Jan 12, 2026 10:23 PM
https://www.eurekalert.org/news-releases/1112338
INTRO: Fabric has long been difficult to render digitally because of the myriad ways different yarns can be woven or knitted together. Now, Cornell researchers, in partnership with the technology company NVIDIA, have developed a method for creating digital images of cloth that more accurately captures the texture of textiles.
The team’s new study, presented Dec. 16 at the Association for Computing Machinery’s SIGGRAPH Asia 2025 meeting in Hong Kong, models how light interacts with yarns – both as it passes through and reflects off the fabric. The advance is the latest to emerge from the lab of Steve Marschner, professor of computer science in the Cornell Ann S. Bowers College of Computing and Information Science, who has worked on this problem for more than two decades.
Film industry people always complain about how difficult it is to render fabric, said Marschner, who was on a team that received a 2004 Technical Achievement Award from the Academy of Motion Picture Arts and Sciences for work on rendering translucent materials. “It’s just hard to get it to look right. It always looks fake.”
At its smallest level, fabric is composed of tiny fibers twisted together to make a strand called a ply. Multiple plies are twisted together to form yarn, which is then woven or knitted to create fabric. Unlike materials such as metal or skin, which have a solid, continuous surface, fabric is “just a bunch of fibers floating in space that happen to be held together by friction,” Marschner said.
The shape of the fibers also varies depending on the material. Cut through a fiber of wool and the end will look almost oval; cotton fibers have a kidney shape and silk looks like three- or four-sided polygons.
“It makes the structure so interesting but so difficult to model,” said Yunchen Yu, a doctoral student in the field of computer science and the study’s first author. “I think there’s never going to be one fabric model that everyone uses.”
Marschner first began researching methods for rendering fabric when he started as an assistant professor at Cornell in 2002. His first doctoral student, Piti Irawan, Ph.D. ‘08, developed a simple method that modeled how light reflects differently off fibers at different points on the cloth’s surface.
After realizing that the underlying fiber structure dictates a fabric’s appearance, Marschner began modeling fabric in a more comprehensive way. Along with Shuang Zhao, Ph.D. ’14, and Kavita Bala, provost and professor of computer science, he used a microCT scanner to image at the scale of the woven fibers. This level of detail allowed them to render fabric more accurately, but scanning was expensive and time-consuming.
Eventually, they made the process more efficient, so they didn’t need to scan each fabric they rendered. This work turned into a side project with Brooks Hagan, a professor at the Rhode Island School of Design, that enabled interior designers to visualize textiles for the production of custom fabrics... (MORE - details, no ads)
INTRO: Fabric has long been difficult to render digitally because of the myriad ways different yarns can be woven or knitted together. Now, Cornell researchers, in partnership with the technology company NVIDIA, have developed a method for creating digital images of cloth that more accurately captures the texture of textiles.
The team’s new study, presented Dec. 16 at the Association for Computing Machinery’s SIGGRAPH Asia 2025 meeting in Hong Kong, models how light interacts with yarns – both as it passes through and reflects off the fabric. The advance is the latest to emerge from the lab of Steve Marschner, professor of computer science in the Cornell Ann S. Bowers College of Computing and Information Science, who has worked on this problem for more than two decades.
Film industry people always complain about how difficult it is to render fabric, said Marschner, who was on a team that received a 2004 Technical Achievement Award from the Academy of Motion Picture Arts and Sciences for work on rendering translucent materials. “It’s just hard to get it to look right. It always looks fake.”
At its smallest level, fabric is composed of tiny fibers twisted together to make a strand called a ply. Multiple plies are twisted together to form yarn, which is then woven or knitted to create fabric. Unlike materials such as metal or skin, which have a solid, continuous surface, fabric is “just a bunch of fibers floating in space that happen to be held together by friction,” Marschner said.
The shape of the fibers also varies depending on the material. Cut through a fiber of wool and the end will look almost oval; cotton fibers have a kidney shape and silk looks like three- or four-sided polygons.
“It makes the structure so interesting but so difficult to model,” said Yunchen Yu, a doctoral student in the field of computer science and the study’s first author. “I think there’s never going to be one fabric model that everyone uses.”
Marschner first began researching methods for rendering fabric when he started as an assistant professor at Cornell in 2002. His first doctoral student, Piti Irawan, Ph.D. ‘08, developed a simple method that modeled how light reflects differently off fibers at different points on the cloth’s surface.
After realizing that the underlying fiber structure dictates a fabric’s appearance, Marschner began modeling fabric in a more comprehensive way. Along with Shuang Zhao, Ph.D. ’14, and Kavita Bala, provost and professor of computer science, he used a microCT scanner to image at the scale of the woven fibers. This level of detail allowed them to render fabric more accurately, but scanning was expensive and time-consuming.
Eventually, they made the process more efficient, so they didn’t need to scan each fabric they rendered. This work turned into a side project with Brooks Hagan, a professor at the Rhode Island School of Design, that enabled interior designers to visualize textiles for the production of custom fabrics... (MORE - details, no ads)