Free during beta · Windows first · Apple Silicon planned

The whole field.
One seamless video.

FullField Stitch is a desktop app that turns two overlapping 4K 60fps recordings of a soccer field — shot on GoPro or similar action cameras — into a single panoramic match video. Synced by sound, stitched on your GPU.

Closed beta · no account, no cloud upload — everything runs on your machine. Step 1 of a two-app pipeline with BallTrack AI.

LEFT CAMERA OVERLAP RIGHT CAMERA

Two fixed cameras, 30–40% overlap, one continuous panorama — no operator, no PTZ rig.

How it works

Set up two cameras once. FullField Stitch handles the rest.

1

Record two cameras

Mount two action cameras side by side at midfield, each covering one half with generous overlap. Hit record on both, clap once, and go watch the match.

2

Auto sync & calibrate

The app finds the clap in both audio tracks to align the clips to the frame, then matches features in the overlap zone to calibrate the stitch geometry — one click each.

3

Export the panorama

A GPU-accelerated pipeline warps, blends, and encodes both halves into a single wide match video, with hardware encoding and optional color grading applied on the way out.

Built for full-match footage, not clips

Ninety minutes of 4K60 from two cameras is a lot of pixels. Every part of the pipeline is designed around that.

GPU-accelerated pipeline

CUDA on NVIDIA cards, OpenCL everywhere else — including Apple Silicon. The app picks the fastest available path on your machine at startup, with a CPU fallback that always works.

Audio-clap sync

No timecode hardware needed. A single clap or whistle at kickoff is detected in both audio tracks and used to align the two recordings to sub-frame accuracy.

Cylindrical projection

A cylindrical stitch keeps the ground plane flat and horizontal lines horizontal across the whole panorama — exactly what automated ball tracking and virtual cameras need downstream.

Color LUTs

Built-in .cube LUT presets fix washed-out flat profiles and match the two cameras, applied at decode or encode. Bring your own LUTs too.

Multi-worker export

Long matches export across several parallel stitch workers that split the timeline and mix GPU backends, then join the pieces — so the export scales with your hardware.

Halves & segments

Mark halftime and stoppages, export each half separately or the whole match in one go, and trim dead time before the encode ever starts.

Real numbers, from real machines

Benchmarks are collected automatically by the app on the developer’s own hardware — no marketing math.

28.4 fps
Cylindrical 4K stitch, RTX 3090
0.95×
of realtime at 30 fps output
< 50 min
to export a full 45-minute half

Two apps, one pipeline

FullField Stitch makes the panorama. Its companion app, BallTrack AI, turns a wide match video into broadcast-style footage that follows the ball. Stitch first, then track — or use either app on its own.

FullField Stitch

Two overlapping 4K60 recordings in, one seamless panoramic match video out — synced by sound, stitched on your GPU. A finished match video on its own, and the ideal source for step 2.

Get the stitcher beta

BallTrack AI

A wide match video in, a broadcast-style MP4 out: AI ball and player detection drives a virtual camera that pans and zooms to follow the play. Works standalone with any wide static-camera recording — a FullField Stitch panorama is ideal. Windows + NVIDIA GPU.

Get the tracker beta

Both betas are separate Windows downloads, free while in beta — installers coming soon.