How to edit image in Google AI Studio

Image generation and editing with AI is already a reality in people's daily lives and, eventually, the main models have made image-focused tools available in their chatbots. For those who like to use the platform Google, Gemini, it is also possible to test AI image manipulation within the platform AI Studio, where users can explore many of the capabilities being developed by the company. In this article, learn how to access and which commands to use to manipulate images in Google AI Studio.

What is Google AI Studio?

O Google AI Studio is an integrated development environment (IDE) provided by Google, which gives developers and enthusiasts the opportunity to experiment with the company's generative AI models, providing a fast and free way to generate AI-based insights using the company's powerful family of models. Gemini do Google.

The tool positions itself as a more user-friendly option for initial exploration and prototyping. Compared to more robust, enterprise-oriented platforms such as Vertex A.I, also do Google, AI Studio is designed for those who are just starting to test different AI models, features and techniques, whether for professional reasons or even for hobby, without the need for corporate data protection.

In this way, the AI Studio ends up being the main platform for interacting and using the family of models Gemini, including those with image editing capabilities.

Once users are satisfied with their prototypes and experiments, the code can be exported (using the Gemini) to be integrated into your own applications. The API of Gemini, including, allows developers to incorporate text and image generation into their applications using a single model.

How to use Google AI Studio

Step 1 – Access the Google AI Studio and connect to your account Google.

Passo 2 – Select the option to use the Google AI Studio to Gemini.

Step 3 – Look for the model selection option in the interface AI Studio and select the model Gemini 2.0 Flash Experimental, latest update of the company's AI.

On desktops, it is usually in the right side of the screen.

On mobile phones, it's in a menu at the top.

Step 4 – Upload the image you want to edit.

The interface of the AI Studio offers the option of upload image that will be modified. You can upload files from your computer, use photos from Google Drive, or even take a photo with your device's camera.

Step 5 – Use commands to edit the image. Example: “Turn this car into a convertible.”

What prompts to use to edit images?

As with the use of other tools for more specific use than the Google currently offers, such as Deep Research do Gemini, the effectiveness of image modification in Google AI Studio It depends a lot on the quality of the prompts provided.

To get the best results, you need to learn to structure your prompts concisely. Start by describing the desired change in a straightforward and objective way. Instead of a vague command like “change the picture,” be specific, like “remove the dog from the background.” When identifying the object or area you want to change, be as specific as possible. For example, if there are multiple people in the picture, specify “the person wearing the red shirt.”  

Use descriptive language to convey the desired outcome. Instead of “enhance the image,” try “brighten the image and increase the contrast.” Details such as color, size, position, style, and mood can make a big difference in the final result. An example would be: “Add a small, vibrant yellow vase to the left side of the table.” To replace the background, clearly describe the new desired setting, such as “Replace the plain wall with a tropical forest.”

It’s important to remember that creating effective prompts is an iterative process. If your first result isn’t exactly what you expected, try rewording the prompt or adding more details. Don’t hesitate to experiment with different wording and levels of detail. If you’re looking for realistic-looking results, consider using modifiers related to photography, such as “high-quality photo,” “studio lighting,” or “close-up photo.”

To help you better understand the importance of being specific when manipulating images, take a look at the huge difference in the final quality of the car edit shown in the tutorial above. When asked to simply change the color of the car to yellow, the tool did not maintain the same detail in the paint job and created additional distortions in the image, such as a sort of spoiler on the car:

With a more specific prompt about the vehicle's details, the final quality was superior, including bringing back details on the hood that were lost when the car was transformed into a convertible:

See below for more possibilities with the tool.

Background replacement

PROMPT INICIAL: Substitua o fundo desta imagem por [descrição do novo fundo].

Examples:

Prompt: "Substitua o fundo desta imagem por uma praia paradisíaca com areia branca e águas cristalinas ao pôr do sol."

Prompt: "Altere o fundo da imagem anexada para um cenário em uma noite estrelada"

And of course, you can also make this change with people. Use a photo from your phone or take a new photo and ask the person to do the same. Gemini change your location. See:

Object removal

PROMPT INICIAL: Remova [descrição do objeto a ser removido] desta imagem.

In some cases, with many elements on the screen, it is necessary to be more specific about what the tool should erase from the image. Examples:

Prompt: Remova o carro vermelho estacionado à esquerda desta imagem.

