# Review — B-final 0 fatal · 1 major · 8 minor · 1 cosmetic · 16 measured but NOT judged engines: {"measure": {"cuts": 9, "shots": 6, "findings": 17}, "say": {"scored": 11, "worst": [["instantly", 0.33], ["defrizzes", 0.5], ["chlorine", 0.55], ["tangling", 0.55], ["roughing", 0.67]]}, "caption": {"events": 21, "ocr_calls": 43, "coverage": 0.208, "errors": 0, "sync": {"median_lag_s": -0.1, "events": 21, "p90_abs": 0.708}}, "overlay": {"boxes": 16, "undeclared": 1, "expected": 4}, "fit": {"devices": 2, "pacing": {"devices": 2, "per_minute": 3.3, "coverage": 0.177}, "major": 0}, "pair": {"body_overlays": 1, "not_hers": 0}, "click": {"clicks": 0, "discontinuities": 0, "seams_hit": 0, "elsewhere": 0}} | At | Kind | Severity | What | |---|---|---|---| | 16.2s | editing | major | the picture changes inside the spoken word "hair." (16.10-16.44 s). The track runs on through it, so the line is whole and the EDIT lands mid-word, which reads as clumsy rather than broken. | | 0.0s | overlay | minor | STICKER a86c7794_B-hook-keyed.png: on screen 4.0 s; the swipes hold an overlay at most 3.0 s (median 1.2 s) | | 0s | overlay | minor | PACING: 2 devices, 3.3 per minute, something added for 18% of the runtime; the swipes run at most 7 per minute and 15% covered | | 8.68s | editing | minor | the cut lands 60 ms after the word "you." ends, so the line has no tail before the shot changes | | 10.32s | editing | minor | the cut lands 0 ms after the word "and" ends, so the line has no tail before the shot changes | | 14.52s | editing | minor | the cut lands 0 ms after the word "causing" ends, so the line has no tail before the shot changes | | 16.2s | editing | minor | the cut lands 100 ms after the word "knotted" ends, so the line has no tail before the shot changes | | 18.72s | editing | minor | the cut lands 20 ms after the word "whole" ends, so the line has no tail before the shot changes | | 32.92s | editing | minor | the cut lands 40 ms after the word "instantly." ends, so the line has no tail before the shot changes | | 27.76s | editing | cosmetic | the cut sits in a silent pause that the transcript spans as the word "You" (27.56-28.06 s): a cut in a gap, not in a word | | 0.0s | overlay | unknown | PERSISTENT GRAPHIC not on the edit track: 182x387 px at (281,68), 3 shot(s) from 0.0s to 27.76s, edge persistence 0.996, pixel std inside 6.82 vs frame motion 5.6, reads "strand,"; crop /root/r-552da4e245/out/_overlay/overlay-281x68-0.3s.png | | 30.72s | skin-tone | unknown | BODY OVERLAY skin a*b* (138.7, 142.9) vs the presenter's cheek (138.6, 142.1): distance 0.8 (a*+0.1, b*+0.8), reported only until an accepted pair is measured | | 0s | cadence | unknown | CADENCE, for the eye only: the first 10 s carry 1 visible change(s), one every 5.0 s; the general lean in short-form is one every 1-2 s. The whole ad runs one every 3.7 s. A yapper or an older audience wants longer holds, so this is case by case and never a defect. | | 0.0s | skin-tone | unknown | skin NOT JUDGED: the shot from 0.0s to 8.68s sits a*+4.3 b*+5.2 from the other shots (distance 7.1). The donor reaches 7.1 at its own cuts, so this scale cannot separate a light change from a skin change. A human decides; positive a* and b* together read as warmer and more yellow. | | 0s | lipsync | unknown | sync NOT judged: peak MFCC R2 0.127 is below the 0.15 floor every control cleared, so the mouth is not readable here; best lag 0 ms; loudness correlation 0.077 at zero lag, best 0.140 at 200 ms, 0 frames loud-with-closed-mouth, 82 open-over-silence, NOT judged (the donor reads 0.087 / 26 / 1); LipLeak 0.636 | | 0s | caption | unknown | CAPTION SYNC, reported only: median -0.10 s from the first spoken word to the caption appearing over 21 events (90th percentile 0.71 s); the correct cut reads +0.00 / 0.19 and unshifted controls read -0.24 to -0.29, so no verdict is drawn | | 8.68s | audio-seam | unknown | the track steps -20.6 dB here, and one side of the cut is silence, so this is a speaker starting or stopping rather than a seam in the sound | | 10.32s | pronunciation | unknown | REDUB SHORTLIST: "chlorine" is heard by an acoustic model with no vocabulary as "cor" (overlap 0.55). Whisper reports the right word because its language model supplies it. The donor control slurs its own words to 0.38, so this is a ranking for a human, not a verdict. | | 14.52s | continuity | unknown | the shot from 14.52s to 16.2s is NOT JUDGED: a face was found in only 4 of the frames sampled from it, which is too little to profile. Its numbers are top_hist 0.394, SFace 0.147, skin distance 11.2. | | 16.76s | audio-seam | unknown | the track steps +24.1 dB here, and one side of the cut is silence, so this is a speaker starting or stopping rather than a seam in the sound | | 19.44s | audio-seam | unknown | the track steps +33.5 dB here, and one side of the cut is silence, so this is a speaker starting or stopping rather than a seam in the sound | | 26.12s | pronunciation | unknown | REDUB SHORTLIST: "defrizzes" is heard by an acoustic model with no vocabulary as "dyfurs as" (overlap 0.5). Whisper reports the right word because its language model supplies it. The donor control slurs its own words to 0.38, so this is a ranking for a human, not a verdict. | | 27.76s | skin-tone | unknown | skin NOT JUDGED: the shot from 27.76s to 32.92s sits a*-2.2 b*-3.0 from the other shots (distance 3.7). The donor reaches 7.1 at its own cuts, so this scale cannot separate a light change from a skin change. A human decides; positive a* and b* together read as warmer and more yellow. | | 32.48s | pronunciation | unknown | REDUB SHORTLIST: "instantly" is heard by an acoustic model with no vocabulary as "ste" (overlap 0.33). Whisper reports the right word because its language model supplies it. The donor control slurs its own words to 0.38, so this is a ranking for a human, not a verdict. | | 32.92s | skin-tone | unknown | skin NOT JUDGED: the shot from 32.92s to 36.72s sits a*-2.3 b*-3.9 from the other shots (distance 4.6). The donor reaches 7.1 at its own cuts, so this scale cannot separate a light change from a skin change. A human decides; positive a* and b* together read as warmer and more yellow. | | 34.18s | pronunciation | unknown | REDUB SHORTLIST: "tangling" is heard by an acoustic model with no vocabulary as "ang" (overlap 0.55). Whisper reports the right word because its language model supplies it. The donor control slurs its own words to 0.38, so this is a ranking for a human, not a verdict. |