Seedance 2.5 grey-peg test: frisbee close-up take E, ducks out of frame then kicks over the railing (15 s)
Disc passes in the foreground from the right, she ducks out of frame, looks sharply up, and the final frame shows a leg and white shoe kicking over the railing.
Task: Close performance beat with a yellow disc, a whip pan, speech to an off-camera listener, then a pull-back to a railing exit.
Outcome: succeeded
0.75 s: Disc enters from the right in the foreground.
3.76 s: Disc crosses the lower frame.
6.77 s: She ducks so only the top of her head is in frame.
8.28 s: Head thrown back, looking up.
11.29 s: Looking up with parted lips.
12.79 s: Waist-up beside the railing.
14.30 s: A leg with a white shoe kicks over the railing as she tumbles out of frame.
Intent vs result: A more physical ending than the other takes (a tumble over the railing). The grey-peg input for this take was not separately viewed.
Publication update (2026-10-29): the founder has now cleared public use of this output, including incorporated input likenesses and soundtrack. Earlier rights/likeness holds described above are resolved by that confirmation. Reference-input files remain private. Audio remains not listened to; visual observations cover sampled frames only.
Structured fields
{
"inputs": [
{
"description": "4 linked image input(s) recorded in the generation metadata (details withheld)."
},
{
"description": "1 linked start_image input(s) recorded in the generation metadata (details withheld)."
},
{
"description": "1 linked video input(s) recorded in the generation metadata (details withheld)."
}
],
"run_at": "2026-09-24T03:04:52Z",
"metrics": [
{
"name": "output_duration",
"unit": "s",
"value": 15.05
},
{
"name": "resolution",
"value": "854x480"
},
{
"name": "frame_rate",
"value": "24/1"
},
{
"name": "scene_detector_transitions_thr0.30",
"value": 0
},
{
"name": "black_intervals",
"value": "none"
}
],
"procedure": "Seedance 2.5 text-and-reference video generation with a grey-peg motion video (a rendered blocking pre-visualisation that fixes camera path and positions), several approved scene stills and per-character face/front/back reference photos, plus a long structured prompt (look statement, reference-use rules, cast, camera, ordered choreography, continuity, sound). Historical founder test period using an older mannequin-based blocking pipeline. Sampled inspection: 10 timestamped frames from 0.75 s to 14.30 s of a 15.05 s file, plus ffmpeg scene detection at threshold 0.30 and blackdetect.",
"parameters": {
"seed": "unset",
"model": "seedance_2_5 (label in job metadata)",
"multi_shots": false,
"aspect_ratio": "16:9",
"audio_requested": true,
"resolution_label": "480p",
"requested_duration_s": 15
},
"limitations": [
"Sampled visual inspection from timestamped frames (contact sheets), not full real-time playback; motion smoothness, micro-timing and lip-sync are unrated.",
"Audio was not listened to; the file has an audio track but sound, dialogue and music are unassessed.",
"Prompt intent is taken from private job metadata and summarised here only in general terms; the founder's taste and intent for each test are unknown.",
"Single historical generation without a fixed seed; not a controlled benchmark and not validation of any current product or pipeline."
],
"artifact_ids": [
"ast_01M64VPY82QSCD4J55J359QHPN"
]
}Evidence claims
- [sourced_fact] Timestamped observations above come from viewing 10 sampled frames of the output (sampled visual inspection, not full playback).
- [upstream_claim] The generation was run by the founder in an earlier test period; H01 did not run or reproduce it.
- [inference] Ducking out of frame mid-take is a usable cut point.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | unknown | permitted | permitted | unknown | unknown | Original analysis written by H01 (an AI agent) for Veyrul from private viewing; no third-party text, frames or audio reproduced. |
| media | unknown | permitted | permitted | unknown | unknown | Founder confirmed public use of all these Higgsfield outputs on 2026-10-29, including incorporated reference inputs, likenesses and soundtrack. Audio has not been listened to; permission does not establish listening or independent review. |