melbar/aitrailer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Retiming long scene matches by action phase

Browse Source
This commit is contained in:
Melbar
2026-05-02 20:47:59 +02:00
parent 252f710396
commit 8415516f89
3 changed files with 158 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cli.py
+80 -11
View File
@@ -640,7 +640,7 @@ def _filter_semantically_invalid_vision_matches(results: list, beats: list, cfg)
from dataclasses import replace
from src.llm.vision_cache import find_action_window_in_scene, validate_match_window_with_vision
from src.cv.scene_indexer import build_scene_index
from src.cv.global_scan import align_in_point_by_content, align_in_point_by_motion
from src.cv.global_scan import align_in_point_by_content_and_motion
logger = logging.getLogger(__name__)
beats_by_id = {beat.beat_id: beat for beat in beats}
@@ -654,19 +654,13 @@ def _filter_semantically_invalid_vision_matches(results: list, beats: list, cfg)
if found is None:
return None
start_s, end_s, semantic_score, reason = found
window_s = max(1.0, min(4.0, (end_s - start_s) * 1.5))
motion_in_s, motion_score = align_in_point_by_motion(
window_s = max(3.0, min(8.0, (end_s - start_s) * 4.0))
aligned_in_s, combined_score, content_score, motion_score = align_in_point_by_content_and_motion(
check_beat,
start_s,
cfg,
search_window_s=window_s,
)
aligned_in_s, content_score = align_in_point_by_content(
check_beat,
motion_in_s,
cfg,
search_window_s=min(window_s, 0.8),
)
aligned_in_s = max(scene.start_s, min(aligned_in_s, max(scene.start_s, scene.end_s - check_beat.duration_s)))
ok, verify_reason = validate_match_window_with_vision(
check_beat,
@@ -685,7 +679,7 @@ def _filter_semantically_invalid_vision_matches(results: list, beats: list, cfg)
verify_reason,
)
return None
score = max(content_score, min(0.99, semantic_score * 0.75 + motion_score * 0.25))
score = max(combined_score, min(0.99, semantic_score * 0.70 + motion_score * 0.20 + content_score * 0.10))
return scene, aligned_in_s, score, f"{reason}; {verify_reason}"
kept = []
@@ -728,7 +722,82 @@ def _filter_semantically_invalid_vision_matches(results: list, beats: list, cfg)
valid = False
break
if valid:
kept.append(result)
repaired = False
if getattr(result, "segments", ()):
new_segments = []
repair_reasons = []
changed = False
for segment in result.segments:
scene = scenes_by_id.get(segment.scene_id)
if scene is None or scene.duration_s <= max(segment.duration_s * 1.6, 6.0):
new_segments.append(segment)
continue
segment_beat = replace(
beat,
start_s=beat.start_s + segment.trailer_offset_s,
end_s=beat.start_s + segment.trailer_offset_s + segment.duration_s,
)
repair = realign_window(segment_beat, segment.scene_id)
if repair is None:
new_segments.append(segment)
continue
repair_scene, aligned_in_s, score, repair_reason = repair
if abs(aligned_in_s - segment.in_point_s) <= 1.0 / cfg.export.edl_frame_rate:
new_segments.append(segment)
continue
changed = True
repair_reasons.append(repair_reason)
new_segments.append(replace(
segment,
scene_id=repair_scene.scene_id,
in_point_s=aligned_in_s,
out_point_s=aligned_in_s + segment.duration_s,
match_score=score,
is_confirmed=score >= cfg.cv.deep_scan.match_threshold,
))
if changed and new_segments:
first = new_segments[0]
repaired_score = min(seg.match_score for seg in new_segments)
logger.info(
"Beat %d: realigned semantically valid long scene by motion/action windows (%s)",
result.beat_id,
"; ".join(repair_reasons),
)
kept.append(replace(
result,
scene_id=first.scene_id,
in_point_s=first.in_point_s,
out_point_s=first.out_point_s,
in_point_frame=int(first.in_point_s * cfg.export.edl_frame_rate),
match_score=repaired_score,
is_confirmed=repaired_score >= cfg.cv.deep_scan.match_threshold,
segments=tuple(new_segments),
))
repaired = True
else:
scene = scenes_by_id.get(result.scene_id)
if scene is not None and scene.duration_s > max(result.duration_s * 1.6, 6.0):
repair = realign_window(beat, result.scene_id)
if repair is not None:
repair_scene, aligned_in_s, score, repair_reason = repair
if abs(aligned_in_s - result.in_point_s) > 1.0 / cfg.export.edl_frame_rate:
logger.info(
"Beat %d: realigned semantically valid long scene by motion/action window (%s)",
result.beat_id,
repair_reason,
)
kept.append(replace(
result,
scene_id=repair_scene.scene_id,
in_point_s=aligned_in_s,
out_point_s=aligned_in_s + result.duration_s,
in_point_frame=int(aligned_in_s * cfg.export.edl_frame_rate),
match_score=score,
is_confirmed=score >= cfg.cv.deep_scan.match_threshold,
))
repaired = True
if not repaired:
kept.append(result)
else:
if getattr(result, "segments", ()):
new_segments = []