Create better AI videos with a more reliable workflow: https://yb.digital/ai
Can Google Flow generate a cinematic AI micro drama with consistent characters? I tested it by creating a 10-second vertical scene using Google's Omni model and my own character reference images.
The scene features three characters: Michelle, Vivian, and Bona. In a tense hotel corridor confrontation, Bona threatens Michelle with evidence of a stolen-money transfer and grabs Vivian by the wrist to force Michelle to reveal his secret.
I provided a detailed prompt, uploaded character references, assigned each character to the scene, and selected the lowest resolution to reduce the credit cost.
But the generation process was far from smooth.
Google Flow rejected the requested 15-second duration because the selected model supports a maximum of 10 seconds. It also required multiple confirmations, experienced generation failures, and took a long time to start rendering.
The final video had major problems:
* The characters did not look like the supplied references.
* Their outfits were different from the requested clothing.
* The wrong character performed an important action.
* The generated scene did not follow the script accurately.
The video eventually generated, but the result took more than a minute and still failed to meet the requirements.
This is my hands-on test of Google Flow for AI video generation, character consistency, and cinematic micro dramas. I also explain why I prefer workflows that let me generate and correct scenes more efficiently.
Try alternative AI video tools here: https://yb.digital/ai
#GoogleFlow #AIVideo #AICharacterConsistency #MicroDrama #OpenArt
