AESplat: Advancing Pose-Free Feed-Forward 3D Gaussian Splatting via Decoupled Appearance Modeling

Shiwei Ren; Zhiang Liu; Yongchun Fang; Hongwei Chen

cs.CVSep 29, 2026

AESplat: Advancing Pose-Free Feed-Forward 3D Gaussian Splatting via Decoupled Appearance Modeling

Authors: Shiwei Ren, Zhiang Liu, Yongchun Fang, Hongwei Chen

Organizations: Nankai University

Abstract

Pose-free feed-forward 3D Gaussian Splatting (3DGS) has demonstrated remarkable potential for generalized novel view synthesis. However, existing methods typically predict Gaussian appearance attributes represented by spherical harmonics (SH) in the same manner, overlooking the fundamental distinction between view-independent and view-dependent appearance, which results in suboptimal rendering quality. In this paper, we present AESplat, a novel and general framework for pose-free feed-forward 3DGS that introduces an effective decoupled appearance modeling strategy based on an analysis of SH, enabling higher-quality rendering. Specifically, AESplat directly derives the zeroth-order SH coefficient, which represents the base view-independent appearance component, from the input images without training. The higher-order SH coefficients are subsequently predicted by a shallow multilayer perceptron equipped with two efficient 3D-aware inductive biases to model view-dependent appearance variations. Extensive experiments across multiple datasets demonstrate that our method significantly outperforms state-of-the-art approaches, achieving a $0.8$ dB improvement in PSNR over the pose-free method NAS3R and a $1.1$ dB improvement over the pose-required method DepthSplat on the RealEstate10K dataset. Project page: https://aesplat.github.io/.

Figures & tables

Figure 1: Our method AESplat consistently outperforms state-of-the-art pose-free feed-forward methods in rendering quality across both indoor and outdoor scenes. Yellow and red boxes highlight differences in diffuse (e.g., walls) and specular (e.g., mirrors and glass) appearance, respectively.

Figure 2: Architectural Differences. Unlike (a) existing pose-free methods that predict all appearance attributes with a unified prediction head, (b) we adopt a decoupled formulation that employs tailored strategies to model view-independent and view-dependent appearance separately.

Figure 3: Pipeline of AESplat . (a) Given unposed images, AESplat jointly predicts camera poses and Gaussian primitives in a single forward pass, with their appearance attributes obtained through decoupled appearance modeling. (b) Leveraging the strong observation provided by the input image itself, the zeroth-order SH coefficient of each pixel-aligned Gaussian is directly derived from its corresponding pixel RGB value. (c) Higher-order SH coefficients are predicted by a view-dependent appearance head, which takes two effective 3D-aware inductive biases, GVSRE and WCM, enabling more accurate modeling of view-dependent appearance.

Supervised Pose-required
Method	RealEstate10K			ACID
Method	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
pixelSplat ( Charatan et al., 2024 )	23.859	0.808	0.184	25.889	0.780	0.194
MVSplat ( Chen et al., 2024a )	24.012	0.812	0.175	25.561	0.775	0.195
DepthSplat ( Xu et al., 2025b )	25.595	0.852	0.145	$-$	$-$	$-$
YoNoSplat ( Ye et al., 2026 )	24.233	0.813	0.162	$-$	$-$	$-$

Table 1: Performance comparison of novel view synthesis on RealEstate10K and ACID datasets . We report the average metrics across all test scenes. The best and second-best results are highlighted.

-

indicates that the result was not reported in the original paper.

Figure 4: Qualitative results of Tab. 1 . The leftmost column shows the two-view context images. The top two rows are from RealEstate10K, and the bottom two are from ACID.

Supervised Pose-required
Method	ACID			DL3DV			ScanNet++
Method	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
pixelSplat	25.477	0.770	0.207	18.688	0.582	0.354	18.422	0.720	0.278
MVSplat	25.525	0.773	0.199	17.786	0.545	0.357	17.138	0.687	0.297
DepthSplat	26.012	0.791	0.185	19.553	0.611	0.285	20.775	0.760	0.254
YoNoSplat	24.246	0.721	0.222	19.636	0.594	0.311	21.075	0.744	0.254

Table 2: Cross-dataset generalization . All methods are trained on RealEstate10K and evaluated in a zero-shot setting on ACID, DL3DV, and ScanNet++.

