Constrain hi-res phase refine and update beat 14

docs/ALGORITHM.md

@@ -166,6 +166,11 @@ eine kurze Geste erst korrekt erkannt und anschließend in eine spätere
 derselben Source-Szene ähnlich gut scoren und die Beat-Dauer abdecken,
 bevorzugt der Matcher die frühere Phase.
 
+Der zusätzliche Hi-Res-Phasenrefine bleibt lokal um den bereits validierten
+Inpoint und übernimmt nur klare Verbesserungen. Er darf keine ganze lange
+Dialogszene nach ähnlichen Layouts durchsuchen, weil sonst dieselbe Location
+mit anderer Gestik als falsche Phase gewinnen kann und die Laufzeit explodiert.
+
 ## Multi-Shot-Beats
 
 Enthält ein Trailerbeat selbst einen harten Umschnitt, werden Kandidaten an

src/cv/global_scan.py

@@ -316,14 +316,31 @@ def _hires_phase_refine(
     scan_templates = [(off, feat, sp) for off, feat, sp, _ in ref_templates]
     max_ref_offset = max(off for off, _, _ in scan_templates)
 
-    # Scan the full scene
+    # Scan only a local window around the already validated in-point.  A full
+    # scene scan can jump to a different phase of the same shot (same room,
+    # same actor, different gesture) and it is unnecessarily expensive.
+    local_window_s = max(0.8, min(2.0, cfg.cv.deep_scan.content_align_window_seconds * 3.0))
+    scan_start_s = max(scene_start_s, in_point_s - local_window_s)
+    scan_end_s = min(scene_end_s, in_point_s + local_window_s)
+
+    original_scores: list[float] = []
+    with open_video(cfg.paths.source_movie) as cap:
+        for off, ref_feat, ref_spatial in scan_templates:
+            src_frame = grab_frame_at(cap, in_point_s + off)
+            if src_frame is not None:
+                original_scores.append(_hires_phase_score(ref_feat, ref_spatial, src_frame))
+    original_score = -1.0
+    if original_scores:
+        original_score = (sum(original_scores) / len(original_scores)) * 0.7 + min(original_scores) * 0.3
+
+    # Scan the local neighbourhood.
     best_t = in_point_s
-    best_score = -1.0
+    best_score = original_score
     scan_step_s = max(1.0 / (cfg.export.edl_frame_rate or 24.0), 0.04)
 
     with open_video(cfg.paths.source_movie) as cap:
-        t = scene_start_s
+        t = scan_start_s
-        while t + max_ref_offset <= scene_end_s:
+        while t + max_ref_offset <= scan_end_s:
             scores: list[float] = []
             all_ok = True
             for off, ref_feat, ref_spatial in scan_templates:
@@ -340,6 +357,9 @@ def _hires_phase_refine(
                     best_t = t
             t = round(t + scan_step_s, 6)
 
+    if best_score < original_score + 0.025:
+        return in_point_s
+
     if best_t != in_point_s:
         logger.info(
             'Beat %d: hi-res phase refine moved in-point %.3fs -> %.3fs '