True Self-Supervised Novel View Synthesis is Transferable
Thomas Mitchel, Hyunwoo Ryu, Vincent Sitzmann
The key criterion for determining whether a models is capable of NVS is transferability, and we present the first fully geometry-free and self-supervised model capable of it.
Abstract
In this paper, we identify that the key criterion for determining whether a model is truly capable of novel view synthesis (NVS) is transferability: Whether any pose representation extracted from one video sequence can be used to re-render the same camera trajectory in another. We analyze prior work on self-supervised NVS and find that their predicted poses do not transfer: The same set of poses lead to different camera trajectories in different 3D scenes. Here, we present XFactor, the first geometry-free self-supervised model capable of true NVS. XFactor combines pair-wise pose estimation with a simple augmentation scheme of the inputs and outputs that jointly enables disentangling camera pose from scene content and facilitates geometric reasoning. Remarkably, we show that XFactor achieves transferability with unconstrained latent pose variables, without any 3D inductive biases or concepts from multi-view geometry — such as an explicit parameterization of poses as elements of SE(3). We introduce a new metric to quantify transferability, and through large-scale experiments, we demonstrate that XFactor significantly outperforms prior pose-free NVS transformers, and show that latent poses are highly correlated with real-world poses through probing experiments.
Presents XFactor, first geometry-free self-supervised model for transferable novel view synthesis without 3D inductive biases.
- Demonstrates transferability as key criterion for true novel view synthesis capability
- Achieves latent pose learning without explicit 3D parameterization or multi-view geometry concepts
- New metric quantifying transferability enables large-scale transfer evaluation
- Self-supervised learning
- Pose estimation
- Novel view synthesis
- Latent variable models
- Augmentation strategies
POSEENC restriction to stereo model precludes ultra-wide baseline pose estimation in single forward pass
from the paperRendering quality exhibits blurring and warping artifacts increasing as target poses diverge from context
from the paperModel is deterministic rather than generative, limiting its ability to resolve uncertainty
from the paper
Integrate recent advances in camera-controllable generative models to address rendering artifacts
from the paper
Author keywords
- Novel View Synthesis
- Self-Supervised
- Unsupervised
- Representation Learning
Related orals
Improving Diffusion Models for Class-imbalanced Training Data via Capacity Manipulation
Capacity manipulation improves diffusion models' handling of class-imbalanced data by reserving capacity for minority classes via low-rank decomposition.
Depth Anything 3: Recovering the Visual Space from Any Views
DA3 predicts spatially consistent 3D geometry from arbitrary camera views using plain transformer and depth-ray targets.
Text-to-3D by Stitching a Multi-view Reconstruction Network to a Video Generator
VIST3A stitches text-to-video models with 3D reconstruction systems and aligns them via reward finetuning for high-quality text-to-3D generation.
Radiometrically Consistent Gaussian Surfels for Inverse Rendering
RadioGS introduces radiometric consistency supervision for inverse rendering to accurately model indirect illumination in Gaussian-based representations.
Locality-aware Parallel Decoding for Efficient Autoregressive Image Generation
Introduces parallel decoding for autoregressive image generation with flexible ordering achieving 3.4x latency reduction.