ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
Unified 3D foundation models aspire to generate 3D assets and reason about them in language within a single backbone, but their text-3D interaction remains largely implicit. Existing methods concatenate text and 3D tokens into a flat sequence and rely on self-attention, collapsing coarse structural cues and fine geometric details into one undifferentiated representation. We introduce ELSA3D, a unified 3D model that addresses this with elastic semantic anchoring, structuring language and geometric reasoning jointly along matched abstraction scales. ELSA3D represents geometry with a scale-aware
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorTianjiao Yu →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorXinzhuo Li →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorYifan Shen →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorOnkar Susladkar →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorYuanzhe Liu →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorXiaona Zhou →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation
- Linked via arxiv authorIsmini Lourentzou →
ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation