Character selection, the presentation bar and the camera

A SON SAYIM match does not start at the table. The character selection screen opens first: everyone gets two offers from the pool, a shared timer runs, and if everyone at the table picks early the phase ends right there. The strip at the bottom of the screen shows who has chosen and who is still thinking.

The network side of that screen was completed in the eleventh wave. Until then free selection was only wired end to end on the single device path; reserving a character from the gallery in an online room did not work. Reservation moved to the server, so two players can no longer take the same character and the bots no longer produce duplicates.

The look of the hand at the table was rewritten in the tenth wave. The opponents' hands were not actually three dimensional: two dimensional card backs were moved to the screen projection of the hand every frame, which is why a card kept riding over the arm. The problem was the method, not the position. The same wave measured and rebuilt the weapon grip as well, after a deviation of one hundred and seventeen to one hundred and thirty two degrees turned up between the bodies.

The camera is a story of its own. Week seven brought a three phase camera that follows the player leaving the table. In week nine the user said the old angle was better, so the side follow was switched off and a distance trigger at the door took its place. In week ten the same framing was verified frame by frame in a real online match.

The animation of that same walk was cleaned up in the eleventh wave: the turn at the threshold was removed, the door gesture moved to its own layer and five more clips got a posture fix. The measurement is clear: knee hyperextension went from eighteen and a half degrees to zero and foot slide came down to a quarter of a centimetre per frame.

The nicest finding of the twelfth wave was the presentation bar. The timer at the top did not come from the animation but from the cost table of the presentation queue, and it printed a fixed duration for every seat. The real scene, meanwhile, varied from seat to seat. The overcount was measured at between two and a half and six and a half seconds and removed; the bar now counts the same scene you see on screen.