Imagen 3

Imagen 3 is Google DeepMind's highest-quality text-to-image model, producing photorealistic images with fine detail, realistic lighting, and few artifacts. Announced at Google I/O 2024 and generally available from December 2024, it improves prompt understanding over Imagen 2 and renders styles ranging from photorealism to animation, with features such as mask-based editing, upscaling, and customization. Every generated image carries SynthID, Google DeepMind's imperceptible watermark, supporting provenance tracking. Imagen 3 handles text rendering within images, multiple aspect ratios including 1:1, 16:9, 9:16, 3:4, and 4:3, and batch generation of up to four images per request.