Google launches Imagen 3 AI Image Generator on Gemini

In a significant development for artificial intelligence enthusiasts and creators alike, Google has officially made its powerful AI image generation tool, Imagen 3, available on its Gemini chatbot platform. This upgrade means that users can now generate images through the AI assistant, paralleling the functionalities observed with DALL-E in ChatGPT. However, potential users should note a key limitation: the generator currently does not support the creation of images featuring people, a feature reserved only for those accessing Gemini Advanced, which offers a one-month free trial followed by a recurring fee of $19.99 per month.

The technology behind Imagen 3 has been highlighted as one of Google’s most effective tools in the AI image generation landscape. PetaPixel reported on its performance in August, noting its competitive edge when compared to competing models like DALL-E and Midjourney. Notably, unlike these rivals, Imagen 3 remains free to use, making it more accessible to a wider audience.

Previously limited to Google’s AI Test Kitchen service, the transition of Imagen 3 to the Gemini platform significantly broadens its user base. Google has confidently asserted that Imagen 3 represents its highest-quality text-to-image model yet. It boasts capabilities of generating images with enhanced detail, improved lighting, and a reduction in distracting artifacts when juxtaposed with earlier models. The company emphasized that tremendous strides had been made in improving Imagen 3’s comprehension of user prompts, allowing for a diverse range of visual styles and the ability to capture intricate details from extended text inputs.

While details on the dataset utilized to train Imagen 3 remain sparse, Google has indicated that it was constructed using a “large dataset comprising images, text, and associated annotations.” Industry insiders speculate that this likely includes numerous copyrighted images, which raises questions about the use of such content in generating AI images.

In addition to generating images, Google also offers an editing feature using the increasingly popular inpainting technique. This allows users to specify parts of an image and dictate the desired changes, providing a versatile tool for creative expression.

However, Google has introduced specific restrictions on the usage of Imagen 3, a contrast to some of the more unregulated options available in the AI landscape, such as Elon Musk’s Grok image generator. For instance, attempts to generate politically charged or sensitive content, like images of political figures engaging amicably, have proven unsuccessful with Imagen 3. An example highlighted by PetaPixel showcases efforts to generate a scene involving Kamala Harris and Donald Trump, which was blocked by the software. Yet, clever users have found workarounds; by utilizing descriptive phrases that reference historical photography styles, they can still achieve desired effects similar to the works of renowned photographers such as Ansel Adams.

Earlier in the year, Google faced significant backlash regarding the AI image generator on Gemini, which was criticized for allegedly overcorrecting for biases to the extent of “erasing white individuals” from generated content. This controversy resulted in Google temporarily withdrawing the image generator from its platform.

With Imagen 3 now integrated into Gemini, users can explore its advanced features and churn out their own artistic creations with just a few input commands. Whether for professional use or personal enjoyment, Google’s latest offering stands to transform the way people create and interact with digital imagery, marking an exciting frontier in AI-generated art. To explore Imagen 3, interested users are encouraged to visit the Gemini platform for a firsthand experience.