Fortnite Drop Physics & Glider Cut Mechanics: 23,930 Evaluated Landings
This primary empirical study measures every phase of a Fortnite Battle Bus drop directly from Unreal Engine replication flags across 246 cleanly parsed 100-player server replays (23,930 first-drop landings) plus 14 local 43.00/43.10 client replays cross-checked against surveyed M1 terrain rasters.
1. Dataset & Exact Engine State Extraction
From an archive of 583 server-replay headers (builds 41.10–42.20, Hera_V2_Terrain), a stratified sample of 250 replays was split deterministically 70% train / 30% untouched holdout per build. 246 replays parsed with zero extractor errors (175 train / 17,053 drops; 71 untouched holdout / 6,877 drops). For Crossbones Isle (Delilah_Terrain), 9 local 43.10 (CL-60118918) and 5 local 43.00 (CL-59596594) replays were extracted at 0 ms sampling and checked directly against shipped M1 terrain rasters.
Extraction reads exact Unreal Engine replication flags (isParachuteOpen, isParachuteForcedOpen, skydivingState, isSkydivingFromBus, replicatedMovementMode, movementBase), accounting for 110 ms client replication delay. 97.8% (16,678 / 17,053) of train landings execute a post-forced glider cut; 90.5% (15,089 / 16,678) execute a final reopen prior to touchdown.
2. Measured Flight Parameters (Builds 41.10 – 43.10)
All skydiving, gliding, open/close/reopen transition, cooldown, redive, and final-reopen mechanics are invariant from build 41.10 through 43.10. Only Battle Bus horizontal speed varies by island (75.0 m/s on Hera_V2_Terrain vs 54.0 m/s on Crossbones Isle).
| Parameter | Server Train (17,053 drops) | 43.10 Cross-Check (9 matches) | Calibrated M2 Profile |
|---|---|---|---|
| Battle Bus speed & Z | 75.0 m/s (Hera), 800.0 m Z | 54.0 m/s (Crossbones Isle), 800.0 m Z | 54.0 m/s · 800.0 m |
| Bus-exit launch (0–1.75 s) | 7 × 0.25 s, Δz = 35.26 m | Matches ego accel ramp (0–1.75 s) | 7 measured 0.25 s slices |
| First auto-deploy AGL | 100.0 m AGL over surface | p50 101.45 m (ground 99.5–103.0 m) | 100.0 m AGL |
| Open deceleration | 3 × 0.25 s (0.75 s), Δz = 15.00 m | Δz = 14.22 m to 5.00 m/s | 0.75 s (3 slices) |
| Steady full-forward glide | 18.00 m/s horiz · 5.00 m/s down (IQR 0.00) | 18.00 / 5.00 m/s (IQR 0.00) | 18.0 / 5.0 m/s |
| Open → post-forced cut | p05 2.17 s · p50 2.61 s (n = 8,235) | min 2.06 s · p50 2.38 s (n = 9) | 2.10 s total cooldown |
| Close + redive accel | 0.25 s close + 1.00 s accel, Δz = 29.64 m | Δz = 28.21 m to 40.00 m/s | 1.25 s transition |
| Terminal end-redive | 18.00 m/s horiz · 40.00 m/s down (IQR 0.00) | 18.00 / 40.00 m/s (n = 2,731) | 18.0 / 40.0 m/s |
| Final reopen AGL | Fast direct p50 14.36 m (all p50 14.98 m) | Ego p50 16.86 m · Lobby p50 18.57 m | 15.24 m AGL (+0.88 m) |
| Final reopen decel | 4 × 0.20 s (0.80 s), Δz = 11.97 m | Δz = 12.71 m to 5.00 m/s | 0.80 s (4 slices) |
3. Measured Steady Dive Polar (262,497 Samples)
Time-weighted medians and interquartile ranges of vertical descent speed across horizontal velocity bins in the server train dataset.
