diff --git a/gnommo/scaffold.py b/gnommo/scaffold.py
index f75a3a0..5e346cc 100644
--- a/gnommo/scaffold.py
+++ b/gnommo/scaffold.py
@@ -181,7 +181,7 @@ def derive_events(
                     vs,
                     t.overrides,
                     default_end_on=(
-                        None if t.marker_id.startswith("narration:") else "next_video"
+                        None if t.marker_id.startswith("narration:") else "next_slide"
                     ),
                 )
                 e["handle"] = pres["handle"]
diff --git a/gnommo/transformer.py b/gnommo/transformer.py
index e20f093..c6af800 100644
--- a/gnommo/transformer.py
+++ b/gnommo/transformer.py
@@ -63,7 +63,7 @@ def resolve_video_presentation(
     marker_id: str,
     video_source,
     overrides: Optional[dict] = None,
-    default_end_on: Optional[str] = "next_video",
+    default_end_on: Optional[str] = "next_slide",
 ) -> dict:
     """Resolve a video marker's per-occurrence presentation to an atomic dict.
 
@@ -74,8 +74,9 @@ def resolve_video_presentation(
     videos.json collision.
 
     Precedence per field: explicit event override > shorthand prefix > videos.json
-    default > built-in default. `default_end_on` is "next_video" for video triggers and
-    None for [narration:] (which runs to the end).
+    default > built-in default. `default_end_on` is "next_slide" for video triggers
+    (so an untagged clip can't overstay and obscure later content) and None for
+    [narration:] (which runs to the end).
 
     Returns {handle, cutout, layer, end_on, take, pause_narration}.
     """
@@ -1009,9 +1010,20 @@ def build_render_plan(
                 if vid_event is event:
                     # Don't shift the pause event by its own pause
                     continue
+                # A clip whose window STRADDLES the freeze (starts before, ends after)
+                # would otherwise be stretched across the whole cutscene. Its overlay
+                # source isn't freeze-spliced like the narration is, so it'd keep
+                # advancing under the cutscene and surface the wrong frame in the sliver
+                # between the cutscene ending and its own (shifted) end — the video7
+                # "appears at the start, vanishes when it should show" artifact. Instead
+                # end it right at the freeze onset: it plays its pre-pause content, then
+                # the cutscene cleanly takes over.
+                straddles = vid_event.start_time < narration_time < vid_event.end_time
                 if vid_event.start_time >= narration_time:
                     vid_event.start_time += pause_duration
-                if vid_event.end_time > narration_time:
+                if straddles:
+                    vid_event.end_time = narration_time
+                elif vid_event.end_time > narration_time:
                     vid_event.end_time += pause_duration
 
             for aud_event in audio_events:
@@ -1407,7 +1419,7 @@ def _extract_video_events(
             marker_id,
             video_source,
             overrides,
-            default_end_on=(None if trigger_type == "narration" else "next_video"),
+            default_end_on=(None if trigger_type == "narration" else "next_slide"),
         )
         cutout_name = pres["cutout"]
         cutout = cutouts[cutout_name]