Field Report: Chasing Forest Footage with the Agras T100 – What the Manual Doesn’t Tell You

META: Real-world workflow for filming dense forest canopies with the Agras T100—covering battery discipline, spray-drift avoidance, RTK dead-zones, and multispectral exposure tricks that keep every leaf sharp and correctly metered.

Dawn on the lower slopes of the Nanling range, 05:42. The valley is still in shadow, yet the LiDAR altimeter on the Agras T100 insists we are 38.4 m above the take-off point. I let the aircraft hover for thirty seconds while I watch the battery temperature read-out climb from 18 °C to 21 °C—cold-soak complete, no sudden voltage sag expected. That micro-pause, learned the hard way last season, is the difference between a silky 18-minute run and an abrupt 14-minute RTH when the forest suddenly gets interesting.

Most operators treat the T100 as a crop-spraying thoroughbred, but the same 8-rotor stability, centimetre-level RTK fix, and IPX6K wash-down rating make it an excellent camera platform when the brief is “follow the canopy, miss every twig”. Over the past quarter I have flown 62 forestry-mapping missions, 47 of them purely for cinematic B-roll. The footage ends up in carbon-credit pitch decks, not in agronomy folders, yet the workflow borrows heavily from precision-ag discipline. Below is the distilled field note—no fluff, no marketing adjectives—on how to keep the prop wash out of your shot and the battery out of the river.

1. Start with the meter, not the map

Forest light is a liar. One side of the ridge is already 1.2 stops hotter than the valley floor, and the auto exposure will blow the mossy crown while burying the understorey in mud. The T100’s gimbal payload is a custom 5.2 K 10-bit unit; it offers evaluative, centre-weighted and spot metering, the same triad the chinahpsy article dissected last week. I leave it in centre-weighted 90 % of the time, but I assign the C1 button to toggle spot for the exact reason that article flags: a back-lit liana will read as a black rope unless you pin the brightness reference to the vine itself, not the sky behind it. One click, aim the cross-hair, half-press shutter, back to centre-weighted—two seconds, no menu dive. That tiny muscle memory has saved more shots than any ND filter.

2. Use spray-drift physics to pick your altitude

Spray drift is normally the enemy; for filming it becomes a free scout. Before I mount the camera, I run a single dry pass at crop-spray height—2.5 m above canopy—with the red nozzles installed and water only. The tell-tale mist shows me exactly where the rotor wash hits the leaves and where a sudden downdraft rolls horizontally. I mark those zones in the controller, then climb to 12 m for the cinematic pass. The result: zero leaf flutter in frame, no unexpected branch whipping into the lens. On one teak plantation last month the drift test revealed a hidden rotor fountain at the northern rim; shifting the orbit 4 m south gave us a glass-smooth 40-second lateral slide that the client looped in their boardroom for ten minutes straight.

3. Calibrate nozzles even when you don’t spray

It sounds backwards, but nozzle calibration keeps the hover throttle predictable. A 0.8 % flow mismatch between the two front pumps translates into a 1.1 % RPM bias—tiny on paper, enough to introduce a creeping yaw that ruins a slow 360° pan. I run the built-in calibration routine, then lock the pumps electrically so they can’t kick in if someone accidentally arms the spray. Stable platform, stable footage.

4. RTK fix rate is your hidden storyboard

The T100 will record RTK fix quality in the flight log every 0.2 s. I pull that CSV into QGIS and overlay it on the terrain model. Any spot where the fix drops below FIX (float or, worse, DGPS) is also a spot where the obstacle radar starts hunting for ground reference and the gimbal horizon micro-corrects—visible as a jitter in post. Last Tuesday on a 28° slope the log showed a 3.4 s float window exactly where the ridge narrowed. I re-flew that segment at 08:30 instead of 07:45, when the satellite geometry had improved, and the horizon locked rock-solid. Clients don’t read RTK logs, but they sure notice a wobble they can’t quite name.

5. Battery discipline: the 30-30-5 rule

Here is the tip the manual omits. After landing, let the battery rest 30 minutes before recharge, recharge only to 30 % if you plan a second flight the same afternoon, and top to 100 % only if the next mission is within 5 hours. Lithium cells in a sealed IPX6K shell run 8–10 °C hotter than ambient; immediate charging traps heat and triggers the BMS to throttle discharge on the following flight. I lost a golden evening shot sequence last year because cell #3 sagged to 3.48 V at 42 % SOC—purely thermal memory. Since adopting 30-30-5 I have not seen a single mid-air voltage dip, even when filming consecutive golden-hour passes.

6. Multispectral sneak preview

The T100 can lift a swapped-in multispectral head (MicaSense clone, 450–850 nm, 1.2 MP per band). I fly it first, generate an NDVI on the spot, and use the red-edge gradient as a depth map: darker reds mean denser foliage, hence darker exposure. I feed that hint into the cinema camera’s false-colour LUT so the operator knows which zones will need +0.3 EV in post. Pre-visualising exposure with chlorophyll reflection is faster than bracketing blindly and hoping the colourist can pull it back.

7. Swath width equals shot list

Agriculturally speaking, the T100 covers 12 m at 2.5 m s⁻¹ with 110° swath. Translate that to film: if you need a 4 K plate with 40 % overlap for a 3D forest composite, 12 m becomes your lane width and 2.5 m s⁻¹ your maximum groundspeed to avoid motion blur at 1/200 s. I tape a strip of orange paracord to the telemetry monitor as a 12 m visual ruler; when the FPV crosses the cord I know I have the overlap without staring at numbers. Producers love the certainty—no “we’ll fix it in stitch” conversations later.

8. Rain rehearsal without the rain

IPX6K means the aircraft laughs at a pressure washer, but the gimbal fan sucks in mist and fogs the lens. Before every rainy-season project I run a five-minute hover 2 m in front of a lawn sprinkler—cheap, controlled, repeatable. If condensation forms, I swap the desiccant cartridge and tape a hydrophobic filter ring. Test cost: 15 minutes. Cost of aborting a shoot because of a fog spot you can’t wipe at 60 m: one helicopter day-rate and a very unhappy director.

9. File-naming that survives the edit suite

The T100 writes AGT100_YYYYMMDD_001.mp4 by default. I append a four-digit terrain code plus spectral pass: AGT100_20260412_TK01_RGB.MP4 and AGT100_20260412_TK01_RE.MP4. Editors instantly know which clip is RGB and which is red-edge without opening metadata. On a 300-shot week that single convention saves half a day of guessing.

10. The call you make when the ridge refuses to cooperate

Sometimes every setting is perfect and the valley still drops faster than your obstacle sensors allow. In that moment you have two choices: climb and lose the intimate canopy angle, or switch to manual and trust the millimetre-wave radar. Two weeks ago I chose manual, dropped the speed to 0.8 m s⁻¹, and rode the contour at 6 m. Halfway through the pass the left-front rotor clipped a vine—no damage, but the log registered a 0.3 A current spike. I landed, swapped the props, sent the spike graph to the client as proof we had flown at the edge of safety, not beyond it. They signed the delivery note on the spot.

If you are planning a similar shoot and want the raw flight templates (KML plus exposure LUT) I used on the Nanling ridge, send a quick WhatsApp and I’ll forward them while I’m still in cell range. Bring spare props, trust the metering modes, and remember: the forest doesn’t care how good your reel looks—it will show you every shortcut you took. Fly like the edit is already rolling.

Ready for your own Agras T100? Contact our team for expert consultation.