Shulou(Shulou.com)11/24 Report--

CTOnews.com, May 24 (Xinhua) researchers at the Massachusetts Institute of Technology have developed a model format with real physical volume, which maps 3D shapes from volume to volume rather than surface to surface, so that the surface of the 3D model can be distorted and squeezed in the same way as the display.

In previous computer graphics and computer-aided design, three-dimensional objects were usually presented by the outline of their outer surface. (CTOnews.com Note: at present, all models are composed of many faces), excluding the actual volume collision box inside the model. This modeling format is the result of previous technology, and can store and present models with high polygons with higher efficiency.

The model demonstration conducted by ▲ CTOnews.com shows that the current model is made up of many faces, and the model itself has nothing to do with actual physics, so the mold-piercing phenomenon in the game is also the reason, but this kind of model also has some limitations in operation because of the lack of internal volume collision box (CTOnews.com Note: this kind of model must be bound to the model bone in order to animate normally), so it has some limitations in operation, for example, for a character model without bound bones. When you move the hands and feet of a character model, you can actually only move the faces or nodes that make up these hands and feet, rather than being able to directly control the movement of the hands and feet full of bones, tendons and muscles as it seems. therefore, it is impossible to directly make the hands and feet take various shapes as in reality.

In terms of animation, these differences are particularly problematic when using these models to develop mapping algorithms, because mapping algorithms can automatically find the relationship between different shapes, but it takes a higher cost to create a model with extremely fine bones all over the body, placing more burden on the budget.

▲ image source mit.edu this technique directly packages the overall shape and mass of the model into a tetrahedral mesh to simulate the fine part of the object more effectively, instead of the traditional model "is actually a structure made up of many faces" (and requires binding bones to achieve similar physical motion effects). Therefore, it can avoid the simple external "many faces" deformation when the model simulates the force.

This technique is useful in many graphics applications such as visual computing, computational manufacturing and engineering, such as transferring the motion of previously animated 3D characters to new 3D models or scans. The same algorithm can transfer textures, annotations, and physical properties from one 3D shape to another.

"switching from the surface to the volume extends the rubber glove to the entire hand. Our method brings geometric mapping closer to physical reality," said Mazdak Abulnaga, a graduate student in Electrical Engineering and computer Science (EECS) and lead author of the mapping paper.

Abulnaga co-authored the paper with former MIT postdoctoral Oded Stein, who now teaches at the University of Southern California; Polina Golland, Professor Sunlin and Priscilla Chou of EECS, lead researcher of the MIT computer Science and artificial Intelligence Lab (CSAIL) and Leader of the Medical Vision Group; and Associate Professor Justin Solomon,EECS and Leader of the CSAIL Geometry data processing Group. The study will be presented at the ACM SIGGRAPH conference.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.