diff --git a/README.md b/README.md
index 191cf99..49b29d9 100644
--- a/README.md
+++ b/README.md
@@ -20,11 +20,12 @@ The agent can also be sent on an autonomous search: an orange triangle is
 hidden on the back side of one of the crates, and the model must explore the
 room with real vision, spot it, and report it. This recording shows the run
 from three angles — the top-down map, the actual first-person frames the
-model saw, and its chat/reasoning:
+model saw, and its chat/reasoning (recorded with smooth motion, ~2 Hz):
 
 ![search mission](search_mission.gif)
 
-Regenerate it with `python -m testbed.record_search --provider polza --out search_mission.gif`.
+Regenerate it with `python -m testbed.record_search --provider polza --goal search
+--out search_mission.gif --capture-hz 2.0 --cruise 0.8 --steps 60`.
 
 ## Smooth & proactive movement
 
diff --git a/search_mission.gif b/search_mission.gif
index fe8a3b0..f10432d 100644
Binary files a/search_mission.gif and b/search_mission.gif differ
diff --git a/smooth_mission.gif b/smooth_mission.gif
index f0290bd..8557302 100644
Binary files a/smooth_mission.gif and b/smooth_mission.gif differ
diff --git a/testbed/record_search.py b/testbed/record_search.py
index b356663..7ae88d8 100644
--- a/testbed/record_search.py
+++ b/testbed/record_search.py
@@ -236,8 +236,8 @@ async def run(args: argparse.Namespace) -> int:
                         samples.append((bridge.world.capsule.x, bridge.world.capsule.z))
                         top = _topdown_frame(bridge, ctl.path, len(frames), marker_seen)
                         frames.append(_compose(top, view, chat.render()))
-                        if len(frames) > 500:  # keep the GIF bounded
-                            frames[:100] = []
+                        if len(frames) > 350:  # keep the GIF bounded
+                            frames[:150] = []
                         await asyncio.sleep(1.0 / args.capture_hz)
 
             agent_done = asyncio.Event()