From 7d23a71947dacf6f81201353c79d5a9d9f600335 Mon Sep 17 00:00:00 2001
From: hbrain <hbrain@piagent.local>
Date: Sun, 31 May 2026 17:25:42 +0000
Subject: [PATCH] Adjust long video caption timing to start after 2s and show
 for 6s

---
 movmaker.py | 60 +++++++++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 51 insertions(+), 9 deletions(-)

diff --git a/movmaker.py b/movmaker.py
index f0550fa..9210f1a 100755
--- a/movmaker.py
+++ b/movmaker.py
@@ -60,6 +60,9 @@ CLIP_AUDIO_SIDECHAIN_THRESHOLD = 0.04
 CLIP_AUDIO_SIDECHAIN_RATIO = 8.0
 CLIP_AUDIO_SIDECHAIN_ATTACK = 5.0
 CLIP_AUDIO_SIDECHAIN_RELEASE = 250.0
+CAPTION_LONG_VIDEO_MIN_DURATION = 10.0
+CAPTION_LONG_VIDEO_DELAY = 2.0
+CAPTION_LONG_VIDEO_SHOW = 6.0
 
 
 ORDER_FILENAME = "order.json"
@@ -988,6 +991,7 @@ def build_video(
     video_intro_duration: float | None = None,
     intro_visible_lead: float = 0.0,
     skip_video_fade_out: bool = False,
+    clip_kinds: list[str] | None = None,
 ) -> None:
     """Assemble normalized clips, overlays, and audio into the final MP4."""
     output.parent.mkdir(parents=True, exist_ok=True)
@@ -1132,17 +1136,26 @@ def build_video(
         if not caption:
             continue
         if n == 1:
-            start = intro_duration
-            end = total_duration
+            base_start = intro_duration
+            base_end = total_duration
         elif idx == 0:
-            start = intro_duration
-            end = intro_duration + transition_offsets[0]
+            base_start = intro_duration
+            base_end = intro_duration + transition_offsets[0]
         elif idx == n - 1:
-            start = intro_duration + transition_offsets[idx - 1] + fade
-            end = total_duration
+            base_start = intro_duration + transition_offsets[idx - 1] + fade
+            base_end = total_duration
         else:
-            start = intro_duration + transition_offsets[idx - 1] + fade
-            end = intro_duration + transition_offsets[idx]
+            base_start = intro_duration + transition_offsets[idx - 1] + fade
+            base_end = intro_duration + transition_offsets[idx]
+
+        start = base_start
+        end = base_end
+        clip_kind = clip_kinds[idx] if clip_kinds and idx < len(clip_kinds) else ""
+        clip_duration = durations[idx] if idx < len(durations) else 0.0
+        if clip_kind == "video" and clip_duration > CAPTION_LONG_VIDEO_MIN_DURATION:
+            start = base_start + CAPTION_LONG_VIDEO_DELAY
+            end = min(base_end, start + CAPTION_LONG_VIDEO_SHOW)
+
         if end <= start:
             continue
         caption_fade = min(1.0, max(0.0, (end - start) / 2))
@@ -1411,6 +1424,7 @@ def main(argv: list[str] | None = None) -> int:
         # Use probed normalized duration for accuracy.
         durations = [ffprobe_duration(out) for out in clips]
         captions = [data.captions.get(item.path.name, data.captions.get(str(item.path), "")) for item in media]
+        clip_kinds = [item.kind for item in media]
         video_audio = data.video_audio or set()
         clip_audio_paths = []
         for item in media:
@@ -1451,6 +1465,7 @@ def main(argv: list[str] | None = None) -> int:
                 intro_clip_path = intro_clip
                 intro_clip_duration = intro_item.duration
                 captions = ["", *captions]
+                clip_kinds = ["image", *clip_kinds]
                 clip_audio_paths = [None, *clip_audio_paths]
                 clip_stamps = ["", *clip_stamps]
                 intro_visible_lead = min(intro_item.duration, INTRO_LOGO_FADE_OFFSET)
@@ -1460,10 +1475,37 @@ def main(argv: list[str] | None = None) -> int:
             clips.append(intro_clip_path)
             durations.append(intro_clip_duration)
             captions.append("")
+            clip_kinds.append("image")
             clip_audio_paths.append(None)
             clip_stamps.append("")
             outro_added = True
-        build_video(clips, durations, audio, clip_audio_paths, output, data, first_media_date, captions, clip_stamps, fade, audio_fade, audio_fade_in, video_fade, args.intro_logo, intro_logo_alpha, width, height, fps, preset, crf, tmp, video_intro_duration, intro_visible_lead=intro_visible_lead, skip_video_fade_out=outro_added)
+        build_video(
+            clips,
+            durations,
+            audio,
+            clip_audio_paths,
+            output,
+            data,
+            first_media_date,
+            captions,
+            clip_stamps,
+            fade,
+            audio_fade,
+            audio_fade_in,
+            video_fade,
+            args.intro_logo,
+            intro_logo_alpha,
+            width,
+            height,
+            fps,
+            preset,
+            crf,
+            tmp,
+            video_intro_duration,
+            intro_visible_lead=intro_visible_lead,
+            skip_video_fade_out=outro_added,
+            clip_kinds=clip_kinds,
+        )
 
     console(f"Done: {output}")
     return 0