Hello,
I am a paying Google AI subscriber and I regularly purchase additional Flow credits.
I am currently using Flow/Veo for a professional fashion film project and have conducted an extensive series of controlled tests.
My objective is very specific:
- preserve a reference image with maximum fidelity,
- preserve the identity of the character,
- preserve a complex environment,
- animate only selected environmental elements (for example translucent flower petals),
- without globally reinterpreting the scene.
I have tested multiple workflows, including:
- single reference image,
- multiple reference assets,
- with and without Instructions,
- different prompt strategies,
- Omni and Lite models.
The observed behavior is always similar:
- if the environment is preserved, the environmental elements remain static;
- if the environmental elements become animated, the entire environment is reinterpreted.
This makes it impossible to produce selective environmental animation while maintaining faithful reconstruction of the original scene.
My question is:
Is this a current limitation of Flow/Veo?
If so, are there any existing or upcoming features that allow selective regional animation (for example masking, motion regions, static regions, or local motion control)?
This capability would significantly expand Flow’s usefulness for professional creative and fashion production.
Thank you for your work and for any clarification you can provide.