Creating Your AI Persona with VASA-1 and Spheria

25 Apr 2024

Microsoft just released a preview of their new AI model called VASA-1, which animates any photo into a “talking face avatar.” It takes a few seconds for our eyes and brain to adjust to this reality, but then the immediate question arises: what are these new avatars supposed to say? Will this new AI milestone be wasted on lip dubs and creating even more fictional Manga-Anime?

What if you could create an AI version of yourself by combining VASA-1 with your own AI to have an official AI that represents you online and is legitimate enough to speak for you?

The future we all knew was coming.

We’re used to the idea of talking images. Remember the talking paintings in Harry Potter?

It was inevitable that large tech companies like Microsoft and Google would develop models to animate pictures and make them “talk.” They have been working on Image Processing, Facial Recognition, Image reconstruction and Classifiers for years. They have the talent and the research to master the processing of visual and audio data… So you can expect more announcements on these two fronts very soon.

With their significant investment in AI research and in the Metaverse (let’s not forget the hundreds of millions poured into it), animating images was destined to happen. It’s no surprise that these tech giants are now leading the charge in transforming static visuals into dynamic, talking heads.

Create the perfect AI version of you with your brain and your talking image!

Create the ai version of you, trained on your data and life experience — screenshot from

Microsoft will not make VASA-1 public because of the ethical risks of creating an infinite amount of deep fakes and misrepresenting people too easily. So, the “talking image” feature will not be available in the near future, but you can absolutely create your own AI version.

Using Spheria to create the AI version of yourself, you can grow your own AI double that learns directly from you, giving you real answers and speaking in your name.

An example of a real person’s AI double, that can reply with images and text — created on

The goal is to one day combine your own AI double with a realistic and animated 3D shell — powered by VASA-1 or another image model. This would really create the futuristic society we’ve seen many times in sci-fi movies.

You would then have a real official AI version of yourself to represent you online and give your real answers based on your personal experience and opinions.

The problem with Avatars and VASA-1

MetaHuman engine developed by Unreal to create 3D avatars

Many companies like NVIDIA, Unreal, and Microsoft focus on creating a 3D rendering of real people or 3D avatars using various technologies like Unreal Engine or VASA-1.

But the result is always the same: an impressive empty shell. What’s the real-life use of a Mona Lisa that can rap? Why would Napoleon suddenly say things that the real Napoleon would never say?

3D avatars and “talking images” are doomed to become empty shells. And these companies, including Microsoft, have no interest in creating convincing fictional characters — it’s too much effort and extremely difficult.

3D avatars gain all their value when they have a distinct personality and when they can deliver meaningful information that is not generic — i.e., information not already available on chatGPT or Wikipedia.

As humans, we immediately differentiate how we choose to interact with generic AI entities or with real people. The line is set by having a personality and an individuality.

So, while VASA-1 evolves, and while 3D avatars become mainstream, you can learn how to easily create the AI version of yourself that captures your unique thoughts and individuality… to connect it in the future with the 3D avatar of your choice.