Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Google upgrades Chatbot Gemini with a new image form for Amnesty International that gives users more accurate in editing photo, a step aimed at catching up with popular Openai popular Photo tools Draw users from ChatGPT.
The update, called Gemini 2.5 Flash Image, starts from Tuesday to all users in the Gemini app, as well as for developers via Gemini API, Google Ai Studio and Vertex Ai.
The new AI image is designed from Gemini to make more accurate adjustments to the images – based on natural language requests from users – while maintaining the consistency of faces, animals and other details, which fights the most competing tools with them. For example, ask Chatgpt or Xai’s GROK to change the color of a person’s shirt in a picture, and the result may include a distorted or variable background.
The new Google tool has already paid attention. In recent weeks, social media users vibrant AI’s AI’s photo editor is impressive on the collective evaluation platform, LMARNA. The model appeared for users anonymously under the name “Nano Banana”.
Google says it’s behind the model (if not Intuitive actually From all banana hints), which are in fact the original image of the image within the pioneer Gemini 2.5 flash Artificial intelligence model. Google says the image model is the latest on LMARNA and other standards.
“We are really paying the visual quality forward, as well as the model’s ability to follow the instructions,” said Nicole Brechtova, a leading company in the visual generation models in Google DeepMind.
“This update does a much better work to make modifications smoothly, and the outputs of the models are useable in everything you want to use for it.”
Artificial Intelligence Photo Models have become a decisive ground for large technology. When Openai launched the original GPT-4O photo generator in March, he led Chatgpt’s Use Through the ceiling thanks to the madness of artificial intelligence Studio GHibli The memes that, according to the CEO of Openaii Sam Al -Tamman, left the company’s graphics processing units.Melting“
To keep up with Openai and Google, Meta announced last week that it would work license Amnesty International Photo Models from Midjourney. Meanwhile, the German -backed UNANCON Black Forest Laborators Continue to control the standards with Flux AI models.
AI’s impressive AI’s photo editor may help Gueini Google bridge his user gap using Openai. ChatGPT now records more than 700 million Weekly users. On Google’s profit call in July, the CEO of Tech Giant Sundar Pichai revealed that Gemini had he 450 million monthly Users – means that weekly users are lower.
Brichtova says that Google designed the photo model specially with consumer use cases in the mind, such as helping users to imagine their home projects and garden. The model also contains a better “global knowledge” and can combine multiple references in one claim; For example, combine a sofa image, a living room image, and a color palette in one coherent width.
Although the new AI’s AI’s photo generator makes it easier for users to create and edit realistic images, the company has guarantees that challenge what users can create. Google has struggled with artificial intelligence images guarantees in the past. At one point, the company I apologize To generate inaccurate images historically for people, and Rolled Fully artificial intelligence image.
Now, Google feels a better balance.
“We want to give users creative control so that they can move from the models they want,” said Brechtova. “But it is not the same as anything.”
The IQ Department of Obstetrics prevent Google from Google users from generating “intimate intimate images”. These types of guarantees do not seem to be present for GROK, which allowed users to create AI Frank photos Little celebrities, such as Taylor Swift.
To process the rise of Deepfake images, which can make it difficult for users to distinguish what is real online, Brichtova says that Google applies visible watermarks to images created from artificial intelligence, as well as identifiers in descriptive data. However, someone may search for a picture on social media about such knowledge.