The camera was built from the spots that actually got filled: the first
filled one for the front, the average of the filled ones for the centre.
So it moved with the finishing order and the head count, and would move
between machines if a remote player's vehicle was not there to place.
Measured on Wiseguy's Wake, where win1 sits at (1199.84, 3, 2.92) and
the eight spots run back to z~38 on the high tier:
one finisher eye 1199.84,15,-33.08 aim z 2.92
eight finishers eye 1199.82,18.44,-9.03 aim z 26.97
Twenty-four units closer to the stand and looking at the middle of the
tiers instead of at the winner - a different photograph of the same
podium depending on how many people showed up.
The stand is fixed furniture on every map, so the shot comes off the
geometry now. The eight dropzones are read once, up front, before
anybody is placed; win1 anchors the framing and the axis from the back
rows out through win1 gives the facing, so a map that mounts its stand
at another angle is still photographed from the front rather than
relying on the old (0,0,-1) fallback. Placement then runs as its own
pass and is the only thing that cares who finished - the log reports
"N placed on M spots" precisely so a changing N beside unchanged camera
numbers is visible.
The framing constants are untouched and so is the shot they produce:
the new numbers for a single finisher are eye 1199.77,15,-33.08 aim
1199.84,5,2.92, which is the old single-finisher shot to within 0.07 in
x. That is the case the standoff/height/aim defaults were dialled in
against, so the approved photograph is what everybody gets now instead
of what one person got.
Verified by running a race to the podium: the eight spot positions log
as expected, the camera numbers match the calculation, and the frame is
the same one as before - rank 1 centred with its callsign, 2 and 3
flanking on the low tier, 4-8 across the high tier behind.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
CydandClaude Opus 5
a1f9c0e3c0