It was just an overview of two VEO 3 videos of 8 seconds, but as for so many things that change life, I will never forget my first generation of audio and video synchronized with a skillfully designed prompt.
I currently perform Google AI Pro, the $ 19.99 per month account which gives you access to the Gemini 2.5 Pro model and, more importantly, a limited trial of the VEO 3 videos.
Veo 3 is the level of generative video creation which, for the first time, allows you to create videos with a dialogue, background noises and sound effects, all synchronized with action.
Although I understood that my Veo 3 access could be limited, I did not know how many videos I could generate with the new model. The answer, it seems, is exactly two. If I want unlimited access, I can go to Google AI Ultra for $ 249.99 per month (there is a three -month offer for $ 124.99 per month). And Veo 3 is currently US only.
From Veo 3 launched at Google I / O 2025, my tiktok flow has been filled with these incredible and often quite realistic AI clips. Some resemble infopublicities or advertisements, others are simply impossible, like a woman questioning a smiling man who is clearly on fire.
I was torn between the creation of realism, hyper-realism and something fantastic. In the end, I built an prompt in the Gemini 2.5 Pro window which supports video creation which was a mixture of science fiction, drama and fantasy.
Writing inside the quick window, however, turned out to be an error because I accidentally struck the return before moving away from my idea, and suddenly, Veo 3 was busy generating my video.
It was my first prompt:
“Bill and Jessica live in a log cabin built on the surface of Mars. Bill emerges from the cabin to find Jessica fighting a Martian using nothing other than a plush animal.
Bill shouts at Jessica: What are you doing?
Jessica: This fucking Martian wants our land and he cannot have it. “”
As you can see, there are not many details, and as easy as generating a video in Veo 3 (and the Veo 2 without audio), you will get a better result including more details and dialogue. Veo 3 will not make the characters say everything you don’t have a script. In this case, because I reached the return too early, Jessica’s dialogue is cut and I could not polish my prompt.
Even so, Veo 3 took the rare details and in about 5 minutes created a striking video. Take a look (ring the full effect).
It is far from perfect. Bill does not really speak his line, although we hear him from the camera outside the camera. Jessica’s cry (or is it the Martian?) Also comes from outside the camera.
There is an unfortunate sound effect that could come from Bill, and that I did not script. In addition, I don’t know why Jessica talks about her lines directly to the camera.
Again, I guess if I had directed to whom she should talk to, Veo 3 could have made a different choice.
However, there are so many more subtle things that are impressive. Veo 3 obtains the correct setting; Notice the reddish cover of daylight March. The Martian is terrifying. I am more impressed, however, by the sound effects such as the sound of the cabin door, the steps on the Martian soil and the sound of the animal by plush the chest of the Martian.
Take 2
For my second invite, I wrote it and edited outside the Gemini. I did my best to define the scene, describe the characters and delimit dialogue and sound effects. Here is the prompt:
The scene is a lush forest with the sunlight flowing from the general costs. We hear the cries of the pterodactyls in the background and the sound of the leaves swinging in a light breeze.
A tyrannosaurus carefully paints a large canvas which represents a colorful image of a man on the point of being destroyed by an asteroid.
The Tyrannosaurus sings quietly for himself: “Pink Pony Club, I will continue to dance at …”
A velociraptor walks and asks: “Why do you comb that?”
The tyrannosaurus: “AI made me do it.”
Velociraptor is back with horror and says: “What? !!”
As you can see, I was, in part, inspired by some of the self-referential videos Veo 3 that I had seen on Tiktok where the characters break the fourth wall and mention that they are in a video. Although my retail work has mainly borne fruit, Veo has made a number of doubtful choices.
I do not know why he chose to dress the T-Rex but neglected to give it a brush, or why the character of the painting looks like a kind of children’s detective from the 1970s. And although Gemini clearly knows one thing or two on the dinosaurs, he hurt the relative sizes of the T-Rex and the Velociraptor. I was also disappointed that instead of “pterodactyl howling”, I obtained a static image of pterodactyls and the sound of the song of birds in the background.
The synchronization of the dialogue is especially good, even if I hoped more emotional of the Velociraptor.
Overall, it took me a few minutes to write these guests and 3 to 5 minutes for Veo 3 to generate each video. I believe that if I spent more time painting a detailed image, even writing a whole news, I could get an even better result.
I would let you know with certainty, but I have just conducted my brief dry test. If you plan to attend a few VEO 3 videos, here are my main tips:
- Write your prompt outside Gemini
- Choose your subjects carefully
- Spell every detail, from the appearance of the characters to the scene
- Details each action, or veo 3 will invent something or will have a character who does nothing
- Speate the dialogue so that it is clear.
- Describe the emotion behind the delivery of the dialogue
- Include details on background noise
- Include sound effect descriptions if you want specific sounds
- Each video measures 8 seconds maximum. Plan
- Try to create several videos that continue a scenario, but keep the descriptions coherent
Good luck with your VEO 3 tests. Let me know how it goes in the comments below.