DiMeR: Disentangled Mesh Reconstruction Model
Lutao Jiang, Jiantao Lin, Kanghao Chen, Wenhang Ge, Xin Yang, Yifan Jiang, Yuanhuiyi Lyu, Xu Zheng, Yingcong Chen
2025-04-25
Summary
This paper talks about DiMeR, a new AI model that can create accurate 3D shapes (called meshes) from just a few pictures, by separately figuring out the shape and the color details of an object.
What's the problem?
The problem is that building a detailed 3D model from only a small number of photos is really hard, especially when it comes to getting both the shape and the texture right. Most current methods mix these two things together, which can lead to less accurate results.
What's the solution?
The researchers designed DiMeR to use two separate streams: one that looks at normal maps to understand the shape and another that uses regular color images for the texture. By keeping these two processes apart, the model can focus on each part more clearly and reconstruct 3D objects much more accurately than before.
Why it matters?
This matters because it makes it easier and faster to create high-quality 3D models from just a few photos, which is super useful for things like video games, movies, virtual reality, and even online shopping.
Abstract
DiMeR, a disentangled dual-stream feed-forward model, improves sparse-view 3D mesh reconstruction using normal maps for geometry and RGB images for texture, outperforming previous methods by over 30% in Chamfer Distance.