render_coordinate_calibration.gd` | Build 12340 camera-coordinate golden point round-trip diagnostic |
| `src/tools/probe_render_terrain_height.gd` | Offline active-mesh terrain height and camera-clearance report | | `src/tools/probe_render_terrain_height.gd` | Offline active-mesh terrain height and camera-clearance report |
| `src/tools/probe_render_camera_occluders.gd` | Scene-tree placement containment and camera-to-target AABB intersection report | | `src/tools/probe_render_camera_occluders.gd` | Scene-tree placement containment and camera-to-target AABB intersection report |
@@ -256,5 +258,6 @@ Exact exported settings and cache versions remain documented in [`../../RENDER.m
- [`../../RENDER.md`](../../RENDER.md) - [`../../RENDER.md`](../../RENDER.md)
- [`../RENDER_BASELINE.md`](../RENDER_BASELINE.md) - [`../RENDER_BASELINE.md`](../RENDER_BASELINE.md)
- [`../CAMERA_POSE_SWEEP.md`](../CAMERA_POSE_SWEEP.md)
- [`../../targets/roadmap/02-rendering-and-graphics.md`](../../targets/roadmap/02-rendering-and-graphics.md) - [`../../targets/roadmap/02-rendering-and-graphics.md`](../../targets/roadmap/02-rendering-and-graphics.md)
- [`../GODOT_BEST_PRACTICES.md`](../GODOT_BEST_PRACTICES.md) - [`../GODOT_BEST_PRACTICES.md`](../GODOT_BEST_PRACTICES.md)
+16 -1
View File
@@ -24,6 +24,9 @@ func _capture_async() -> void:
var revision := _arg(args, "--revision", "worktree") var revision := _arg(args, "--revision", "worktree")
var cache_state := _arg(args, "--cache-state", "existing") var cache_state := _arg(args, "--cache-state", "existing")
var camera_fov_override := float(_arg(args, "--camera-fov", str(manifest.get("camera_fov", 62.0)))) var camera_fov_override := float(_arg(args, "--camera-fov", str(manifest.get("camera_fov", 62.0))))
var camera_yaw_offset_degrees := float(_arg(args, "--camera-yaw-offset", "0.0"))
var camera_pitch_offset_degrees := float(_arg(args, "--camera-pitch-offset", "0.0"))
var single_pass := args.has("--single-pass")
var headless := DisplayServer.get_name().to_lower() == "headless" var headless := DisplayServer.get_name().to_lower() == "headless"
var dry_run := args.has("--dry-run") or headless var dry_run := args.has("--dry-run") or headless
var viewport: Array = manifest.get("viewport", [1280, 900]) var viewport: Array = manifest.get("viewport", [1280, 900])
@@ -86,6 +89,8 @@ func _capture_async() -> void:
"dry_run": dry_run, "dry_run": dry_run,
"cache_state": cache_state, "cache_state": cache_state,
"camera_fov": camera_fov_override, "camera_fov": camera_fov_override,
"camera_yaw_offset_degrees": camera_yaw_offset_degrees,
"camera_pitch_offset_degrees": camera_pitch_offset_degrees,
"environment": _environment_metadata(), "environment": _environment_metadata(),
"comparison_budgets": manifest.get("comparison_budgets", {}), "comparison_budgets": manifest.get("comparison_budgets", {}),
"cache_contract": manifest.get("cache_contract", {}), "cache_contract": manifest.get("cache_contract", {}),
@@ -94,6 +99,8 @@ func _capture_async() -> void:
} }
var captured := 0 var captured := 0
var passes := ["cold_process", "warm_revisit"] var passes := ["cold_process", "warm_revisit"]
if single_pass:
passes = ["cold_process"]
if dry_run: if dry_run:
passes = ["dry_run"] passes = ["dry_run"]
@@ -114,6 +121,7 @@ func _capture_async() -> void:
camera.global_position = _vector3(checkpoint.get("camera", [0.0, 0.0, 0.0])) camera.global_position = _vector3(checkpoint.get("camera", [0.0, 0.0, 0.0]))
