Same bench, day and dusk

The light moves from midday sun to dusk on the same bench with identical framing. Checked frame by frame.

Midday sun with hard shadows, then dusk after a cut at about 3.5 seconds, on the same bench with identical framing. Checked frame by frame.

Midday then dusk on the same bench with matching framing and surroundings.

Midday then dusk on the same bench with matching framing, ironwork and background trees.

Bench geometry and framing stay locked across the midday-to-dusk change.

Bench geometry and framing stay locked, only ambient lighting changes.

Bench geometry and framing stay locked across the midday-to-dusk lighting change.

Framing and bench geometry stay locked while lighting genuinely shifts to dusk.

Framing and bench geometry stay locked while lighting genuinely shifts from midday to dusk.

Bench geometry and framing identical between midday and dusk shots.

Real cut with an actual midday-to-dusk lighting change (lamps lit, atmosphere shifted) -- the only one of the models benchmarked so far to render this prompt as intended rather than a single static/continuous shot.

Rendered as a single static time-lapse (day-to-night) rather than multiple shots; a defensible reading of the prompt, lighter fidelity discount than the other multishot prompts.

The requested midday-to-dusk lighting change never happens -- verified via frame extraction: first and last frames both already read as dusk/blue-hour, no midday state shown at any point. Confirmed prompt-adherence failure, not a hallucination.

FAILURE MODE: single continuous shot, stays in midday light, the requested dusk transition never happens.

FAILURE MODE: single static shot, no midday-to-dusk transition delivered at all, the prompt's core requirement (time change) is simply missing, though what is there is stable.