The new parents curious about what they should expect from their newborn should probably avoid Tiktok and AI. There is now an incalculable number of videos representing newborns working in factories and burger, walking through the dusty plains of March and commenting on the desolate state in the world in which they have just entered.
Since Gemini unveiled his VIDE 3 video model, which allows you to generate an audio and a synchronized video, people have generated very convincing but impossible 8 -second video clips. The number has only increased since Google has opened VEO 3 access to users of GemiMi Flash 2.5 Pro.
After seeing more than some of these videos in my flow, I decided to try it. I have now spent the past few weeks to create VEO 3 clips of newborns to do all kinds of things and often comment their situation.
Here is what I learned, as well as some tips to help you create similar clips or really improve your generation of gemini veo 3 videos.
Planning
Unless you pay an account, you get a maximum of three VEO clips 3 seconds a day, so you want to use this number judiciously and, if possible, pre-plane the three clips.
Start with a concept or an idea
What do you want your newborn to do and say? Try to think about the frame and action on the three clips so that together they understand a mini story.
When I started to generate my baby clips, I did not consider this and I created a series of clips representing an information journal interviewing a newborn in the hospital, but with different levels of success. Finally, however, I chose a theme, like a baby astronaut on Mars, or a newborn baby working in a robot factory.
Once I had a mini story in mind, my results have improved exponentially.
Imagine a style
One of my favorite clip sets is the one in which I imagined a baby in the role of Dr. Frankenstein. I specifically asked Veo to represent the baby in a “style laboratory of the 1930s” and I added this critical line, “the film is in black and white with a contrast and strong shadows.”
Do not be afraid to describe the details of the scene that could improve credibility. In my baby’s astronaut video, I described “a newborn baby using a selfie stick for the film itself” and I noted that the earth was visible in the sky. Veo 3 did a great job including these details.
Be precise
I found that less to describe the main character as a newborn baby, I ended up with a toddler, and sometimes the one who looked like a pixar character.
Also make sure to describe the action. What is the baby doing? A sequence of events is the best way to make sure that Veo 3 builds the scene you imagined.
Keep the dialogue tight
You only have 8 seconds of video, so be sure to timed your dialogue to make sure that it is not 15 seconds of jokes. Whatever Veo 3 cannot adapt, he will cut.
Keep the characters at least
One of the reasons why these videos are so convincing is that they have newborns speaking like adults. The achievement of these results has sometimes been difficult.
I had a hard time making sure that it is the baby who speaks and not another character that I included in the prompt. I found that it is better to describe at most two characters and to make sure that your prompt makes it clearly clear which should speak the lines, which should always be between the quotes.
Even with all this meticulous planning, I have always had trouble making sure that only the newborn spoke. In my series Frankenstein, the last clip presents a speaking Frankenstein monster, even if I specifically attribute this dialogue to the baby.
Tell a story
If you create the three clips of the same session, one after the other, Gemini Veo 3 will remember the style and even the appearance of the characters included, which means that you do not have to repeat all these details in the following invites.
That said, be sure to describe the baby and other characters with the same words; Otherwise, AI can change things. If possible, design the three guests so that they can advance your 24 -second story.
Call that “realistic”
To ensure photo realism, use the term “photo-realistic” or “realistic” in your prompt. Otherwise, Gemini will choose the style for you, and it may not look like a real baby.
Download the video
Gemini does not record your videos generated indefinitely. They generally left in one day, although the prompts remain. If you like what the prompt has created, be sure to download the video.
These tips could undoubtedly be applied to any generative video style, but if you want to join the craze for the same AI Baby, this should help you start.
