Getty Images has announced the launch of an updated model for its generative AI services and tools, promising significant advancements in speed, quality, and accuracy.
The company's latest enhancement is set to provide image generation speeds of approximately six seconds for four images, effectively doubling the performance of the previous model.
This updated model has been developed in partnership with NVIDIA, utilising their Edify model architecture, part of NVIDIA Picasso. The updates to NVIDIA Edify offer substantial improvements in output speed and quality, better alignment with user prompts, longer prompt capabilities, enhanced 4K upscaling, and the ability to fine-tune models.
Grant Farhall, Chief Product Officer of Getty Images, commented on the new advancements, stating, "We continue to push the boundaries of what's possible with responsible AI-generated imagery. This updated model, and the new capabilities we are launching, will provide more opportunities for our customers to leverage AI alongside our amazing pre-shot content, to bring their visions to life more efficiently than before, and in a manner that is commercially safe with the legal protections Getty Images is known for."
The updated model introduces several key features aimed at enhancing user experience and output. These include industry-leading generation speeds of around six seconds, advanced 4K generation detail, expanded support for more detailed prompts, support for longer prompts up to 250 words, and greater control over output with advanced camera controls such as shot type and depth of field.
In addition to these updates, Getty Images is also rolling out AI modification capabilities. These features enable customers to modify both generated AI images and existing pre-shot creative images. Modification options include the ability to add or change individual elements, extend the canvas, or remove backgrounds in a single click. These functionalities are currently available on iStock and will soon be introduced to Getty Images.
Farhall highlighted the benefits of these new features, saying, "The updated model and ongoing rollout of additional AI functionalities are designed to benefit customers across a broad range of industries. These advancements provide professionals and enthusiasts with powerful tools to enhance their creative workflows, enabling safe and effective experimentation with AI."
One significant aspect of this updated model is its training source. The AI tools are trained exclusively on Getty Images' creative library, which includes premium content. This ensures that the generative AI services are clean of trademarks, logos, and other elements protected by third-party intellectual property rights. Additionally, Getty Images promises indemnification for commercial use, offering further reassurance to its users.
Moreover, the company has introduced fine-tuning capabilities that allow enterprise customers to build tailored models by incorporating their own brand assets into Getty Images' generative AI model. This feature aims to provide a more customised and brand-specific experience for enterprise clients.
The launch of this updated model is part of Getty Images' ongoing efforts to push the boundaries of AI technology while maintaining a focus on commercial safety and legal protections. Through its various brands, including iStock and Getty Images, the company seeks to offer powerful and comprehensive content solutions for customers worldwide.
