Voyager vertical-video comparison, September 18 2026

Source: generated moving-target diagram, 3840x2160, 24fps, 12 seconds.
It is not a person-tracking, object-detection, or AI Smart Reframe benchmark.

Open Voyager-vertical-comparison.drp in DaVinci Resolve. Relink horizontal-source.mp4
from this folder if media is offline. The three timelines are:
01 - Fit whole frame: full source retained, letterboxing.
02 - Center crop: fills 1080x1920, subject leaves the frame.
03 - Follow the subject: native Position keyframes keep the target centered.

The three outputs were rendered in Resolve Free 21.1. The MP4 copies preserve
the native H.264 picture without re-encoding; the clips are silent.
The optional FCPXML files reference absolute source paths, which need updating
on another computer. Use the DRP and relink media for the simplest handoff.

For the verified follow timeline the Inspector Position X is +270 at 0s, -270 at
6s, and +270 at the 12s end boundary. Zoom is 1 under the imported fill sizing.
These values are specific to this source and sizing mode. The last visible frame
is 11:23, and the final keyframe defines interpolation to the out boundary.

verification.json records a color-based target check over all 288 frames using
135x240 preview frames. This verifies the controlled teal target, not arbitrary
footage. create_fixture.py creates the source and combined FCPXML; it needs
FFmpeg and the macOS Arial font. verify_renders.py expects the native MOV renders
and needs NumPy and FFmpeg.
