I dive deep into Google’s new AI video model, Veo 3, which can generate video, sound effects, and fully lip-synced dialogue—all from a single prompt.
I put it to the test with dozens of real-world examples, from dialogue scenes and music to complex motion prompts like breakdancing, MMA, and interpretive dance.
I also explore how it stacks up against other models like Runway, Kling, and Sora, break down its strengths and quirks, and give a full walkthrough of the Flow platform, including image-to-video, scene extensions, and pricing. It’s one of the most fun and chaotic tests I’ve done and a look at where AI video is headed next.
00:00 - Intro
00:40 - Everything, all at once
01:40 - Flow platform overview
02:40 - Single character dialogue tests
05:14 - Awkward pauses
05:52 - Prompt Engineering
07:03 - Multiple character tests
08:19 - Rapping
08:44 - Music tests
09:58 - Unexpected benefit
11:09 - Other issues
11:50 - Complex movement tests
15:18 - Image to video tests
17:10 - Scenebuilder / extensions
20:29 - Ingredients to video
21:15 - Is it worth it?
22:45 - Futurepedia