Feedforward 3D Editing via Text-Steerable Image-to-3D
This addresses the need for easy editing of 3D assets in applications like design, AR/VR, and robotics, representing a novel method for a known bottleneck.
The paper tackles the problem of editing AI-generated 3D assets by introducing Steer3D, a feedforward method that adds text steerability to image-to-3D models, enabling language-based editing with results that are 2.4x to 28.5x faster and more faithful to instructions and original assets compared to competing methods.
Recent progress in image-to-3D has opened up immense possibilities for design, AR/VR, and robotics. However, to use AI-generated 3D assets in real applications, a critical requirement is the capability to edit them easily. We present a feedforward method, Steer3D, to add text steerability to image-to-3D models, which enables editing of generated 3D assets with language. Our approach is inspired by ControlNet, which we adapt to image-to-3D generation to enable text steering directly in a forward pass. We build a scalable data engine for automatic data generation, and develop a two-stage training recipe based on flow-matching training and Direct Preference Optimization (DPO). Compared to competing methods, Steer3D more faithfully follows the language instruction and maintains better consistency with the original 3D asset, while being 2.4x to 28.5x faster. Steer3D demonstrates that it is possible to add a new modality (text) to steer the generation of pretrained image-to-3D generative models with 100k data. Project website: https://glab-caltech.github.io/steer3d/