Seedance 2.5 gothic prompt comparison B: low opening, silhouette and profile close-up
Historical B output with sampled-frame observations and an accessible original. Comparison is uncontrolled; audio remains unreviewed.
Task: Same painterly gothic sequence with added constraints: dust motes not fog, a window as a place not a viewpoint, a red eye as a lit object, figures reading as people.
Outcome: partial
0.63 s: Low close shot of a dark cloak hem beside a glowing floor light pool; rose window softly visible upper right.
3.13 s: Hem moves through the light; the glow stays warm and clean.
8.15 s: Cut to a floor-level horizontal light streak across stone; fine sparks.
11.90 s: Cut to bright rose-window tracery with a dark hooded silhouette in front; no checkerboard texture visible.
16.92 s: Cut to a profile close-up: pale face with dappled light and a glowing red eye in the dark.
19.42 s: Profile remains present in the last sampled frame at 19.42 s.
Intent vs result: Detected cuts at 7.88, 11.58 and 15.83 s, within about 0.3 s of version A; the compositions differ (low hem vs wide nave; silhouette vs flat pattern; frontal vs profile close-up) and version B shows the requested lit red eye.
Main clarification (2026-10-29): the two contact sheets and full-resolution frames at 17 s were inspected. A ends on a three-quarter face view; B ends on a tighter profile with a separate red eye-like light in the dark area beside the face. This confirms a composition difference, not the cause of it. No listening or full real-time playback review was performed.
Structured fields
{
"inputs": [
{
"description": "1 linked 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": "2026-09-02T18:45:20Z",
"metrics": [
{
"name": "output_duration",
"unit": "s",
"value": 20.05
},
{
"name": "resolution",
"value": "1280x720"
},
{
"name": "frame_rate",
"value": "24/1"
},
{
"name": "scene_detector_transitions_thr0.30",
"value": 3
},
{
"name": "black_intervals",
"value": "none"
}
],
"procedure": "Seedance 2.5 with the same reference still and earlier generated video as version A, a prompt about 24% longer that adds explicit never/not constraints, same parameters, no seed set. Sampled inspection: 16 timestamped frames from 0.63 s to 19.42 s of a 20.05 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": "720p",
"requested_duration_s": 20
},
"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_01M64JC3FCMQQYEAHNZG53S8RR"
]
}Evidence claims
- [sourced_fact] Timestamped observations above come from viewing 16 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] The two sampled sequences have similar four-part structure and different framing. The changed prompt, unspecified seed and separate runs prevent attributing the differences to prompt constraints.
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 | The founder confirmed public use on 2026-10-29, covering the uploaded inputs, depicted likenesses and generated soundtrack. Founder-attested clearance; no independent legal assessment. Audio has not been listened to. |