3DrawAgent: Teaching LLM to Draw in 3D with Early Contrastive Experience

• URL: http://arxiv.org/abs/2604.08042v1
• Date: Thu, 09 Apr 2026 09:47:00 GMT
• Title: 3DrawAgent: Teaching LLM to Draw in 3D with Early Contrastive Experience
• Abstract summary: We introduce 3DrawAgent, a training-free, language-driven framework for 3D sketch generation.<n>Unlike prior 2D sketch agents, our method introduces a relative experience optimization strategy.<n>We show that 3DrawAgent can generate complex and coherent 3D Bezier sketches from diverse textual prompts.
• Score: 17.17661155254756
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Sketching in 3D space enables expressive reasoning about shape, structure, and spatial relationships, yet generating 3D sketches through natural language remains a major challenge. In this work, we introduce 3DrawAgent, a training-free, language-driven framework for 3D sketch generation that leverages large language models (LLMs) to sequentially draw 3D Bezier curves under geometric feedback. Unlike prior 2D sketch agents, our method introduces a relative experience optimization strategy that adapts the recently proposed Group Reward Policy Optimization (GRPO) paradigm. Instead of relying on explicit ground-truth supervision, we construct pairwise comparisons among generated sketches, with each pair consisting of a relatively better and a worse result based on CLIP-based perceptual rewards and LLM-based fine-grained qualitative assessment. These experiences are then used to iteratively refine the prior knowledge of 3D drawing, enabling black-box reinforcement of the model's 3D awareness. This design allows our model to self-improve its spatial understanding and drawing quality without parameter updates. Experiments show that 3DrawAgent can generate complex and coherent 3D Bezier sketches from diverse textual prompts, exhibit emergent geometric reasoning, and generalize to novel shapes, establishing a new paradigm for advancing the field of training-free 3D sketch intelligence.

Related papers

• 4-Doodle: Text to 3D Sketches that Move! [60.89021458068987]
  4-Doodle is the first training-free framework for generating dynamic 3D sketches from text.<n>Our method produces temporally realistic and structurally stable 3D sketch animations, outperforming existing baselines in both fidelity and controllability.
  arXiv  Detail & Related papers  (2025-10-29T09:33:29Z)
• UniUGG: Unified 3D Understanding and Generation via Geometric-Semantic Encoding [65.60549881706959]
  We introduce UniUGG, the first unified understanding and generation framework for 3D modalities.<n>Our framework employs an LLM to comprehend and decode sentences and 3D representations.<n>We propose a spatial decoder leveraging a latent diffusion model to generate high-quality 3D representations.
  arXiv  Detail & Related papers  (2025-08-16T07:27:31Z)
• Shape from Semantics: 3D Shape Generation from Multi-View Semantics [30.969299308083723]
  Existing 3D reconstruction methods utilize guidances such as 2D images, 3D point clouds, shape contours and single semantics to recover the 3D surface.<n>We propose a novel 3D modeling task called Shape from Semantics'', which aims to create 3D models whose geometry and appearance are consistent with the given text semantics when viewed from different views.
  arXiv  Detail & Related papers  (2025-02-01T07:51:59Z)
• Sketch3D: Style-Consistent Guidance for Sketch-to-3D Generation [55.73399465968594]
  This paper proposes a novel generation paradigm Sketch3D to generate realistic 3D assets with shape aligned with the input sketch and color matching the textual description. Three strategies are designed to optimize 3D Gaussians, i.e., structural optimization via a distribution transfer mechanism, color optimization with a straightforward MSE loss and sketch similarity optimization with a CLIP-based geometric similarity loss.
  arXiv  Detail & Related papers  (2024-04-02T11:03:24Z)
• Control3D: Towards Controllable Text-to-3D Generation [107.81136630589263]
  We present a text-to-3D generation conditioning on the additional hand-drawn sketch, namely Control3D. A 2D conditioned diffusion model (ControlNet) is remoulded to guide the learning of 3D scene parameterized as NeRF. We exploit a pre-trained differentiable photo-to-sketch model to directly estimate the sketch of the rendered image over synthetic 3D scene.
  arXiv  Detail & Related papers  (2023-11-09T15:50:32Z)
• Sketch-A-Shape: Zero-Shot Sketch-to-3D Shape Generation [13.47191379827792]
  We investigate how large pre-trained models can be used to generate 3D shapes from sketches. We find that conditioning a 3D generative model on the features of synthetic renderings during training enables us to effectively generate 3D shapes from sketches at inference time. This suggests that the large pre-trained vision model features carry semantic signals that are resilient to domain shifts.
  arXiv  Detail & Related papers  (2023-07-08T00:45:01Z)
• SENS: Part-Aware Sketch-based Implicit Neural Shape Modeling [124.3266213819203]
  We present SENS, a novel method for generating and editing 3D models from hand-drawn sketches. S SENS analyzes the sketch and encodes its parts into ViT patch encoding. S SENS supports refinement via part reconstruction, allowing for nuanced adjustments and artifact removal.
  arXiv  Detail & Related papers  (2023-06-09T17:50:53Z)
• Make Your Brief Stroke Real and Stereoscopic: 3D-Aware Simplified Sketch to Portrait Generation [51.64832538714455]
  Existing studies only generate portraits in the 2D plane with fixed views, making the results less vivid. In this paper, we present Stereoscopic Simplified Sketch-to-Portrait (SSSP), which explores the possibility of creating Stereoscopic 3D-aware portraits. Our key insight is to design sketch-aware constraints that can fully exploit the prior knowledge of a tri-plane-based 3D-aware generative model.
  arXiv  Detail & Related papers  (2023-02-14T06:28:42Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.