The new model, Titan Image Generator v2, introduces several advanced features. According to AWS principal developer advocate Channy Yun, users can now guide image generation with reference images, edit existing visuals, remove backgrounds, and create image variations.

Yun explains that Titan Image Generator v2 can intelligently detect and segment multiple foreground objects. It also allows users to generate color-conditioned images based on a color palette and use image conditioning to shape their creations.

Titan Image Generator v2 supports image conditioning, enabling it to focus on specific visual characteristics such as edges, object outlines, and structural elements. The model can be fine-tuned with reference images, like a product or company logo, to maintain a consistent aesthetic. AWS has not disclosed the exact data used to train its Titan Image Generator models, stating only that it uses a combination of proprietary and licensed data.

Many vendors keep their training data confidential, viewing it as a competitive advantage and a potential source of IP-related lawsuits. To address concerns, AWS offers an indemnification policy that protects customers if a Titan model inadvertently replicates a copyrighted training example.

During Amazon's recent second-quarter earnings call, CEO Andy Jassy expressed strong confidence in generative AI technologies like AWS' Titan models, despite some enterprise hesitation and the high costs associated with training and maintaining these models. Jassy emphasized that the generative AI space is poised for rapid growth, primarily driven by cloud-based solutions.