The human eye sees a lightning flash as a single instantaneous burst. Two Vision Research Phantom T2410 cameras record at up to 525,000 frames per second โ revealing stepped leaders probing downward, a bright return stroke racing upward, and dart leaders re-igniting spent channels in milliseconds.
2T2410 units
525kfps max
24kfps @ 1 Mpx
24Gpx/s
How it works
The Phantom T2410 uses a back-side illuminated (BSI) CMOS sensor with a global electronic shutter and 24 Gpx/s throughput. At 24,270 fps (1280 ร 800) each frame lasts ~41 ฮผs โ enough to resolve stepped-leader structure. In binned mode at 100,000 fps (640 ร 384) the return-stroke luminosity wave, travelling at roughly one-third the speed of light, can be tracked frame by frame up the channel.
What is captured
Stepped leaders โ the invisible-to-eye downward-propagating discharge that branches and pauses in 50 ฮผs steps. Return stroke โ the bright upward-propagating wave triggered when a leader contacts ground. Dart leaders โ fast restrikes that follow the existing channel in subsequent strokes. Corona and recoil leaders visible in upper branches.
Role in RTL3D
Two Phantom T2410 units provide visual ground truth for the 3D source locations produced by the LF and VHF networks. Frame-by-frame channel images are IRIG/GPS time-stamped and overlaid on the 3D point cloud, confirming that mapped sources trace the actual visible channel geometry and branching structure.
Camera platform
Vision Research Phantom T2410 โ compact T-Series high-speed camera with BSI sensor for light-starved outdoor capture.
30 fps (standard video) โ the entire lightning event collapses into one or two saturated white frames. Channel structure, branching, and stroke sequence are completely lost.
1,000โ10,000 fps โ individual strokes become separable; stepped-leader tip luminosity becomes visible as a faint advancing point in each frame.
50,000โ525,000 fps โ the return-stroke luminosity wave can be tracked upward frame by frame; dart-leader propagation speed can be measured directly from the footage.
Overlaying the frame sequence on RTL3D 3D source maps confirms spatial accuracy of the time-of-arrival solution to within the channel width.