Amazon Web Services (AWS) introduced the Amazon Titan Image Generator on November 29 during its AWS re:Invent 2023 conference. Amazon Titan Image Generator helps content creators quickly create and refine images using English natural language prompts. The new tool is currently available in preview for AWS Bedrock customers.
Amazon claims that its Titan models incorporate 25 years of artificial intelligence (AI) and machine learning (ML) innovation and offer various high-performing image, multimodal, and text model options. AWS mentioned in its blog that the Amazon Titan Image Generator helps firms in the advertising, e-commerce, and media sectors create studio-quality, realistic images at low cost and in large volumes. Also, the Titan model can understand complex prompts with multiple objects and generate relevant images.
#AWS is putting the “art” back into “artificial intelligence” with the new Amazon Titan Image Generator. 🎨✨🖼️ #MachineLearning #AI
Learn more. 🔗 https://t.co/pOzi2TaPPk pic.twitter.com/NvRdCtLuiu
— Amazon Web Services (@awscloud) November 29, 2023
Notably, the Amazon Titan model is similar to Meta’s Llama, OpenAI’s GPT, and Google’s Palm large language models. Amazon says that it is trained on high-quality, diverse data to develop accurate outputs, such as realistic images with limited distortions. With Titan Image Generator’s image editing features, users will be able to automatically edit an image with a text prompt using a built-in segmentation model. It also supports inpainting with an image mask and outpainting to change or extend an image background. Users can also configure image dimensions and specify the number of image variations.
According to AWS, the AI-powered image generator also mitigates harmful content generation to support the responsible use of AI. The images developed by this tool will have an invisible watermark. This will help to reduce the misinformation spread. Along with the image generator, AWS has also launched Titan Multimodal Embeddings, which will help to develop more accurate and contextually relevant multimodal search and recommendation experiences for end users.