It was just a glimpse, two 8 seconds of VEO 3 videos, but as many things that change life, I would never forget to produce my first time -to -create audio and video, with a great easily developed indicator.
I’m currently running Google AI Pro, which is a month’s account $ 19.99, which gives you access to the Gemini 2.5 Pro model, and even more importantly, the VEO 3 video is a limited test of video generation.
VEO 3 Generative Video is the tiping point level of creation that makes it possible for the first time, making videos with dialogue, background noise and sound effects, which are all synchronized in action.
Although I understood that access to my VEO 3 could be limited, I was not sure how many videos I could produce with a new model. The answer, it seems, are exactly two. If I want unlimited access, I can be converted to Google AI Ultra to drink water in a month (there is a three -month contract for 4 124.99 a month). And the VEO 3 is currently only we are.
Since VEO 3 has launched Google I/O 2025, my tricktok feed is full of these incredible and often realistic AI clips. Some look like informal or advertising, it is impossible for others to be just as impossible, as a woman who interviewed a smiling man who was clearly on fire.
I was bursting between realism, hyper realism, and creating something terrific. Finally, I created a gesture in the Gemini 2.5 Pro window that supports the video creation that was a mixture of science -fi, drama and seriousness.
Immediately writing inside the window, though, proved to be a mistake because I hit the mistakenly before eliminating my idea completely, and suddenly V3 was busy preparing my video.
It was my first gesture:
“Bill and Jessica live in a log cabin built on the surface of Mars. The bill emerged from the cabin so that Jessica is fighting Martin using nothing but a full animal.
Bill screams in Jessica: What are you doing?
Jessica: This kicker Martin wants our land and cannot have it. “
As you can see, it does not have much detail, and as easy is to produce the video in VEO 3 (and audio -free VEO 2), you will get a better result by adding more detail and dialogue. VEO 3 will have nothing to do with the characters you did not script. In this case, because I soon targeted the return, Jessica’s dialogue was disconnected and I did not get to polish my gesture.
Nevertheless, V3 took very little details and made an amazing piece of video in about 5 minutes. Take a look (take the sound for full effect).
My first VEO 3 video: a cabin pic.twitter.com/at63w2lqdm on MarsMay 28, 2025
It is far from perfect. The bill does not actually speak the line, though we hear it from off -camera. Jessica’s scream (or is it from Martin?) Also comes from the camera.
There is an unfortunate sound effect that is probably coming from the bill, and that I did not script. Also, I do not know why Jessica speaks her lines directly on the camera.
Once again, I assume that if I had instructed who should talk to, V3 may have made a different choice.
Nevertheless, there are many and more subtle things that are impressive. VEO 3 setting is fine. See the red cloud of Mars Day Light. Martin is terrible. I am more impressed, however, the sound of the cabin door, the sound of the feet on the mud, and the sound of a filled animal like I’m hitting Martin’s chest.
Take 2
For my second gesture, I wrote and edited it outside Gemini. I tried my best to set the scene, explain the characters, and to describe the dialogue and any sound effects. Indicated. It is:
This scene is a lush forest in which the sunlight runs through the overhead. We listen to the highlights of pterodactyls in the background and the sound of the leaf in light air.
A tirenosis is carefully painting a large canvas that shows the colorful image of a man, which is about to be destroyed by a temptation.
Tirenosis sing quietly, “Pink Pony Club, I will continue to dance on it …”
A Velverspatter rotates and asks, “Why are you painting it?”
Tirennosus: “Ai forced me to do it.”
The Velvisrapeter retreated in a frightening way and said, “What? !!”
As you can see, I, partially, was impressed with some self -related VEO 3 videos that I was watching on Techtok, where the characters break the fourth wall and mention that they are Ais in a video. Although the work is mostly eliminated in my detail, the VO has made several objectionable choices.
I do not know why he chose Tea Rex to wear clothes but was ignored by giving him a paint brush, or why the role in painting looks like a spy of children of the 1970s. And while Gemini clearly knows one or two things that look like a dinosaur, he found the relatively size size of T -Rex and Velverspatter. I was also disappointed that instead of “Petrodcticles’ clashes”, I found a stable picture of the sound of Petrodctilis and Birds in the background.
Dialogue synchronization is mostly good, though I was hoping to have more emotions from the Velvisrapor.
Overall, writing these indicators to prepare each video took me a few minutes and 3 to 5 minutes for View 3. I am sure that if I just spent more time painting a detailed picture, even writing a whole short story, I could get even better results.
I’ll definitely tell you, but I just dried up my short test. If you plan to attend a couple of VEO 3 videos, my main points are:
- Write your gesture outside of the gym
- Choose your titles carefully
- Spell every detail from the form of roles to the scene
- Each action description, or VEO 3 will make something or no character is doing anything
- Spell the dialogue so it is clear.
- Explain the feelings behind the dialogue delivery
- Add details related to the background noise
- If you wish specific sounds, add a description of the sound effect
- Each video is more than 8 seconds. Plan accordingly
- Try to create multiple videos that continue the story, but keep the explanation permanent
Good luck with your VEO 3 test drives. Tell me how it goes into the comments below.