_orient_camera_without_roll(camera, _vector3(checkpoint.get("target", [0.0, 0.0, 0.0]))) _orient_camera_without_roll(camera, _vector3(checkpoint.get("target", [0.0, 0.0, 0.0])))
_apply_camera_pose_offsets(camera, camera_yaw_offset_degrees, camera_pitch_offset_degrees)
camera.make_current() camera.make_current()
if player != null: if player != null:
player.global_position = _vector3(checkpoint.get("player", checkpoint.get("target", [0.0, 0.0, 0.0]))) player.global_position = _vector3(checkpoint.get("player", checkpoint.get("target", [0.0, 0.0, 0.0])))
@@ -122,11 +130,13 @@ func _capture_async() -> void:
world.call("_refresh_streaming_targets_at", camera.global_position, true) world.call("_refresh_streaming_targets_at", camera.global_position, true)
if dry_run: if dry_run:
print("RENDER_CHECKPOINT dry_run name=%s coverage=%s camera=%s target=%s time=%.2f" % [ print("RENDER_CHECKPOINT dry_run name=%s coverage=%s camera=%s target=%s yaw_offset=%.2f pitch_offset=%.2f time=%.2f" % [
checkpoint_name, checkpoint_name,
str(checkpoint.get("coverage", [])), str(checkpoint.get("coverage", [])),
camera.global_position, camera.global_position,
_vector3(checkpoint.get("target", [0.0, 0.0, 0.0])), _vector3(checkpoint.get("target", [0.0, 0.0, 0.0])),
camera_yaw_offset_degrees,
camera_pitch_offset_degrees,
float(checkpoint.get("time_hours", 13.0)), float(checkpoint.get("time_hours", 13.0)),
]) ])
(report["results"] as Array).append(_result_record(checkpoint, pass_name, 0.0, {}, "")) (report["results"] as Array).append(_result_record(checkpoint, pass_name, 0.0, {}, ""))
@@ -215,6 +225,11 @@ func _orient_camera_without_roll(camera: Camera3D, target_position: Vector3) ->
camera.global_basis = Basis(right, corrected_up, -forward) camera.global_basis = Basis(right, corrected_up, -forward)
func _apply_camera_pose_offsets(camera: Camera3D, yaw_offset_degrees: float, pitch_offset_degrees: float) -> void:
camera.global_rotate(Vector3.UP, deg_to_rad(yaw_offset_degrees))
camera.rotate_object_local(Vector3.RIGHT, deg_to_rad(pitch_offset_degrees))
func _result_record(checkpoint: Dictionary, pass_name: String, load_time_ms: float, metrics: Dictionary, sha256: String) -> Dictionary: func _result_record(checkpoint: Dictionary, pass_name: String, load_time_ms: float, metrics: Dictionary, sha256: String) -> Dictionary:
return { return {
"name": checkpoint.get("name", "checkpoint"), "name": checkpoint.get("name", "checkpoint"),
@@ -0,0 +1,118 @@
[CmdletBinding()]
param(
[Parameter(Mandatory = $true)]
[string]$ReferenceDirectory,
[Parameter(Mandatory = $true)]
[string]$Checkpoint,
[string]$GodotPath,
[string]$Output = "user://render_camera_pose_sweep",
[double[]]$YawOffsets = @(-10.0, 0.0, 10.0),
[double[]]$PitchOffsets = @(-10.0, 0.0, 10.0),
[double]$CameraFov = 62.0,
[double]$WaitSeconds = 2.0,
[double]$MeasureSeconds = 0.1,
[switch]$PlanOnly
)
$ErrorActionPreference = "Stop"
$repoRoot = Split-Path -Parent $PSScriptRoot
if (-not $GodotPath) {
$godotCommand = Get-Command godot -ErrorAction SilentlyContinue
if ($godotCommand) {
$GodotPath = $godotCommand.Source
} else {
$GodotPath = Join-Path $env:TEMP "godot-4.6.1-openwc\Godot_v4.6.1-stable_win64.exe"
}
}
if (-not (Test-Path -LiteralPath $GodotPath)) {
throw "Godot executable not found: $GodotPath"
}
if (-not (Test-Path -LiteralPath $ReferenceDirectory)) {
