Fluid motion in one region of a generated clip can look detailed while the contact between the water and an object is impossible. Use a static waterline test before relying on a scene where the material change is central. OpenArt lists Veo and Wan versions as external research candidates.
Build the test input
Draw an unbranded clear tumbler with a pale blue block standing half-submerged. Use a controlled script where a hand gently raises the block straight upward. Avoid splashing initially. Mark the original fill line on a reference overlay independent of the scene area.
Run the proposed comparison
Use identical crop and clip duration where the model operations share them. If a shorter clip is the common option, test only the lift. Keep camera movement off. Retain versions and operations as separate labels, because platforms can expose different modes under related family names.
Decide what passes
Approve a take only if the waterline remains attached to the glass, the block visibly crosses it once, and the visual wet edge stays coherent through the break. Flag a static stain that floats up with the entire container. Look for quick rebound artifacts just after the block clears the surface.
Act on the outcome
If the contact point is sloppy but lifting is natural, a cut into a close image at the moment of contact can preserve the sequence. Choose the winner by the boundary viewers would see. Judge more turbulent variants separately after the basic contact condition has passed.