Figure 5: Qualitative results of Tab. 2 . From top to bottom, the three rows show zero-shot results on ACID, DL3DV, and ScanNet++ when trained on RealEstate10K.

Figure 7: Qualitative results of Tab. 4 . GVSRE and WCM model view-dependent appearance variations (red boxes), while I2DC better preserves the base appearance component (yellow boxes).

Method	PSNR $\uparrow$	SSIM $\uparrow$	LPIPS $\downarrow$
Full	24.268	0.821	0.160
(1) w/o GVSRE	23.999	0.816	0.163
(2) w/o WCM	24.012	0.818	0.161
(3) I2DC only	23.610	0.808	0.173
(4) Baseline (NAS3R)	23.754	0.808	0.171
(5) Baseline (NAS3R) + I2DC	23.908	0.813	0.165

Table 4: Ablations. We evaluate the contribution of the proposed method on RealEstate10K.

Explore similar work

QuerySplat: Decoupling Geometry and Appearance Representations in 3DGS Prediction
Aug 2, 2026Yinglong Li, Donghui Shen, Xiaoyu Zhang +5Feed-Forward 3D Gaussian Splatting 3D Priors
State Key Laboratory of Virtual Reality Technology and Systems, Beihang University · InSpatio Research · State Key Lab of CAD&CG, Zhejiang University
ViewSplat: View-Adaptive 3D Gaussian Splatting for Feed-Forward Synthesis
Mar 26, 2026Moonyeon Jeong, Seunggi Min, Suhyeon Lee +1Feed-Forward 3D Gaussian Splatting Novel View Synthesis
University of Seoul · Korea Electronics Technology Institute
ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion
Jul 22, 2026In Cho, Jeonghwan Cho, Mijin Yoo +2Feed-Forward 3D Gaussian Splatting Feed-Forward 3D Gaussian
Yonsei University, South Korea · National University of Singapore, Singapore

Aug 2, 2026, cs.CVJ/K move · Enter open · S save

QuerySplat: Decoupling Geometry and Appearance Representations in 3DGS Prediction

Yinglong Li, Donghui Shen, Xiaoyu Zhang, Zhichao Ye+4

State Key Laboratory of Virtual Reality Technology and Systems, Beihang University · InSpatio Research · State Key Lab of CAD&CG, Zhejiang University

Feed-Forward 3D Gaussian Splatting 3D Priors Appearance

Open paper AskPDF arXiv

While feed-forward 3D Gaussian Splatting (3DGS) enables efficient 3D reconstruction, achieving high-fidelity rendering remains challenging. Existing pixel-aligned approaches suffer from spatial inflexibility and massive structural redundancy, whereas query-based methods lack 3D priors and entangle geometry with appearance, yielding blurry, pose-dependent results. To overcome these deficiencies, we propose \textbf{QuerySplat}, a feed-forward 3DGS framework driven by geometric priors and explicit appearance decoupling. Specifically, we design a dual-branch query-based decoder: the geometry branch leverages a pretrained Vision Geometric Model for spatial understanding, which intrinsically endows QuerySplat with pose-free modeling capabilities, while the appearance branch recovers high-frequency details through a dedicated pathway separated from geometric attribute regression. Extensive experiments demonstrate that QuerySplat mitigates the blurry rendering issues of early query-based models and consistently outperforms pixel-aligned approaches in rendering fidelity. On the challenging DL3DV benchmark, it achieves state-of-the-art novel view synthesis performance, with average PSNR gains of 2.30 dB and 1.04 dB over the best pose-free and pose-required baselines, respectively. Project Page: https://inspatio.github.io/querysplat.

Abstract

Explore similar work

QuerySplat: Decoupling Geometry and Appearance Representations in 3DGS Prediction

ViewSplat: View-Adaptive 3D Gaussian Splatting for Feed-Forward Synthesis

ATSplat: Compact Feed-forward 3D Gaussian Splatting with Adaptive Token Expansion