Fast High Dynamic Range Radiance Fields for Dynamic Scenes

1Schoo of CS, HUST 2School of EIC, HUST

*Equal Contributions. +Project Lead. Corresponding Author.

HDR-HexPlane can reconstruct dynamic scene within hours and render HDR image.


Neural Radiances Fields (NeRF) and its extensions have shown great success to represent 3D scenes and synthesis novel-view images. However, most NeRF methods take in low-dynamic-range (LDR) images, which may lose details especially with nonuniform illumination. Some previous NeRF methods attempt to introduce high-dynamic-range (HDR) techniques but mainly target at static scenes. To extend HDR NeRF methods to wider applications, we propose a dynamic HDR NeRF framework, named as HDR-HexPlane, which can learn 3D scenes from dynamic 2D images captured with various exposures. A learnable exposure mapping function is constructed to obtain adaptive exposure values for each image. Based on the monotonically-increasing prior, a camera response function is designed for stable learning. With the proposed model, high-quality novel-view images at any time point can be rendered with any desired exposure. We further construct a dataset containing multiple dynamic scenes captured with diverse exposures for evaluation. All the datasets and code will be released.

The framework of Our Model.

Balanced HDR Images Rendering

HDR-HexPlane can render tonemapped HDR Image Given both overexposure and underexposure images.

Multi Exposure Images Rendering

HDR-HexPlane can render novel views with different exposures.

Scene Depth Rendering

With Multi-views temporal images input, HDR-HexPlane also propose precise depth estimation.