Three table scenes ask for stylised faces and show photographic-looking ones

Three Seedance 2.5 runs of 27 August 2026 (UTC), each from text alone, set a woman at a round table in a dark amphitheatre with red gems, then two armoured men in close-up. Their prompts ask for stylised faces and exclude photographic skin (one names an anime-first target, one a 3D target, one an anime-influenced target). H01 viewed 12 timestamped frames of each. Seen in the sampled frames: the set, the spotlit disc and the red light read as lit and modelled, and the three faces (the woman and the two men) read as photographic, with skin texture and film-like lighting, rather than stylised. The beat order is the same in all three: a wide, a table shot, a red-light insert, a spiky grey-haired man and a long-haired pointed-ear man. Caution: 'reads as photographic' is H01's judgement at the sampled size (frames about 250 px wide), not a measurement; no person is identified. Reading: in these three single runs (no seed set) the face part of the look target was not met; this describes these runs, not the model or the prompt style. 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_01M64VDHXEFWM8H1N50S73JWWF (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64VD6X6TV65PQ1MQEVY41D0 (candidate; Main resolves to a record ID)"
    },
    {
      "note": "https://veyrul.com/experiments/exp_01M64VDJJGR97BMG12ZNNFKRDY (candidate; Main resolves to a record ID)"
    }
  ]
}