MagicClay introduces a novel sculpting tool that combines neural fields and triangular meshes. The hybrid approach allows for precise sculpting of regions based on textual prompts while maintaining consistency between representations. By optimizing both mesh and Signed Distance Field (SDF) jointly, MagicClay achieves superior geometry generation compared to existing methods. The framework balances consistency and regularization throughout the shape optimization process, resulting in high-quality generated geometry. Leveraging recent advancements in differentiable mesh reconstruction, MagicClay enables adaptive allocation of triangles based on the SDF, leading to improved geometric quality. The tool offers novel control over sequential prompt-based edits on the same mesh for the first time.
In un'altra lingua
dal contenuto originale
arxiv.org
Approfondimenti chiave tratti da
by Amir Barda,V... alle arxiv.org 03-06-2024
https://arxiv.org/pdf/2403.02460.pdfDomande più approfondite