Rendering synthetic objects into legacy photographs
A group of students from the University of Illinois at Urbana-Champaign have demonstrated a ground-breaking system that allows a user realistically insert synthetic objects into legacy photographs quickly and easily.
The system requires that the user first annotate an image, pointing out light sources and object boundaries using software-based tools. Once an image in annotated, the system accurately renders synthetic objects placed ‘into’ the photograph, accounting for lighting, shading, reflection, illumination and other individual object properties like transparency, diffuse or specular.
The system has obvious applications in the movie and gaming industry, as well as interior design and user content creation. The video below speaks for itself,
In the archives of nearly every major heritage brand – Louis Vuitton, Mercedes-Benz, Coca-Cola or…
You woke up this morning, made a series of choices, and ended up here reading…
Writing sucked long before LLMs showed up. Sure, today's doomsayers love pointing at ChatGPT as…
DARPA's O-Circuit program looks to build a new class of biologically inspired computer equipped with…
As AI races into classrooms worldwide, Google is finding that the toughest lessons on how…
The push to eat bugs is not an organic movement coming from the people, but…
View Comments
very cool! I am assuming the objects would be modeled in a different program and inserted? how are theh new materials scaled/degenerated to match the original image quality? I am interested to follow the development of this program and see the far reaching effects that it can and will have.
very cool! I am assuming the objects would be modeled in a different program and inserted? how are theh new materials scaled/degenerated to match the original image quality? I am interested to follow the development of this program and see the far reaching effects that it can and will have.
very great post.