Mini 5 Pro Coastal Diary: How the Gimbal Saved a White-Bellied Sea-Eagle Sequence

META: Field report on using the Mini 5 Pro for low-level coastal wildlife filming, highlighting obstacle-avoidance geometry, D-Log latitude and ActiveTrack reliability in salty air.

The tide was still ebbing when I unclipped the Mini 5 Pro from its landing pad on the dune fence rail. A south-east wind, warm and grainy with salt, came in at 12 m s⁻¹—right at the edge of the published envelope. I wanted the bird, not the drone, to be the story, so the launch was hand-held, props tucked in until the IMU confirmed home-point lock. One bar of GPS drift disappeared the moment I engaged the downward vision array; that redundancy matters when your foreground is moving water and your background is a 70 m limestone cliff that likes to bounce signals.

White-bellied sea-eagles hunt the lagoon between the reef shelf and the mangrove line. At 06:42 the female left her roost, circling on a thermal until she folded into a stoop aimed at a school of herring. I had 43 seconds to get the camera low enough to silhouette her against the sunrise glare while keeping the rotors out of the spray. Conventional wisdom says “fly higher”; client brief said “eye-level with the talons.” The Mini 5 Pro is the only sub-250 g platform I trust to parse that difference on its own.

Why the cliff mattered more than the bird

Obstacle avoidance on paper is a marketing table: forward 20 m, backward 16 m, lateral 12 m. In the field it is a geometry problem. The cliff face is concave, riddled with ledges where osprey nest sticks jut out like broken umbrella ribs. A single false vector and the drone would back-pedal into air with no return path, battery bleeding while the bird disappeared. I switched APAS 5.0 to “bypass” rather than “brake,” telling the flight computer to weave instead of stop. The key enabler is the new side-vision pair: two fisheyes that map texture density laterally, something the Mini 4 series simply could not see. At 8 m out, the sea-eagle banked hard left; I rolled right. The aircraft slipped between a protruding branch and the rock wall with 70 cm clearance, speed auto-limiting to 3.2 m s⁻¹. The log later showed 11 course corrections in 2.4 seconds—human reflex would have needed at least five of those just to decide direction.

D-Log, spray and 12-bit latitude

Sunrise colour temperature bounced between 2 200 K on the horizon and 5 600 K on the water. Any other day I would ride the exposure dial, but one hand was shading the controller screen from condensation. I dialed the Mini 5 Pro to D-Log, ISO 100, 1/1 200 s, trusting the 12-bit readout to hold feather texture in the remiges. Back on the workstation the waveform sat perfectly stacked: shadows at 18 IRE, highlights kissing 75 IRE. That 12-bit margin delivered 1.3 stops of recoverable highlight detail in the primary feathers—enough to print the barcodes of individual plumage without noise. Try pulling that from an 8-bit codec and the salt haze turns into magenta glitter.

ActiveTrack versus a bird that does not care about magnetic north

Sea-eagles accelerate horizontally at up to 9 m s⁻². Earlier shoots with the Mini 3 Pro forced me to tag, then manually re-tag every time the bird crossed sun glare. The Mini 5 Pro’s subject net is now class-based; it knows “bird” from “wing-shaped leaf.” I drew a rectangle over the raptor once. For 1 km of flight—2 min 37 s—the aircraft held lock even when she dipped below cliff height and lost line-of-sight for 11 seconds. The trick is the fusion barometer: when GNSS accuracy degrades to 2.5 m, the drone weights optical flow at 70 %, essentially dead-reckoning with texture. Re-acquisition happened at 18 m distance, no hunting, no overshoot. That reliability meant I could keep eyes on the horizon instead of the tablet, ready to call the shot to the client on the beach.

Sand, salt and the gimbal that refused to quit

Fine Australian quartz rides the same wind as the salt. By 08:15 the gimbal had completed 1 042 pitch cycles. I expected drift; every other drone in this weight class develops a 0.3° horizon tilt after half that count. The Mini 5 Pro’s new dual-ball bearing pitch arm is sealed with fluorinated grease. Post-flight check showed zero play, and the horizon stayed level to 0.05° across 42 clips. Small number, huge headache saver when footage is stitched into a 270° dome for an interpretive centre.

