This new model is currently only available in select regions in the United States but Amazon is likely to expand its customer base in the rest of the regions. Amazon Web Services (AWS) has not explicitly mentioned what data they train their models on but it has been reported that AWS may use a mix of both licensed and propriety data.

Amazon Titan Image Generator v2 and Its New Features

The updated model comes with new features including image consistency, image guidance with colour palette, background and subject consistency. The new model still carries the features of the Amazon Titan Image Generator v1.

Image Conditioning enables users to use reference images to generate new images. Users can also ask the model to focus on particular details or elements in the image such as the object outlines, segmentation maps, edges and more. v2 currently supports two types of image conditioning such as the Canny edge and segmentation.

Another new feature in v2 is colour conditioning which assists users with generating images with certain colour palettes in mind. This feature is helpful for designers who want to create images with colour branding in mind. Users can also remove backgrounds in the new Titan v2 upgrade. v2 can automatically detect the background elements and effectively remove them to make the final image look clean without any overlaps.

v2 also allows the seamless incorporation of subjects into the various visual scenarios. Users can fine-tune the model and write a text prompt which lets the model generate images. The generated images will feature the subject in the desired contexts.

Suggested:

Generative AI image Tool Introduced by Amazon for Advertisers.

Amazon Prime will introduce ads on Prime Video in 2024.

The post Amazon Titan Image Generator v2 Has Been Released appeared first on Tricky Enough.