Four text-only look tests share one order: a wide, two portraits and a closing wide

Four Seedance 2.5 runs of 29 August 2026 (UTC), each from text alone, share a severe-faced look target (everything designed, never photographed) and a very different set. H01 viewed 12 timestamped frames of each, about 1.7 s apart. Seen in the sampled frames: each opens on a wide for three to five samples (a violet shaft over dark roots, a magenta lightning crack, a dark hall with a glowing wheel, a blue hall with an ice anvil and hammer strikes), follows with two close portraits of two different characters at about 9.2 to 14.2 s, and ends on a wide or two-shot with figures (two figures walking forward, a pale and a bald dark figure under lightning, two silhouettes on the wheel, a raised hammer). The faces read as drawn or painted in all four. Caution: four single runs (no seed set); beats between samples are not seen. Reading: with a text-only prompt and a fixed look target, the three-part order repeats across very different sets; this describes these four runs, not a rule. These are retrospective observations of founder tests, not a benchmark or product validation. Audio was not listened to.

{
  "citations": [
    {
      "note": "https://veyrul.com/experiments/exp_01M64VEPKGNCHS0SXVGA3T7SD2 (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64VEVNK6XE7SZEY3HMT46YG (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64VEWX5KBESBQNGJR12N5W8 (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64VEY43H92SY6EJBBAS1ENX (candidate; Main resolves to a record ID)"
    }
  ]
}