Gemini image generation is adding more editing tools (www.engadget.com)

Google’s DeepMind team has introduced powerful new image editing tools in the Gemini app, enhancing how AI-generated images can be refined and personalized. These updates focus on maintaining visual consistency, especially for human subjects, allowing users to modify backgrounds, outfits, or settings while preserving the original appearance of characters. This capability marks a significant advancement in creating realistic, editable AI-generated portraits that remain authentic across different scenes. The latest model also introduces sophisticated multi-stage editing features, enabling users to merge elements from separate images into entirely new compositions and use visual traits from one image as prompts for further design. These nuanced controls allow for step-by-step adjustments without losing previous edits, offering a flexible workflow that enhances creative precision. Google has committed to transparency by watermarking all AI-generated or edited images in Gemini, addressing concerns about content authenticity. This release comes after earlier challenges with image generation of humans, which was temporarily disabled due to inaccuracies in diversity representation. The reintroduction of human image generation with the Imagen 3 model alongside these new editing tools signals Google’s renewed push to refine ethical and technical standards in AI-driven image synthesis.