{ "encoders": { "official_nvfp4": { "exact_requested_timing_pass": false, "median_frame_mae": 3.06270667989418, "requested_3s_window_peak": { "hard_cut_detected": false, "score": 3.2759641617063493, "seconds": 3.2083333333333335 }, "requested_7s_window_peak": { "hard_cut_detected": false, "score": 4.487263351521164, "seconds": 6.5 }, "separated_global_peaks": [ { "score": 56.54574859457672, "seconds": 6.375 }, { "score": 53.21586681547619, "seconds": 8.583333333333334 }, { "score": 6.335105613425926, "seconds": 4.875 }, { "score": 4.9722118882275135, "seconds": 5.625 }, { "score": 4.4160724619708995, "seconds": 7.458333333333333 } ] }, "pruned24_balanced_ara": { "exact_requested_timing_pass": false, "median_frame_mae": 2.393596023478836, "requested_3s_window_peak": { "hard_cut_detected": false, "score": 2.6754712301587302, "seconds": 3.5 }, "requested_7s_window_peak": { "hard_cut_detected": true, "score": 71.38343253968254, "seconds": 7.25 }, "separated_global_peaks": [ { "score": 71.38343253968254, "seconds": 7.25 }, { "score": 63.06867972883598, "seconds": 4.458333333333333 }, { "score": 8.930333374669312, "seconds": 5.708333333333333 }, { "score": 5.625527033730159, "seconds": 8.541666666666666 }, { "score": 2.892893311838624, "seconds": 1.25 } ] }, "recovered_8b": { "exact_requested_timing_pass": false, "median_frame_mae": 3.9301163607804233, "requested_3s_window_peak": { "hard_cut_detected": false, "score": 4.111085276124339, "seconds": 3.0416666666666665 }, "requested_7s_window_peak": { "hard_cut_detected": false, "score": 10.220491278108465, "seconds": 6.583333333333333 }, "separated_global_peaks": [ { "score": 55.339544064153436, "seconds": 8.416666666666666 }, { "score": 10.676406456679894, "seconds": 5.708333333333333 }, { "score": 10.220491278108465, "seconds": 6.583333333333333 }, { "score": 8.637467964616402, "seconds": 4.875 }, { "score": 5.989288814484127, "seconds": 7.458333333333333 } ] } }, "format": "minimax_h3_neon_detective_three_te_cut_diagnostics_1344x768_v1", "metric": "Mean absolute RGB frame difference after decoding each 240-frame presentation video at 336x192. Separated peaks are at least 0.75 seconds apart. This identifies abrupt transitions, not scene semantics.", "requested_boundaries_seconds": [ 3.0, 7.0 ], "structural_interpretation": "Exact timing is separate from semantic action compliance; human review remains required." }