Filmstrip & Swing Position Detection — Getting the Best Results

MotionEdge's filmstrip automatically detects 10 key swing positions — Address, Takeaway, Backswing, Top of Backswing, Transition, Downswing, Pre-Impact, Impact, Follow-Through, and Finish. This works by tracking your hands and body through the video. Here's how to get the most accurate results.

The Two Things That Matter Most

1. Use a Stable Camera (Tripod)

This is the single biggest factor in filmstrip accuracy.

Why it matters: MotionEdge tracks your wrist position frame-by-frame to find the top of your backswing (hands highest), impact (hands return to ball), and finish (hands settle). When the camera shakes, every joint in your body appears to bounce up and down in the video — even if you're standing perfectly still. The algorithm can't tell the difference between "your hands went up because of your backswing" and "your hands went up because the camera dipped down."

What goes wrong with a shaky camera:

• Top of backswing gets placed in the follow-through instead of the actual backswing — because the camera drift makes the follow-through hands appear higher
• Impact gets placed several seconds after the real impact — because the camera motion corrupts the hand-return-to-address signal
• The filmstrip shows 10 frames that don't match the actual swing positions

The fix: Use a tripod, a phone mount on your golf bag, or lean your phone against something stable. Even a cheap $15 phone tripod dramatically improves results. If you're having someone hold the phone, ask them to brace their elbows against their body and stay as still as possible.

2. Trim the Video Close to the Swing

Why it matters: MotionEdge analyzes the entire video to find your swing. If you record 30 seconds of standing around, then a 2-second swing, then 30 seconds of watching the ball and walking to pick up the tee — the algorithm has to sort through all that extra footage. Post-swing activities like watching the ball flight, walking, or bending down can look similar to swing positions when viewed through pose tracking data alone.

What goes wrong with long, untrimmed video:

• Address may be detected during a post-swing standing moment (it looks "still" — just like address)
• Impact may be detected when you bend to pick up your tee (hands return to ground level — same signature as impact)
• The algorithm wastes its analysis on footage that isn't part of the swing

The fix: Before analyzing, trim the video to start a second or two before address and end a second or two after the finish. The built-in trim tool in Photos works perfectly. You don't need to be frame-precise — just cut out the long lead-in and the ball-watching aftermath.

Quick Checklist

"My Filmstrip Looks Wrong — What Happened?"

If the 10 positions don't match what you expect, check these in order:

1. Was the camera shaky? Look at the background — if the treeline or buildings bounce up and down, camera shake is likely the cause. Re-record on a tripod.

2. Is there a lot of extra footage? If the video is more than ~8 seconds for a single swing, try trimming it tighter.

3. Is it a DTL (down-the-line) video? DTL detection uses additional hand-path tracking that's more sensitive to camera stability than face-on.

4. Is it 30fps? Higher frame rates (120fps, 240fps) give the algorithm more data to work with. At 30fps, the entire downswing may only be 4-5 frames, giving less margin for detection. If possible, record at 120fps+ in slo-mo mode.

5. Is it a slow-motion video with wrong frame rate metadata? Some slow-motion videos downloaded from the internet or exported from video editors have their slow-motion "baked in" — the video plays at 30fps but the actual content was captured at 120fps+. MotionEdge's auto-detection relies on the video's frame rate metadata to calibrate its timing. When a swing that should take 1.5 seconds instead spans 6 seconds of video, the algorithm's timing windows are all wrong and impact will be placed in the downswing instead of at the ball. If you have a baked-in slow-motion video, you'll need to set the filmstrip positions manually. Videos recorded in MotionEdge or with the iPhone's native slo-mo mode preserve the correct frame rate metadata and work automatically.

How the Detection Works (Behind the Scenes)

For the curious: MotionEdge tracks your wrist position through every frame of the video and builds a "wrist curve." During a golf swing, this curve has a distinctive shape:

• Address: Hands are still at hip level
• Backswing: Hands rise (wrist Y decreases in video coordinates)
• Top: Hands reach their highest point
• Downswing: Hands drop rapidly
• Impact: Hands return to approximately their address position (near the ball)
• Follow-through: Hands rise again past the body

The algorithm finds these landmarks by matching this expected pattern against the actual wrist curve. Camera shake adds noise to the curve. Extra footage adds false patterns. A clean, trimmed, stable video gives the cleanest curve — and the most accurate filmstrip.

See also: Camera Placement Best Practices for detailed guidance on camera height, angle, and distance.