Addressing Janus Issue in Text-to-3D via Orientation-Controlled Diffusion Models

In the evolving landscape of text-to-3D technology, DreamFusion [9] optimizes implicit representations like neural radiance fields using score distillation sampling but faces limitations in both fidelity and speed. Specifically, it encounters the multihead Janus issue and exhibits a relatively slow optimization process. We present OrientDream, a camera orientation conditioned framework for efficient, multiview consistent 3D generation from text prompts. OrientDream achieves this by pretraining a 2D text-to-image diffusion module with camera orientation features and utilizing data from MVImgNet [17]. To shorten training time, we also introduced a decoupled backpropagation technique, allowing for multiple updates of implicit parameters per optimization cycle. Our experiments reveal that our method not only produces high-quality 3D models with consistent multiview properties but also achieves an optimization speed significantly greater than existing methods, as quantified by comparative metrics.