| v_h bin (m/s) | Samples (n) | v_h p50 (m/s) | v_down p50 (m/s) | v_down IQR (m/s) |
|---|---|---|---|---|
| [0, 2) | 1,735 | 0.17 | 60.00 | 0.05 |
| [2, 5) | 2,857 | 3.80 | 59.38 | 0.52 |
| [5, 8) | 7,857 | 6.84 | 57.99 | 0.88 |
| [8, 11) | 17,144 | 9.69 | 55.89 | 1.34 |
| [11, 14) | 25,337 | 12.58 | 52.85 | 1.89 |
| [14, 16) | 18,983 | 15.04 | 49.38 | 1.63 |
| [16, 17.5) | 16,960 | 16.77 | 46.32 | 1.48 |
| [17.5, 18.5) | 171,624 | 18.00 | 31.25 (p50) / 41.76 (p90) | 12.41 |
4. Untouched Holdout Validation & Cursed Cavern Regression
Evaluated with zero refitting on the untouched 71-match server holdout, the M2 phase model achieves a 0.50 s median absolute arrival error (p90: 1.28 s) on fast clean 1-cut drops.
| Cohort | Evaluated | Median |err| | p75 |err| | p90 |err| | Median Bias |
|---|---|---|---|---|---|
| Server Holdout · Fast Clean 1-Cut (71 matches) | 1,130 / 1,140 | 0.50 s | 0.88 s | 1.28 s | +0.15 s |
| Server Holdout · All Direct 1-Cut (71 matches) | 2,775 / 2,789 | 0.65 s | 1.23 s | 2.11 s | -0.31 s |
| Server Holdout · Top-10 Fastest/Match (71 matches) | 647 / 710 | 0.57 s | 0.97 s | 1.33 s | +0.46 s |
| 43.10 Local Direct Ego (CL-60118918, dir >= 0.90) | 7 / 7 | 0.64 s | – | 0.91 s | -0.34 s |
Regression Benchmark: Cursed Cavern (Moosinsel Chest, 59.21 m Z)
Crossbones Isle scenario (58.81 m surveyed M1 terrain, 54.0 m/s bus at 800.0 m Z): legacy rough estimate lands at 36.7 s; M2 single-glide lands at 34.74 s (jump 8.92 s); M2 Cut + End-Redive + Catch touches down at 31.00 s (jump 10.52 s, cut at 26.33 s / 135.78 m AGL, final catch at 29.74 s / 15.24 m AGL) — 3.73 s faster with verified 200/200-cell terrain clearance.
Primary source: Fortiq Dropmap Benchmark Record (specs/160-professional-dropmap/benchmark.md, 2026-11-21). Aggregated evidence receipt SHA-256: ff0a3f7d1b96ef4fb0489fd1f779a71b9c6f4b3c7d9e163777473d710c9a3c04.
Drop Physics Research FAQ
How often do competitive Fortnite players cut their glider on initial drop?
Across 17,053 training landings from 175 server replays, 97.8% (16,678 / 17,053) executed a post-forced glider cut, and 90.5% (15,089 / 16,678) triggered a final reopen before touchdown.
At what height AGL does final glider reopen happen in Fortnite?
The calibrated M2 model places final reopen at 15.24 m AGL (14.36 m measured direct-redive median plus 0.88 m character capsule offset), followed by 0.80 s deceleration to 5.00 m/s descent.
What is the Battle Bus speed on Crossbones Isle in Patch 43.10?
On Crossbones Isle (Delilah_Terrain, 43.00 and 43.10), the Battle Bus travels at 54.0 m/s at 800.0 m world Z, compared to 75.0 m/s on Hera_V2_Terrain. Skydiving, gliding (18.00 / 5.00 m/s), and terminal end-redive (18.00 / 40.00 m/s) velocities are invariant across builds 41.10–43.10.
How much faster is glider cutting compared to continuous gliding?
In the verified Cursed Cavern regression scenario (Moosinsel chest spawn at 59.21 m Z over 58.81 m M1 terrain), M2 Cut + End-Redive + Catch touches down at 31.00 s bus clock vs 34.74 s for single-glide — saving 3.73 seconds.
Apply These Measurements on the 43.10 Map
Calculate your exact jump vector in guest mode with no account wall, or explore all 12 Crossbones Isle POIs.