Grey-peg motion references in Seedance 2.5: what transferred and what drifted across six 6-20 second tests

Six founder-account Seedance 2.5 tests from a historical test period each fed a rendered grey-peg blocking video (camera path and positions), approved scene stills and character reference photos with a long structured prompt. H01 viewed 8-16 timestamped frames of every output and of five of the peg videos. This is a retrospective observation set, not a benchmark and not validation of any current pipeline. What transferred: (1) camera path - the rising reveal at a table-tennis court (13.2 s of 15 s), the hallway move with the same wall sign, and the high ledge view all follow the peg video; (2) foreground occluders - a large grey foreground mass became a real person's arm and shoulder in a fence shot; (3) identity and wardrobe across 20 s of the hallway take. What drifted: (1) a hero prop - a ball changed from smooth grey to worn cream leather between 3.40 and 4.15 s in a shot with no cut visible in the sampled frames; (2) background characters left the table-tennis frame as the camera rose; (3) a close frisbee shot ended on a railing vault with lawn geography that the peg video did not show; (4) a wide ledge vantage kept composition but left the subject too small to judge action. Retiming: the output durations (6.0, 10.0, 15.0 and 20.0 s) differ from the peg videos (6.6, 9.0, 12.7, 14.8 and 19.3 s), so beats shift by up to several seconds; compare beats, not clock times. Practical reading (inference from these sampled takes, untested elsewhere): use a peg video for camera and geography; restate the hero prop in every beat (an untested precaution inferred from the drift above); keep key subjects large enough in the peg framing to judge their action; expect late-shot divergence in long takes. Audio was not listened to.

{
  "citations": [
    {
      "note": "https://veyrul.com/experiments/exp_01M64GNB1DR8WGRH5V8ZB71W70 (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64GN72H9WFRDW83Z3JBSP0W (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64GN883AB03PJYWD3RRS733 (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64GN9X6VHQFC5QM63Y3D56H (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64GN8V1T1SKM67Q29B9BZDC (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64GN7PH6FJ9A9TJ73MN6JRF (candidate; Main resolves to a record ID)"
    }
  ]
}