Improve model card: Add pipeline tag, paper, project page, code, usage, and abstract

README.md

@@ -2,9 +2,64 @@
 tags:
 - model_hub_mixin
 - pytorch_model_hub_mixin
+pipeline_tag: text-to-3d
 ---
 
-This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
-- Code: [More Information Needed]
-- Paper: [More Information Needed]
-- Docs: [More Information Needed]
+# FastMesh: Efficient Artistic Mesh Generation via Component Decoupling
+
+This repository contains the official implementation and model weights for **FastMesh**, presented in the paper [FastMesh: Efficient Artistic Mesh Generation via Component Decoupling](https://huggingface.co/papers/2508.19188). Our approach efficiently produces 3D objects by substantially reducing the number of tokens required for generation.
+
+-   🏠 [Project Page](https://jhkim0759.github.io/projects/FastMesh/)
+-   💻 [GitHub Repository](https://github.com/jhkim0759/FastMesh)
+
+<p align="center">
+    <img width="90%" alt="pipeline", src="https://github.com/jhkim0759/FastMesh/raw/main/assets/Teaser.png">
+</p>
+
+## Abstract
+
+Recent mesh generation approaches typically tokenize triangle meshes into sequences of tokens and train autoregressive models to generate these tokens sequentially. Despite substantial progress, such token sequences inevitably reuse vertices multiple times to fully represent manifold meshes, as each vertex is shared by multiple faces. This redundancy leads to excessively long token sequences and inefficient generation processes. In this paper, we propose an efficient framework that generates artistic meshes by treating vertices and faces separately, significantly reducing redundancy. We employ an autoregressive model solely for vertex generation, decreasing the token count to approximately 23% of that required by the most compact existing tokenizer. Next, we leverage a bidirectional transformer to complete the mesh in a single step by capturing inter-vertex relationships and constructing the adjacency matrix that defines the mesh faces. To further improve the generation quality, we introduce a fidelity enhancer to refine vertex positioning into more natural arrangements and propose a post-processing framework to remove undesirable edge connections. Experimental results show that our method achieves more than 8$\times$ faster speed on mesh generation compared to state-of-the-art approaches, while producing higher mesh quality.
+
+## Quick Start
+
+### Installation
+
+Our environment has been tested on CUDA 11.8 with A6000.
+(It would be greatly appreciated if you report an issue when you find any errors.)
+
+```bash
+conda create -n fastmesh python=3.10
+conda activate fastmesh
+pip install torch==2.7.1 torchvision==0.22.1 torchaudio==2.7.1 --index-url https://download.pytorch.org/whl/cu118
+pip install -r requirements.txt
+pip install flash-attn --no-build-isolation
+```
+
+### Inference
+
+Once the environment is set up and activated, you can run inference from sampled point clouds using the provided scripts:
+
+**Generate meshes from sampled point cloud with V1K variant:**
+```bash
+python inference.py --mesh_path assets --variant V1K --batch_size 3
+```
+
+**Generate meshes from sampled point cloud with V4K variant:**
+```bash
+python inference.py --mesh_path assets --variant V4K --batch_size 1
+```
+
+## Citation
+
+If you find our work helpful, please consider citing:
+
+```bibtex
+@misc{kim2025fastmesh,
+      title={FastMesh: Efficient Artistic Mesh Generation via Component Decoupling}, 
+      author={Jeonghwan Kim and Yushi Lan and Armando Fortes and Yongwei Chen and Xingang Pan},
+      year={2025},
+      eprint={2508.19188},
+      archivePrefix={arXiv},
+      url={https://arxiv.org/abs/2508.19188}, 
+}
+```