Atmospheric video rewards subtle movement, but wisps can look realistic even when the container they emerge from changes shape. Compare Grok Imagine and Veo candidates on versions and modes listed by your surface. A fixed outline gives the review a stable anchor across soft-detail outputs.
Build the test input
Use an original high-contrast still of a plain black mug on a white shelf. No textures or reflecting objects should compete with rising steam. Request slow pale steam from the opening and a locked-off camera. Place a small paper ruler near the base to track any unrequested movement.
Run the proposed comparison
Use the same aspect and duration when eligible operations agree. Record settings that might affect exposure or animation strength. Review frame order at normal speed, then mark the root of the steam and the depth of the mug lip on a small sequence of frames.
Decide what passes
Pass if the steam stays plausibly attached to the opening and dissipates rather than flashing into new positions. Reject a clip where mug walls stretch, the lip begins to resemble a different material, or steam comes from the shelf. Compare cloud density only after deciding whether geometry held.
Act on the outcome
A result with weak steam may make a good stable plate for a separately composited layer. A dramatic movement is less valuable if it forces you to repair the mug on every other frame. Real production review should use the proposed still and chosen model rather than extending a single studio test to all atmospheric shots.

