Google has introduced Imagen 3, an advanced AI system that generates high-quality, photorealistic images up to 1024x1024 pixels from text prompts, outperforming similar tools like Midjourney, DALL-E 3, and Grok-2.
Imagen 3 uses a latent diffusion model trained on a large dataset of images, text, and annotations, with improved prompt understanding and safeguards to prevent offensive or illegal content generation.
It allows users to create, edit, and upscale images by providing detailed instructions and selecting specific areas for adjustments, with an interactive approach and user-friendly design. Some users were frustrated with its censorship mechanisms, as non-problematic prompts are also being blocked.
Google included its digital watermark SynthID for origin tracking to minimize harmful content. Imagen 3 is initially available only in the U.S. through the AI Test Kitchen service, ImageFX, and Vertex AI, with plans to expand to other countries.
Google may be taking a cautious approach with Imagen 3's rollout, possibly influenced by backlash against the Gemini model for producing historically inaccurate images, and trying to avoid controversy experienced with earlier image creation models.
Sources: ZDNet, PCMag, Android Central, Times of India, Deccan Herald, PetaPixel, Mobile Bulgaria, Chip Turkey, TechGear, Analytics Insight, RM Update, Times Now, PCM, Rosario3, Planetared, Notebookcheck
This article was written in collaboration with Generative AI news company Alchemiq.
