I test AI tools for a living and have had my fair share of “wow” moments. But nothing quite prepared me for what Google’s new Veo 2 can do. Let me just say this: if you’ve ever dreamed of directing a movie scene without touching a camera, this tool comes pretty close.
With so many free tools available from Gemini, it’s understandable why Veo 2 is part of Google’s latest rollout for Gemini Advanced users. It’s worth the subscription. This tool lets you generate 8-second, 720p cinematic videos using a single prompt.
Think ChatGPT, but instead of spitting out action figures, it gives you mini films.
Released April 15, the feature quietly appeared as a dropdown option in Gemini, and naturally, I had to try it.
1. Animals doing silly things
I started off with something silly and could not believe how well Veo 2 nailed it. I’m talking sunglasses sparkle, palm trees swaying, and that dog looking like he’s about to order a piña colada. I’m having so much fun with these videos and I’m only just getting started!
Prompt: A pampered French Bulldog wearing sunglasses, lounging on a plush daybed by a pool at a luxury resort.
2. Science fiction
Prompt: Nico Carmichael is an 11-year-old gaming prodigy who would rather be dominating virtual battlefields than navigating the unpredictable world of real life (or "IRL," as he calls it). Nico’s world is thrown into chaos when he accidentally brings his video game controller to school—only to discover that it holds real-world superpowers.
For my next prompt, I went sci-fi. I’ve always wanted to turn my young adult science fiction book into a movie, and this was the best opportunity to try it.
I simply described the synopsis of my book and within seconds, Veo 2 returned a realistic, otherworldly clip that looked like it belonged in a big-budget sci-fi series. Okay, maybe not that intense, but I can dream, right? The movements were smooth as butter. The atmosphere? Chef’s kiss.
3. People and children
Prompt: Children playing on a playground. Please include typical equipment such as swings and sandbox.
I was curious how Gemini would do with generating a video that included people. From my experience, most AI tools still struggle to accurately depict humans and many won't generate images of children at all.
The AI did a fairly decent job but it wasn’t perfect. The little girl on the swing suddenly shifted to a boy in a sandbox. But the children were surprisingly realistic.
4. Transportation and motion
Prompt: Create a scene with traffic that includes multiple cars and a train passing next to the highway.
My goal with this prompt was to determine how well the AI could capture speed and motion.
This video looks like it was a clip from the news or a television show. It’s wild how well Gemini captured the realistic movement and the direction of the traffic in conjunction with the train.
5. Pushing the limits of imagination
Prompt: An ocean scene with mermaids chatting with each other and living their best life.
For the last prompt I wanted to test Gemini’s ability to make unreality come to life. What better way than to prompt it to create a video of mermaids in the ocean?
It was interesting how it decided to create a scene that was more Bubble Guppies than National Geographic, but it was well done, nonetheless. It did seem to go overboard (no pun intended) with the number of mermaids.
To be fair, my prompt was vague, which I did on purpose because I like to see where the AI takes the story. But I know that tweaking the prompt would curate a much better scene.
Whisk Animate
Google’s other new AI video image generator, Whisk Animate, is tucked inside Google Labs and is equally impressive. I tried it by uploading a static image of myself and adding what I wanted the image to do.
To be completely honest, I wasn’t as impressed with this feature as much as Veo 2. But I plan to continue playing around a little more because I have seen what it can do.
1. "Me" walking through the city
Prompt: Using the image of me, create a scene where I am window shopping on a city street.
The image it created from my picture did not look like me, not even a little bit. However, the video of “me” from behind, walking down a street and window shopping was cute and nuanced.
2. A cat walking down the street
Prompt: Using this image of my cat, create a scene where she is walking down the same street that I was in the last video.
Next, I tried uploading an image of my cat. Gemini captured here perfectly. I clicked “Animate” and Google made the image come to life. While it looked amazingly like my cat, it was fun to see her in a situation that wasn't snoozing on my desk.
How to access
To access these exciting features including Veo 2 and Whisk, you need a Google One AI Premium subscription. You can give it a try with a free trial, but to continue, it's $20/month.
With so many free features available within the Google AI suite, I have to say, this one is definitely a game-changer that might be worth the subscription.
Final thoughts
While the videos are short (just eight seconds) and have no sound with a resolution that maxes out at 720p, the results are striking — especially considering this is all powered by AI and done with a simple text prompt.
I kept my prompts extremely simple, but you can get as detailed as you’d like to create incredibly realistic videos that fit your style.
I was impressed that Gemini did not glitch even once during the generation. I’ve honestly never had an AI tool not glitch when making several videos in a row, so that was huge. The interface is simple, the turnaround is fast, and the creative possibilities are endless.
Bottom line? Veo 2 is the most fun I’ve had with an AI tool in a while. The possibilities are endless and I can't wait to keep playing with this tool.
More from Tom's Guide
- OpenAI just launched its smartest AI yet that can think with images — here's how to try it
- I've subscribed to every AI tool — only one was worth the money
- I let Google AI take over my desktop — it found files I thought were gone forever and other mind-blowing tricks
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.
