Korean Researchers Develop AI Model That Can Control Lighting Effects in Photos and Videos With Text Input
Hwang Sujin Reporter
hwang075609@gmail.com | 2025-02-23 14:57:11
A research team at the Ulsan Institute of Science and Technology (UNIST) in South Korea has developed an artificial intelligence (AI) model called "Text2Relight" that can control lighting effects in photos and videos with text input. This new technology is expected to be useful in a variety of fields, including film production, photo editing, and virtual and augmented reality.
The Text2Relight model can understand creative natural language text and express various lighting characteristics, such as color, brightness, and emotional atmosphere. This makes it possible to easily correct the color of photos and videos without using editing tools, and even add emotional touches to the color. For example, you can use text like "warm chicken" or "cold blue light" to easily adjust the color tone of a photo or video. The AI model can also adjust the color tone of both the subject and background simultaneously without distorting the original image.
This technology is expected to save time in photo and video editing and increase the immersion of virtual and augmented reality content. The research team believes that Text2Relight has great potential in the content field.
The research was conducted in collaboration with Adobe and was supported by the Ministry of Science and ICT. The research findings will be presented at the 2025 AAAI conference in Philadelphia, USA.
WEEKLY HOT
- 1The flowers at Magok Seoul Botanic Garden are blessing the approach of spring.
- 2Spring has already arrived at Incheon Grand Park, and the flowers are bursting into bloom.
- 3Iran’s Retaliation Deals $800M Blow to U.S. Bases; Key Missile Defense Systems Hit
- 4Trump Deploys ICE to Airports as Budget Standoff Leaves Security Understaffed
- 5Naver D2SF Launches 18th Campus Tech Startup Competition to Foster Next-Gen Innovators
- 6Vishay Unveils Ultra-Compact 0404 RGB LED with Independent Chip Control for Enhanced Color Precision