Mono4DGS-HDR: High Dynamic Range 4D Gaussian Splatting from Alternating-exposure Monocular Videos

Abstract

We introduce Mono4DGS-HDR, the first system for reconstructing renderable 4D high dynamic range (HDR) scenes from unposed monocular low dynamic range (LDR) videos captured with alternating exposures. To tackle such a challenging problem, we present a unified framework with two-stage optimization approach based on Gaussian Splatting. The first stage learns a video HDR Gaussian representation in orthographic camera coordinate space, eliminating the need for camera poses and enabling robust initial HDR video reconstruction. The second stage transforms video Gaussians into world space and jointly refines the world Gaussians with camera poses. Furthermore, we propose a temporal luminance regularization strategy to enhance the temporal consistency of the HDR appearance. Since our task has not been studied before, we construct a new evaluation benchmark using publicly available datasets for HDR video reconstruction. Extensive experiments demonstrate that Mono4DGS-HDR significantly outperforms alternative solutions adapted from state-of-the-art methods in both rendering quality and speed.

Method Overview

The overview of Mono4DGS-HDR:

(a) We infer vision foundation models on the input alternating-exposure video to extract 2D priors, which provide scene initialization and regularization.
(b) We propose a novel two-stage Gaussian optimization procedure, which includes video Gaussian training in the first stage, world Gaussian fine-tuning in the second stage, and a video-to-world Gaussian transformation strategy. The HDR Gaussians are optimized through 2D prior supervision, Gaussian motion regularization, temporal luminance regularization and HDR photometric reprojection loss.

Demo Videos of Mono4DGS-HDR

We provide demo videos for our Mono4DGS-HDR, including the precomputed 2D prior information and our rendered results. Click the scene buttons to switch between scenes. All four video types are displayed simultaneously.

Select Scene:

Input Video with Prior Track

Prior Depth

Prior Flow

Prior Mask

Rendered HDR Video

Rendered Depth

Rendered HDRV (Dynamic)

Rendered HDRV (Static)

Novel View and Time Rendering

Induced 3D Tracking by Gaussian Motion

Comparison Results on Recovered HDR Videos

Comparison of our method with state-of-the-art approaches on recovered HDR videos. We extend GFlow, SplineGS and MoSca to HDR mode for fair comparison. Click the scene buttons to switch between different scenes.

Select Scene:

Note: Students_2 scene has no camera motion (fixed camera).

GaussHDR

GFlow-HDR

SplineGS-HDR

HDR-HexPlane

MoSca-HDR

Mono4DGS-HDR (Ours)

Comparison Results under Fix-View-Change-Time Setting

Comparison of our method with MoSca under the fix-view-change-time setting. We extend MoSca to HDR mode for fair comparison. Click the scene buttons to switch between different scenes.

Select Scene:

MoSca-HDR

Mono4DGS-HDR (Ours)

Comparison Results under Fix-Time-Change-View Setting

Comparison of our method with MoSca under the fix-time-change-view setting. We extend MoSca to HDR mode for fair comparison. Click the scene buttons to switch between different scenes.

Select Scene:

MoSca-HDR

Mono4DGS-HDR (Ours)

BibTeX

@article{liu2025mono4dgshdr,
      title={Mono4DGS-HDR: High Dynamic Range 4D Gaussian Splatting from Alternating-exposure Monocular Videos}, 
      author={Jinfeng Liu and Lingtong Kong and Mi Zhou and Jinwen Chen and Dan Xu},
      journal={arXiv preprint arXiv:2510.18489},
      year={2025},
}