AI-generated dinosaur image

(Image credit: Future)

How Veo 3 helped me create these mind-blowing artificial intelligence videos and what I gleaned

The narrative is everything.

Though it only took a few seconds—two brief, 8-second videos—I was hooked when my first Veo 3produced video came to life. The integration of images, sound, and speech sourced from a basic text prompt felt like enchantment.

Google AI Pro, a $19.99/month plan with access to Gemini 2.5 Pro, and — more excitingly — a trial copy of Veo 3, Google’s newest generative video model, is what I am currently using.

Let me say as well: Veo 3 is quite unlike anything I have ever employed. It’s not only a video maker. It even provides synced audio, background noise, and character conversation. You are creating miniscenes that come alive rather than just producing animated stills.

The trial revealed neither how many videos I would get to produce. Two turned out to be all there were. I would have to upgrade to Google AI Ultra at $249.99 monthly (or $124.99 monthly throughout a threemonth promotional) to open limitless access. Veo 3 is only accessible in the United States for the time being.

Veo 3 has overtaken my TikTok feed since Google I/O 2025 unveiling. Some of the videos I have seen look like trailers for major films; others are strange, funny, and entirely absurd.

I wasn’t sure if I should get really surreal, cinematic, or naturalistic. I ultimately arrived somewhere between a little drama, a little scifi, and a dash of fantasy. My initial try, though? Let’s just say… I found out some things.

 

Take One: A Martian Showdown Gone Somewhat Bad

First Prompt of mine? Though creatively conceived, its execution was ungainly. I wrote it within Gemini’s text window and unfortunately pressed Enter too soon, sending the incomplete scene. Darn.

Bill and Jessica live in a wooden cabin on Mars’ surface.
Bill goes outside to see Jessica battling a Martian only with a stuffed animal.
Bill yells: “What are you doing?”
> Jessica answers: ‘That Martian wants our land and he can’t have it.’”

Because I didn’t complete or refine the prompt, the outcome was slightly messy. For one, Jess’s last line was severed. Veo 3 also handled the visuals well—the flaming red Martian dust, the timber hut, the eerie alien design—but some things felt… off.

Bill’s voice is heard, but his lips never budge. Jessica stares directly into the camera as she speaks, which breaks immersion unless you’re going for that effect. And some sound effects, like a peculiar clunking noise from Bill, did not seem sensible depending on the prompt.

I was nonetheless amazed by the level of detail. The soft crunch of Martian dirt, the creaking cabin door, and Jessica’s “thud” throwing of the stuffed animal are all part of this. Those little touches gave it a feeling of life, even if it was imperfect.

Take Two: An artist of dinosaurs with existential fear

Eager to improve, I created my second prompt outside of Gemini first—edited and reworked before pasting it into Veo 3. This time, I leaned into something more surreal and self- aware:

Scene: A bird-song-filled sundappled forest with faraway dinosaur screams.
> A TRex is creating a large canvas portraying a man just before an asteroid strikes.
He softly hums, “Pink pony club, I’ll keep on dancing in…”
A Velociraptor arrives and questions, “Why are you painting this?”
The TRex stops and declares, “AI made me do it.”
The Velociraptor stares, surprised: “What?!”

Veo 3 got the scene mostly right this time, but not without weirdness. The TRex was carrying garments but for some reason lacked paintbrush. The man on the canvas seemed less like a doomed character and more like a 1970s detective from a children’s television program.

The noises were also erratic. I asked for dinosaur noises, but I got bird chirps instead. The forest was stunning visually, although some of its scale was incorrect. One shot saw the Velociraptor seem bigger than the TRex. Although the conversation was synchronized, it lacked emotional expression, so the scene came off a bit dull.

 

Better Veo 3 Video Lessons Learnt and Advice

Each of the films was made in less than five minutes; writing the challenges took only somewhat longer. Though the trial limits me, I’m ready to attempt once more should I ever have more ages.

What I discovered—and what you need to know if you want the best results out of Veo 3:

First compose your prompt in a different text editor. It is extremely simple to accidentally press “Enter” too early.
Define precisely. If necessary, detail backgrounds, activities, characters, and camera angles.
Define speaker sentiments and conversation. Say so if you want a character to shout, whisp, or cry.
Include ambient sounds and sound effects. I am rather astonished by how Veo translates these into the finished movie.
Maintain your clip’s 8 seconds or less. Plan appropriately; the limit counts.
Create a more complex narrative using many clips. Veo will be kept in continuity by regular prompts.

 

Final Remarks

Though 3 isn’t ideal—yet—it’s a remarkable stride in video produced by artificial intelligence. You can write something hilarious, dramatic, ghostly, or poetic with just a few lines of text. Though it varies depending on your prompt, when it does, it truly does.

If you’re worn out on infinite scrolls or looking for a new creative outlet, try Veo 3—even if it’s just to see a dinosaur painting the end of the earth.

 

Planning to try Veo 3? Share your best (or strangest) prompt ideas in the comments!

Leave a Reply

Your email address will not be published. Required fields are marked *