Trace a real flight path over a Google Maps satellite screenshot and hand it to your AI video model as a navigation reference — turning a vague "fly over the city" prompt into an exact, repeatable route from ground level to a landmark, with a real cameo dropped in along the way.
A single continuous FPV drone shot that follows an exact, pre-planned route — no vague "fly over the city" guesswork. You screenshot Google Maps in satellite/3D view, trace your flight path directly on top of it in a photo editor, then feed that annotated image to Seedance 2.0 as a navigation reference alongside your written prompt. The model follows the drawn line instead of inventing its own path, so beach launches, favela weaves, and a landmark orbit all land exactly where you planned them.
Why a hand-drawn map beats a text-only prompt
Turn a satellite screenshot into a navigation reference image
Open Google Maps, switch to Satellite view, and tilt into 3D if the area supports it. Zoom until your start point and end landmark are both visible in one frame — or take two overlapping screenshots if the route is long.
Screenshot the map, then open it in any photo editor or markup tool — Photos markup on iPhone, Procreate, or even Google Maps' own Measure/Add a Missing Road line tool works in a pinch.
Draw your route as a single continuous line, thick enough to read clearly, from ground level at the start point, winding through any streets or terrain you want the drone to weave through, and ending with a small loop or spiral around your landmark to signal an orbit.
Label the landmark clearly if it isn't obvious from the map already, and keep the line color high-contrast (orange or red works best against satellite imagery).
Pro tip: Loops and spirals in the drawn line get read by the model as orbit or wraparound camera moves — use them deliberately at your key landmark to cue a slow reveal instead of a straight flyover.
Set these before pasting the prompt
| Setting | Value | Why |
|---|---|---|
| Model | Seedance 2.0 | Image-to-video with reference support — reads the traced map as a navigation cue |
| Reference Image | Your annotated map screenshot | Anchors the route, order of landmarks, and camera direction before generation starts |
| Duration | 8–15 seconds | Long enough to cover ground-to-landmark without feeling rushed; longer routes may need to be split into linked clips |
| Aspect Ratio | 9:16 vertical | Matches real FPV drone footage shared on TikTok/Reels/Shorts; use 16:9 for YouTube |
| Style | Photorealistic, golden-hour | Keeps the generated footage looking like real drone capture rather than stylized animation |
Pro tip: If the model struggles to hold a long or winding route in one generation, split it into two linked clips at a natural midpoint (like ground-level city flying → landmark climb) and stitch them together with a hard cut disguised as a bank or dive.
Upload your traced map as the reference image, then paste this into Seedance 2.0
Upload your annotated Google Maps screenshot as the reference/navigation image in Seedance 2.0.
Copy the prompt below exactly as written and paste it into the prompt field — this example is a real Rio de Janeiro route from Copacabana Beach to Christ the Redeemer.
Swap the city, landmark, and opening scene for your own route — the structure (reference the traced map, describe ground level, describe the climb, describe the landmark reveal, describe the exit) is what carries over to any location.
Tap Generate and wait for the render.
Using the attached Google Earth screenshot as the navigation reference, generate a single continuous unbroken FPV drone shot that follows the exact red flight path. No cuts, no edits, no transitions, no camera resets. The entire sequence must feel like one uninterrupted real-world FPV drone flight. The flight begins low on Copacabana Beach, Rio de Janeiro, Brazil, just a few metres above the sand. In the very first second, people walk toward the camera along the shoreline before the drone banks upward and climbs above them, revealing the wider beach scene below: people playing volleyball and football, sunbathers, surfers carrying boards, joggers along the promenade, and cars driving along the beachfront road. Continue skimming past crashing waves, beach umbrellas, and swaying palm trees before accelerating inland. Climb aggressively through Rio's colourful hillside neighbourhoods, weaving smoothly between the tightly packed favela buildings, power lines, rooftop terraces, narrow streets, and staircases. Every movement should feel physically accurate with natural FPV banking, dives, climbs, and momentum. As the drone gains altitude, reveal the city alive with the spectacle of Rio Carnival. Fly above vibrant streets filled with enormous parade floats, thousands of samba dancers in dazzling costumes, live drummers, confetti cannons, colourful flags, fireworks, cheering crowds, and an electric festival atmosphere stretching across the city. Without breaking the shot, continue climbing towards the iconic Christ the Redeemer. As the statue comes into full view, reveal a man sitting on top of the statue's head, perched confidently on the crown of the monument, wearing a black hoodie and holding up a black flag on a pole with his raised arm, the flag rippling in the wind, catching the warm golden hour light against the sky. Execute one smooth, high-speed cinematic orbit around the statue and the man, before flowing into a graceful looping manoeuvre that wraps around the monument and exits with a breathtaking panoramic view overlooking Sugarloaf Mountain, Guanabara Bay, Copacabana Beach, the sprawling city, and the Carnival celebrations below. Maintain uninterrupted forward momentum throughout the entire flight. The camera should never stop, jump, teleport, or cut. Every movement should connect naturally as though captured by a professional FPV drone pilot in one flawless take. Photorealistic. Ultra-realistic FPV drone physics. Golden-hour lighting. HDR. Volumetric sunlight. Natural atmospheric haze. Motion blur. Cinematic lens characteristics. Realistic reflections. Dynamic shadows. Ultra-detailed textures. Hollywood-quality colour grading. Style: Real FPV drone footage. Shot: Single continuous one-take, no edits. Aspect Ratio: 9:16 vertical. Resolution: 8K HDR.
Keep the drawn path a single unbroken line with no gaps — a break in the line is often read by the model as a cut, which fights against the "one continuous take" instruction in the prompt.
No editing needed — what you get is the final clip
Once generation completes, preview the clip. If the drone deviates from the drawn path, regenerate — results vary slightly between runs.
Tap Download to save the MP4 — ready for TikTok, Instagram Reels, and YouTube Shorts.
Posting tip: "I traced my flight path on Google Maps before generating this" is a strong hook on its own — showing the map screenshot side-by-side with the finished FPV clip makes for a great two-slide carousel or before/after post.
Four traced maps and the matching FPV footage they produced — same technique, four different cities
Notice the pattern: a straight line means a fast, direct approach; loops and spirals mean orbit or wraparound moves at the landmark. The shape you draw is the camera move you get.
Ways to remix this concept
| Swap This | For This | Effect |
|---|---|---|
| Rio de Janeiro | Any city with 3D Google Maps coverage | The same tracing technique works anywhere satellite/3D imagery exists |
| Christ the Redeemer | Any landmark — tower, stadium, bridge, monument | Swap the destination while keeping the ground-to-landmark structure |
| Beach launch | Rooftop, street level, or car park start | Changes the opening hook without touching the traced route |
| Loop-orbit at the end | Straight flyover or hard stop | Different drawn line shapes cue different camera moves at the destination |
| Character cameo | No cameo, or a product/logo reveal | Swap the landmark reveal for a UGC-style brand moment instead of a personal cameo |
Generated your version? Tag us — we repost the best results across all our channels.