Prompt: Apague a pessoa vestindo uma camisa azul no canto direito da imagem.

Object addition

PROMPT INICIAL: Adicione [descrição do objeto a ser adicionado] [posição e detalhes do objeto] nesta imagem.

To add the object, there are contexts in which the tool will respond better if you detail the object and the conditions in which the object should be in the image. As in:

Prompt: Adicione um gato preto pequeno dormindo no sofá ao lado da mulher nesta imagem.

Prompt: Adicione uma vaca marrom no lado direito da foto. A vaca deve ser parecida com as outras duas. A vaca deve estar de frente, como as outras. A vaca deve estar na mesma posição e escala que as outras duas. Integre-a corretamente ao ambiente, com luz e sombras condizentes com o resto da imagem

You can also add a whole set of objects to a scene, directly indicating what type of location the AI ​​should be inspired by. To furnish a room, for example:

Prompt: Transforme este cômodo em uma cozinha

Object replacement

PROMPT INICIAL: Substitua o [descrição do objeto a ser substituído] por um [descrição do novo objeto] nesta imagem.

You can use commands that ask the AI ​​to replace objects in the scene with other models or a completely different object. Use prompts like:

Prompt: Substitua o vaso de flores na mesa por um laptop moderno nesta imagem.

Prompt: Troque o cachorro pequeno na imagem por um gato siamês.

style change

PROMPT INICIAL: Transforme esta imagem no estilo de [estilo artístico desejado].

ou

Aplique o estilo de [estilo artístico desejado] à imagem anexada.

You can change the art style of an image, suggesting inspirations from artistic movements (such as cubism, surrealism, impressionism…) and painting techniques.

Prompt: Transforme esta imagem no estilo de uma pintura, utilizando técnica de aquarela.

Image expansion

PROMPT INICIAL: Expanda esta imagem adicionando conteúdo [direção da expansão] no estilo da imagem original [adicione mais detalhes sobre o que deve ser incluído]. 

If you’ve taken a photo and want to show a little more of what’s out of the frame, you can ask the AI ​​to expand the image. Use prompts like:

Expanda esta imagem adicionando mais céu azul e algumas nuvens na parte superior no estilo da imagem original.

It is also possible to ask for the Gemini complete a specific object that was cut out of the image, as in the example below:

Prompt: Expanda a imagem para a esquerda e reconstrua a outra metade da lamparina, mantendo o estilo e as configurações de iluminação da imagem original

Color Adjustment

PROMPT INICIAL: Ajuste as cores desta imagem para torná-la [descrição do ajuste desejado].

You can directly tell the AI ​​how you want the photo to be adjusted, asking it to make the image more or less saturated, adjusting color temperature, brightness, filters, and more.

Still on the subject of colors, the tool can be used to colorize drawings or enhance them. See:

Prompt: Colorize este desenho com tons de vermelho

And of course, as an iterative tool, use more commands to improve the paint characteristics. Such as:

Prompt: Faça um sombreamento suave no desenho, considere que o ponto de luz está no canto superior direito 
Prompt: Deixe o desenho menos brilhante, simulando uma pintura a mão

In addition to decolorized images, you can also “play” a little with colorizing old black and white photos, with commands like “colorize this photo”. In some cases, the Gemini will not apply colors properly, so it is necessary to elaborate the command a little better. How:

Prompt: Colorize esta foto antiga com tons historicamente precisos, baseando-se em referências fotográficas da época. Preserve a textura original e evite cores artificiais ou saturadas. Aplique tons realistas.

O Google AI Studio, powered by artificial intelligence Gemini, opens up a range of possibilities for image modification. The ability to use natural language to perform creative transformations on images makes the process intuitive and accessible, even for those without prior experience with complex editing software.

However, it is important to emphasize that the tool is intended as a testing tool, so do not rely on it for professional editing. As an experimental version, it is still common for you to need to interact with the platform a few times to get a result close to what you want, and even then, it is not uncommon to see distortions, especially on the faces of people or animals. Using commands in English, on some occasions, the result is better.

As well as other models available on AI Studio, it is expected that the ability to modify images will be made available in the Gemini in the future. Tell us in the comments below how you intend to use the tool.

Learn more about Google Gemini:

How to Use Gemini's Deep Research to Do Research and Work for You (for Free)

Text revised by Alexandre Marques in 20 / 03 / 2025.

Sources: Google, TechCrunch.