throw "Reference directory not found: $ReferenceDirectory"
}
$invariantCulture = [Globalization.CultureInfo]::InvariantCulture
$cameraFovText = $CameraFov.ToString($invariantCulture)
$waitSecondsText = $WaitSeconds.ToString($invariantCulture)
$measureSecondsText = $MeasureSeconds.ToString($invariantCulture)
$absoluteOutput = if ($Output.StartsWith("user://")) {
Join-Path $env:APPDATA "Godot\app_userdata\OpenWC\$($Output.Substring(7))"
} else {
[IO.Path]::GetFullPath((Join-Path $repoRoot $Output))
}
$ranking = [Collections.Generic.List[object]]::new()
Push-Location $repoRoot
try {
foreach ($yawOffset in $YawOffsets) {
foreach ($pitchOffset in $PitchOffsets) {
$yawText = $yawOffset.ToString("0.###", $invariantCulture)
$pitchText = $pitchOffset.ToString("0.###", $invariantCulture)
$candidateName = "yaw_$($yawText.Replace('-', 'n').Replace('.', '_'))__pitch_$($pitchText.Replace('-', 'n').Replace('.', '_'))"
$candidateOutput = "$Output/$candidateName"
$comparisonReport = Join-Path $absoluteOutput "$candidateName.json"
if ($PlanOnly) {
$ranking.Add([pscustomobject]@{
yaw_offset_degrees = $yawOffset
pitch_offset_degrees = $pitchOffset
camera_fov_degrees = $CameraFov
candidate_directory = $candidateOutput
})
continue
}
& $GodotPath --path . --script res://src/tools/capture_render_checkpoints.gd -- `
--output $candidateOutput --only $Checkpoint --single-pass `
--camera-fov $cameraFovText `
--camera-yaw-offset $yawText --camera-pitch-offset $pitchText `
--wait $waitSecondsText --measure $measureSecondsText
if ($LASTEXITCODE -ne 0) {
throw "Capture failed for yaw=$yawText pitch=$pitchText"
}
& $GodotPath --headless --path . --script res://src/tools/compare_render_checkpoints.gd -- `
--reference $ReferenceDirectory --candidate $candidateOutput `
--only $Checkpoint --output $comparisonReport
if ($LASTEXITCODE -notin @(0, 1)) {
throw "Comparison failed for yaw=$yawText pitch=$pitchText"
}
$comparison = Get-Content -Raw -LiteralPath $comparisonReport | ConvertFrom-Json
if ($comparison.compared_count -lt 1) {
throw "No paired image was compared for yaw=$yawText pitch=$pitchText"
}
$meanError = ($comparison.results | Measure-Object -Property mean_perceptual_error -Average).Average
$changedRatio = ($comparison.results | Measure-Object -Property changed_pixel_ratio -Average).Average
$ranking.Add([pscustomobject]@{
yaw_offset_degrees = $yawOffset
pitch_offset_degrees = $pitchOffset
camera_fov_degrees = $CameraFov
mean_perceptual_error = $meanError
changed_pixel_ratio = $changedRatio
candidate_directory = $candidateOutput
comparison_report = $comparisonReport
})
}
}
$rankedCandidates = if ($PlanOnly) {
@($ranking)
} else {
@($ranking | Sort-Object mean_perceptual_error, changed_pixel_ratio)
}
$rankingPath = Join-Path $absoluteOutput "ranking.json"
$rankingDocument = [ordered]@{
schema_version = 1
checkpoint = $Checkpoint
reference_directory = [IO.Path]::GetFullPath($ReferenceDirectory)
candidates = $rankedCandidates
}
New-Item -ItemType Directory -Force -Path $absoluteOutput | Out-Null
$rankingDocument | ConvertTo-Json -Depth 6 | Set-Content -Encoding UTF8 -LiteralPath $rankingPath
$rankedCandidates | Format-Table yaw_offset_degrees, pitch_offset_degrees, mean_perceptual_error, changed_pixel_ratio
$completionKind = if ($PlanOnly) { "plan" } else { "ranking" }
Write-Output "Camera pose sweep $completionKind completed: $rankingPath"
} finally {
Pop-Location
}