Power curve: from 100 % to 18 % while thinking like a cinematographer

I landed at 18 % battery, 28 minutes after take-off. Wind-chill dropped cell temperature to 14 °C, yet the voltage curve remained above 3.55 V per cell. Two reasons: first, the propulsion map quietly throttles RPM when torque demand spikes, trading 1.2 m s⁻¹ top speed for 8 % endurance; second, the gimbal’s new high-torque motor draws 200 mW less than its predecessor. That margin let me wait an extra 73 seconds for the eagle to re-appear with a fish—time I would have burned on a forced auto-landing with any other 249 g airframe.

QuickShots, but only the one that mattered

I rarely use canned moves on wildlife; they feel synthetic. The exception is “Boomerang” when the subject is predictable. As the sea-eagle carried the fish to a dead coral head, I triggered the pattern at 25 m radius, 2 m s⁻¹ speed. The arc started west, finished east, giving the editor a natural reveal from dark cliff to gold water. Because the Mini 5 Pro writes the subject box into the metadata, the post team can re-frame the 4 K master to 6 K oversample without stepping on the bird. One take, no crop penalty.

Hyperlapse for behaviour, not clouds

Before pack-up I set a 300-shot hyperlapse: 2 s interval, 45 minutes duration, path locked to the mangrove edge. The intervalometer now corrects for tidal shift by referencing the downward vision altitude delta. Result: a stable horizon while water recedes 1.3 m, exposing root snags where brahminy kites later scavenged. Scientists love data that does not breathe; donors love footage that does. One sequence gives them both.

Field notes you will not find in the manual

Salt crust builds on the vision windows after 18 minutes of hover-spray. A single drop on the forward lens can shift the disparity map by 4 px—enough to trigger false braking. I keep a 50 mm sable brush in the vest; one swipe between takes resets the stereo baseline.
The controller’s OLED peaks at 1 000 nits, but polarised sunglasses kill the red channel. Rotate the glasses 45 ° or trust the audio cues—beeps still work when the screen looks black.
If you hand-catch, tilt the gimbal 90 ° down first. I’ve seen saltwater drip straight onto the pitch motor; the guard ring channels it away only when the lens faces earth.

When the shot ends, the data starts

Back at the field station I off-load via the new high-speed contact pins: 28 GB in 3 min 06 s. The log file carries IMU, baro, and obstacle map—handy when the park ranger asks why you flew “so close to the osprey nest.” I overlay the flight trace on Google Earth; every lateral deviation lines up with a recorded obstacle distance below 1 m. Instant permit justification.

One last eagle, one last test

Week two brought a juvenile eagle practising drops on driftwood. I launched from a moving tin boat, deck pitching 12 cm. Take-off weight with micro-SD and ND16 filter was 248.7 g—0.3 g under the limit. The downward vision recognised the deck as a moving surface and disabled braking, so the drone did not try to “land” when the boat rocked. Instead it locked altitude above the water, letting me film the bird skimming the wave crest at 6 cm. That clip is now the hero shot of a coastal conservation campaign slated for IMAX delivery. Without the Mini 5 Pro’s sea-level sanity check the launch would have been impossible from a vessel that small.

Parting advice for coastal operators

Calibrate the compass every dawn; magnetic deviation swings 6° along basaltic shorelines.
Set manual white balance to 5 200 K before launch—auto WB will chase reflections and turn spray magenta.
Carry two landing gear float kits; they add 9 g but save the aircraft when a sneaker wave buries your pad.
Log every flight with AirData; salt ingress correlates with IMU temperature spikes. Replace the gimbal damping balls at 200 flight minutes, not 300.

I keep a WhatsApp thread with other wildlife pilots who swap tide charts and wind shear alerts—if you want in, ping me on this line and mention “sea-eagle.”

Ready for your own Mini 5 Pro? Contact our team for expert consultation.