TIMI: Training-Free Image-to-3D Multi-Instance Generation with Spatial Fidelity
This addresses the need for accurate spatial fidelity in image-to-3D generation for real-world applications, offering an incremental improvement by leveraging underutilized priors in existing models.
The paper tackles the problem of generating 3D models from images with multiple instances while ensuring precise spatial fidelity, proposing a training-free framework that achieves better performance in layout and instance quality without additional training and with faster inference.
Precise spatial fidelity in Image-to-3D multi-instance generation is critical for downstream real-world applications. Recent work attempts to address this by fine-tuning pre-trained Image-to-3D (I23D) models on multi-instance datasets, which incurs substantial training overhead and struggles to guarantee spatial fidelity. In fact, we observe that pre-trained I23D models already possess meaningful spatial priors, which remain underutilized as evidenced by instance entanglement issues. Motivated by this, we propose TIMI, a novel Training-free framework for Image-to-3D Multi-Instance generation that achieves high spatial fidelity. Specifically, we first introduce an Instance-aware Separation Guidance (ISG) module, which facilitates instance disentanglement during the early denoising stage. Next, to stabilize the guidance introduced by ISG, we devise a Spatial-stabilized Geometry-adaptive Update (SGU) module that promotes the preservation of the geometric characteristics of instances while maintaining their relative relationships. Extensive experiments demonstrate that our method yields better performance in terms of both global layout and distinct local instances compared to existing multi-instance methods, without requiring additional training and with faster inference speed.