# Initial visual findings

These are qualitative observations from sampled frames, not a statistical ranking. All takes remain available in the comparison.

- The character set is substantially stronger for review than the earlier mixed batch: the elf, wizards, sorceresses, succubus, and armored fighters keep recognizable silhouettes across both variants.
- The object cases make small motions easy to inspect. Chest lids, the potion liquid, the compass needle, and the ramen-stall lanterns move without changing the source framing.
- The requested frame counts scale the returned clips as expected: 8/12/16/20/24 requested becomes 9/13/17/21/25 returned, including the initial frame.
- Action-driven effects appear where the action asks for a spell cast, staff sweep, or sword draw. The breathing golem, compass, potion, and city-light cases do not need an effects instruction, and their sampled frames stay comparatively restrained.
- The strongest visible differences are motion choices rather than global image quality. For example, the staff and sword cases show clearer directional movement in some Nik samples, while several original samples retain a more conservative pose.
- The largest caution is prompt compliance: every Nik output has the exact locked-camera ending, but five prompts exceed the requested 80–100 words. The animation model can also introduce or omit effects even when the prompt is conditional.

No overall winner is asserted from one generation per variant. The paired controls and exported votes are provided for review.