WildSplat: Feedforward Gaussian Splatting from Unposed In-the-Wild Images
It solves the problem of feedforward 3D reconstruction under varying illumination for unposed in-the-wild images, which is a known bottleneck in the field.
WildSplat is the first feedforward 3D Gaussian Splatting framework for appearance-conditioned novel-view synthesis from unposed in-the-wild images, achieving state-of-the-art performance in both novel view synthesis and appearance editing from sparse inputs in a single forward pass.
While feedforward 3D reconstruction excels at efficient novel view synthesis, it typically falters when faced with scenes under varying illumination. To this end, we introduce WildSplat, the first feedforward 3D Gaussian Splatting framework capable of appearance-conditioned novel-view synthesis for unposed in-the-wild images. To handle inconsistent photometric conditions, we propose a dual-branch architecture that explicitly decouples geometry from appearance. The geometry branch extracts an appearance-invariant 3D structure and jointly predicts camera poses. To govern the rendering appearance, the appearance branch injects target appearance cues into the content features via a globally pre-modulated cross-attention mechanism. To further prevent feature entanglement, we introduce a joint multi-reference training strategy that stabilizes the training process. Extensive experiments show that WildSplat surpasses existing optimization-based and feedforward methods, achieving state-of-the-art performance in in-the-wild novel view synthesis and appearance editing from sparse inputs in a single forward pass.