Change photos with voice or text with this impressive free artificial intelligence from Google

THE Google bet for artificial intelligence It's huge. Not in vain technology transformer, which is the essence of the generative AI, left its laboratories. We are currently seeing a very strong integration in Android because Sameer Samat, president of the Android ecosystem in Google, said in an interview with The Spanish – Omicrono.

We spent a little time to use chatbots such as Chatppt in office applications to which Gemini works as the successor of the Google assistant Or that Samsung positions Galaxy Ai as the largest differential element of its new smartphones. And successfully.

The revolution is not only rapid, but constant. And companies like Google continue to innovate, although The most striking functions are previously tested in controlled development environments And then bequeathed to applications and mobiles. A good example is Google AI Studio, a platform designed to start compiling with Gemini, the new generation of multimodal generative models of the alphabet subsidiary.

Edition with voice

In this web platform, the company test different models of AI that specialize in different things. One of the latter who has made available to users and developers is Gemini 2.0 Flash (generation of images) experimentalCentered on the editing and generation of images. This model is able to change things from an image that we simply download by writing or dictating what we want.

Both in computers and mobiles We can use the vocal dictation to create complex requests. As you can see, with simple sentences, it is already possible to change elements of an image, as if we control Photoshop remotely and the program has done what we ask.

In this first image, for example, we changed the color of the jersey of one of the people, and we did it correctly and extremely fast, because the execution of the order led Between 5 and 10 seconds. Of course, it doesn't always work well.

For example, the second order told him to put sunglasses for the two people. He understood what we wanted, but not that we wanted it in the two people, so he did not put them to man. If we ask him to correct and put them on the woman, strangely deletes them from the other person. We have also changed the background.

It shows that At all times, he works with the original image, Since he has no failure to remove the people's glasses again and show his face. Of course, we can continue to make iterations on changes by adding more and more.

The fact is that, If we detail the petition much moreexpressly saying that we want red glasses in the case and blue in the case, The system includes and the fact. It is clear that it is always an experimental system, but the operation is impressive.

In addition to being able to change the colors or certain elements of the images, others can be added With a very strong consistency. In the second example, we took a photo of the Cathedral of Mallorca and we asked him to add a red ship.

The system understood what we wanted and created it, but also know that I also had to create reflection in waterBecause this is how it would have remained in a real photograph. This gives us an idea of ​​the quantity of autonomous decisions that these models can take with a simple order.

The Google AI Studio interface is not complicatedBut it is not as simple as that of Gemini. It shows that it is designed for a more advanced type of user, in particular developers, and gives tools that allow you to modify the results. For example, we can extract the programming code, modify the model or output format.

It is also possible to see the number of tokens that a specific command has used or to modify the security parameters so that the system does not launch violent or unpleasant images. In the upper part, we can even activate the mode of comparison, which allows us to execute an order in two models at the same time to see what differences they show.

Creation of stories

This AI model, Gemini 2.0 Flash (image generation), He has other possibilities in addition to modifying the images. In the interface, they show us that birthday cards can be created, for example, or much more interesting, as Create a story in some vignettes simply with an order.

What the system does is create its own commands for each sticker, ending what it has to put according to the original command, which does not have to be particularly concrete or extended. The result is a story with images that could be used to tell a child at night, for example.

One of the examples we have created is a story on a ghost that wants to escape hell. To remove a small iron, the illustrations were requested in a pixel art style, as if it was almost the story of a game of the 80s. maintains consistency throughout the creations of the different vignettes.

Although all these tools are fully usable and free (we only need a Google account to use Google AI Studio), They will become massive when they are integrated into the company's systemsIn applications such as Google Drive or on Android via Gemini.