AI
AI

Getty Images Enhances Its AI-Driven Image Generation Model with New Editing Features

Photo credit: www.gadgets360.com

On Tuesday, Getty Images announced a significant enhancement to its artificial intelligence (AI) image generation model, now utilizing the Nvidia Edify architecture. This upgrade promises not only faster image generation speeds but also a wider variety of outcomes from a single input prompt. According to the company, the revised model enhances the detail in 4K images and maintains a high level of adherence to longer prompts.

Getty Images Updates AI Model

The announcement was made in a newsroom post, where Getty Images confirmed that this updated model is now available for testing on its platform. Built on Nvidia’s Edify architecture, which is part of Nvidia Picasso—an AI framework designed for developing generative AI models for visual arts—this new system introduces several key advancements.

Among the enhancements, the Edify architecture significantly boosts the model’s speed and output quality. The AI can now generate four images in just six seconds, a performance that the company claims is twice as fast as its predecessor. Furthermore, the model’s ability to interpret and execute prompts has been notably improved, allowing it to handle prompts of up to 250 words and process more complex requirements.

The new model also incorporates advanced image modification features powered by a large language model (LLM). Users are now equipped to easily add or change specific elements within an image, expand the canvas size, or remove backgrounds, providing enhanced flexibility in image creation.

Regarding the training data for the model, Getty Images stated, “The model was only trained on high-resolution, licensed or owned images and metadata from Getty Images’ vast creative library.” The company emphasized that no data or images were taken from the internet, nor were they generated artificially or sourced from other generative models.

Importantly, Getty Images has also trained the image generator to avoid producing outputs that might infringe upon intellectual property rights or artist rights. Consequently, requests for images featuring recognizable individuals, protected locations, or trademarked brands will not be honored.

Source
www.gadgets360.com

Related by category

Amazon’s Top TV Receives Exciting Free Upgrades

Photo credit: www.techradar.com Amazon's Flagship Fire TV Omni with Mini-LED...

Samsung Maintains Leading Position as Global Smartphone Shipments Increase 0.2% Year-over-Year in Q1 2025, Reports Canalys

Photo credit: www.gadgets360.com Global smartphone shipments saw a minimal year-on-year...

Duolingo Announces Doubling of Language Courses Thanks to AI

Photo credit: www.theverge.com Duolingo has announced a significant increase in...

Latest news

How to View Star Wars: Tales of the Underworld in Fortnite

Photo credit: dotesports.com Fortnite is gearing up to offer an...

Ajith Kumar’s Wife Breaks Her Silence with First Post Following Actor’s Hospitalization Reports

Photo credit: www.news18.com Last Updated: April 30, 2025, 21:47 IST Tamil...

10 Iconic ’90s Movies That Split Critics and Audiences

Photo credit: movieweb.com Film critics play a vital role in...

Breaking news