# Resolving Representation Ambiguity in Feedforward Novel View Synthesis Transformer via Semantic-Spatial Decoupling

arXiv: https://arxiv.org/abs/2605.18599 · HTML: https://mlsanity.com/paper/2605.18599/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 47 of its 62 references that are in the map.
- First posted on arXiv 18 May 2026.

## Builds on (most cited first, 30 of 47)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Perceptual Losses for Real-Time Style Transfer and Super-Resolution](https://mlsanity.com/paper/1603.08155/) (2016)
- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [pixelNeRF: Neural Radiance Fields from One or Few Images](https://mlsanity.com/paper/2012.02190/) (2020)
- [DUSt3R: Geometric 3D Vision Made Easy](https://mlsanity.com/paper/2312.14132/) (2023)
- [Objaverse: A Universe of Annotated 3D Objects](https://mlsanity.com/paper/2212.08051/) (2022)
- [Depth Anything V2](https://mlsanity.com/paper/2406.09414/) (2024)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://mlsanity.com/paper/2401.10891/) (2024)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [LRM: Large Reconstruction Model for Single Image to 3D](https://mlsanity.com/paper/2311.04400/) (2023)
- [Structured 3D Latents for Scalable and Versatile 3D Generation](https://mlsanity.com/paper/2412.01506/) (2024)
- [MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereo](https://mlsanity.com/paper/2103.15595/) (2021)
- [LGM: Large Multi-View Gaussian Model for High-Resolution 3D Content Creation](https://mlsanity.com/paper/2402.05054/) (2024)
- [IBRNet: Learning Multi-View Image-Based Rendering](https://mlsanity.com/paper/2102.13090/) (2021)
- [Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think](https://mlsanity.com/paper/2410.06940/) (2024)
- [Depth Anything 3: Recovering the Visual Space from Any Views](https://mlsanity.com/paper/2511.10647/) (2025)
- [pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction](https://mlsanity.com/paper/2312.12337/) (2023)
- [ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depth](https://mlsanity.com/paper/2302.12288/) (2023)
- [MVSplat: Efficient 3D Gaussian Splatting from Sparse Multi-View Images](https://mlsanity.com/paper/2403.14627/) (2024)
- [CAT3D: Create Anything in 3D with Multi-View Diffusion Models](https://mlsanity.com/paper/2405.10314/) (2024)
- [Depth Pro: Sharp Monocular Metric Depth in Less Than a Second](https://mlsanity.com/paper/2410.02073/) (2024)
- [GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting](https://mlsanity.com/paper/2404.19702/) (2024)
- [GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control](https://mlsanity.com/paper/2503.03751/) (2025)
- [MoGe: Unlocking Accurate Monocular Geometry Estimation for Open-Domain Images with Optimal Training Supervision](https://mlsanity.com/paper/2410.19115/) (2024)
- [Diffusion Transformers with Representation Autoencoders](https://mlsanity.com/paper/2510.11690/) (2025)

## Abstract (arXiv)

Transformer-based models have advanced feedforward novel view synthesis (NVS). Current architectures such as GS-LRM and LVSM mix semantic information (e.g., RGB) and spatial information (e.g., Pl"ucker rays) into a shared feature space. Since Pl"ucker rays naturally carry lattice-like spatial structure, these designs can make the spatial bias interfere with appearance representation and degrade rendering fidelity. To this end, we propose to decouple the representation of feedforward NVS transformers into separate semantic and spatial tokens. The decoupled design keeps semantic and spatial information explicit in their branches while preserving cross-branch interaction through shared attention routing. Built on this design, we introduce optional categorized supervision and bidirectional modulation: the former provides branch-specific training signals, while the latter improves interaction between the two branches. Notably, the base decoupled design introduces virtually zero additional inference latency due to its architectural design. The proposed designs achieve consistent improvements, demonstrating effectiveness across decoder-only and encoder-decoder feedforward NVS models.