Google’s new AI version of everything is wild

Last year I My son’s flesh is deep to make it look like her cute deer is on vacation.

It was trying to see if I could recreate the events shown in the Gemini Google travel ads, and I had never shown the videos of Buddy the deer in his activities to my four-year-old son. But it was an exercise that made me think deeply about the difference between harmless fun and AI spawning and completely falling apart. Maybe the Venn diagram is the perfect circle! Maybe not. But what I do know for sure is that the tools for creating great looking videos are amazing, requiring very little effort and skill. And this continues until the Gemini period of the Omni.

Omni is a new family of production models which he says will one day transform any form – image, video, text – into anything else. But for starters, I’m just making a video. Omni Flash is the first of these models released by Google, which is now available in the company’s AI video production and editing platform, Flow. You can still use the older version, the Veo, if you want, but the Omni improves on the Veo in a number of ways.

With Omni, you can upload a video and use it along with audio as the starting point for your AI-powered creation. Google also says that Omni integrates global information into video production and can do a better job of keeping characters consistent throughout the video. There was only one way to find out if the claims were true: I brought AI Buddy back to carry his little AI-designed bags on another trip.

The results are so mixed that it’s confusing. The rest was great – more stable and faster than when I tested the Veo five months ago. But even the best videos that Omni cooked up for me still have some AI scares, like when Buddy suddenly transforms while flying.

For another video, I gave Omni the creative rights. “Make a scene of Buddy packing up for a vacation and boarding a cruise ship to a tropical vacation. The atmosphere is cute and fun. Buddy is packing something funny in his suitcase that will be played later in the clip.” They had Buddy carrying a jar of honey; later in the clip he reaches for a bottle of sunscreen. “Ah,” the man says as he pours honey on his hoofs.

Honestly, not a bit bad. Except that the honey bottle changes every time in the movie, from a jar, to a squirt bottle filled with water, and back to a squeeze bottle filled with honey. And I can’t even begin to explain how this metaphor came to be with the final frame of the movie – almost like it just blocked a group of things that had just been created.

You can use voice-based suggestions to make changes to your videos, and I’ll give Google credit: This works better with the Omni than it did when I tried the Veo 3. But the results were good. bad and Veo – so bad that I found it easier to just start a new video every time I wanted to change. Omni will take your changes, but the results are not always there.

I inspired Buddy’s facial expressions in his holiday videos, and the results just looked weird. It also gives Buddy horns from time to time, which he doesn’t have. Buddy is childThank you very much. When I got it to remove the horns that were visible in one image, it obliged – and added all the other moons.

The truth is, none of this is free. Video production costs money, from 15 to 40 depending on the length of the event and the “ingredients” you start with. One change costs $40. I have the $20-a-month AI Pro plan that comes with 1,000 credits per month. After 20 parts made and a little change on others, I’m down to 145. If you have a specific idea about the video you want the Omni to make, you can look at the cuts back and forth with the model to find a video that is close to your vision.

One of Omni’s strengths is adding AI-generated content to real-time videos, so I gave Buddy a break and put it on myself. Starting with a selfie video with a neutral tone, I had Omni make videos of me eating a bowl of spaghetti, sitting in an airplane seat, standing in front of the Eiffel Tower and biting into a baguette. And I can honestly say that I was not prepared for what I saw.

There is an AI telling me in my deepfake videos. The sound of a fork hitting a plate of pasta is a bit contrived. There is a woman in the background of the video who appears twice. But aside from those little quirks and the vague feeling about them, they’re satisfying as hell.

I showed my husband the pasta clip; he knew I was testing the AI ​​video tool but I didn’t tell him what the AI ​​was made of. Not knowing what the AI ​​was made of, he bought that I was sitting in front of the camera and eating pasta, and said that his only clue was that the dish looked unfamiliar. Eating the pasta itself looked real enough to justify it my husband. The man who looked at me in real life every day for the past ten years.

My other obsessions are of the “good enough to fool people on social media” variety. Several of the Eiffel Tower animations look a bit cartoony, but one of them is so convincing that you’ll have to watch it a few times to make sure it’s AI. Me Note that it’s not me when the AI ​​turns its head to reveal its hair pulled back into a ponytail. But I don’t know that anyone can tell the difference, and that makes me feel weird.

I’m a little tired of it all, to be honest. I was very surprised when I tried Veo 3 in reality it could give birth. I have been amazed at how easy it has become to create fake people in fake photos over and over again over the last few years. I also have to wonder about the Omni, and I think I will, but the edge is gone.

It wasn’t as easy to create an AI-generated video masterpiece as Google would have you believe. But the Omni outperforms the Veo in notable ways. If you have a Google account and a credit card, then you can take a video from home and make it look like you’re jetting off to Maui with little effort. I don’t think we are “mountains of singularity” exactly, but we are definitely in the middle of an uncanny valley.

All images and videos in this article were created by Google Gemini.