dorsal/arxiv
View SchemaEZBlender: Efficient 3D Editing with Plan-and-ReAct Agent
| Authors | Hao Wang, Wenhui Zhu, Shao Tang, Zhipeng Wang, Xuanzhao Dong, Xin Li, Xiwen Chen, Ashish Bastola, Xinhao Huang, Yalin Wang, Abolfazl Razi |
|---|---|
| Categories | |
| ArXiv ID | 2601.07143vv1 |
| URL | https://arxiv.org/abs/2601.07143 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
As a cornerstone of the modern digital economy, 3D modeling and rendering demand substantial resources and manual effort when scene editing is performed in the traditional manner. Despite recent progress in VLM-based agents for 3D editing, the fundamental trade-off between editing precision and agent responsiveness remains unresolved. To overcome these limitations, we present EZBlender, a Blender agent with a hybrid framework that combines planning-based task decomposition and reactive local autonomy for efficient human AI collaboration and semantically faithful 3D editing. Specifically, this unexplored Plan-and-ReAct design not only preserves editing quality but also significantly reduces latency and computational cost. To further validate the efficiency and effectiveness of the proposed edge-autonomy architecture, we construct a dedicated multi-tasking benchmark that has not been systematically investigated in prior research. In addition, we provide a comprehensive analysis of language model preference, system responsiveness, and economic efficiency.
{
"annotation_id": "33a89c52-2c03-45d2-bc71-73320d91d0af",
"date_created": "2026-02-17T05:53:12.427000Z",
"date_modified": "2026-02-17T05:53:12.427000Z",
"file_hash": "5ee082dd08bfbee155d40ead7e19713d770840f6ed4b79c88a5ff30a97120cf4",
"private": false,
"record": {
"abstract": "As a cornerstone of the modern digital economy, 3D modeling and rendering demand substantial resources and manual effort when scene editing is performed in the traditional manner. Despite recent progress in VLM-based agents for 3D editing, the fundamental trade-off between editing precision and agent responsiveness remains unresolved. To overcome these limitations, we present EZBlender, a Blender agent with a hybrid framework that combines planning-based task decomposition and reactive local autonomy for efficient human AI collaboration and semantically faithful 3D editing. Specifically, this unexplored Plan-and-ReAct design not only preserves editing quality but also significantly reduces latency and computational cost. To further validate the efficiency and effectiveness of the proposed edge-autonomy architecture, we construct a dedicated multi-tasking benchmark that has not been systematically investigated in prior research. In addition, we provide a comprehensive analysis of language model preference, system responsiveness, and economic efficiency.",
"arxiv_id": "2601.07143",
"authors": [
"Hao Wang",
"Wenhui Zhu",
"Shao Tang",
"Zhipeng Wang",
"Xuanzhao Dong",
"Xin Li",
"Xiwen Chen",
"Ashish Bastola",
"Xinhao Huang",
"Yalin Wang",
"Abolfazl Razi"
],
"categories": [
"cs.HC"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "EZBlender: Efficient 3D Editing with Plan-and-ReAct Agent",
"url": "https://arxiv.org/abs/2601.07143",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "fd831508-d8da-4973-90cf-7f2122bc6aa5",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}