Minimal parametrization of 3D lines plays a critical role in camera localization and structural mapping. Existing representations in robotics and computer vision predominantly handle independent lines, overlooking structural regularities such as sets of parallel lines that are pervasive in man-made environments. This paper introduces RiemanLine, a unified minimal representation for 3D lines formulated on Riemannian manifolds that jointly accommodates both individual lines and parallel-line groups. Our key idea is to decouple each line landmark into global and local components: a shared vanishing direction optimized on the unit sphere, and scaled normal vectors constrained on orthogonal subspaces, enabling compact encoding of structural regularities. For n parallel lines, the proposed representation reduces the parameter space from 4n (orthonormal form) to 2n+2, naturally embedding parallelism without explicit constraints. We further integrate this parameterization into a factor graph framework, allowing global direction alignment and local reprojection optimization within a unified manifold-based bundle adjustment. Extensive experiments on ICL-NUIM, TartanAir, and synthetic benchmarks demonstrate that our method achieves significantly more accurate pose estimation and line reconstruction, while reducing parameter dimensionality and improving convergence stability.

The growing demand for immersive 3D content calls for automated monocular-to-stereo video conversion. We present a controllable, direct end-to-end method for upgrading a conventional video to a binocular one. Our approach, based on (conditional) latent diffusion, avoids artifacts due to explicit depth estimation and warping. The key to its high-quality stereo video output is a novel, guided VAE decoder that ensures sharp and epipolar-consistent stereo video output. Moreover, our method gives the user control over the strength of the stereo effect (respectively, the disparity range) at inference time, via an intuitive, scalar tuning knob. Experiments on three different datasets of real-world stereo videos show that our method outperforms both traditional warping-based and recent warping-free baselines and sets a new standard for reliable, controllable stereo video conversion.

EgoCross is a cross-domain egocentric video question answering benchmark designed to evaluate whether multimodal large language models can generalize beyond common daily-life scenarios. The first EgoCross Challenge was hosted at the Third EgoVis Workshop at CVPR 2026 and evaluated models on first-person videos from four target domains: surgery, industrial assembly, extreme sports, and animal perspectives. Each test example consists of an egocentric video clip, a question, and four candidate answers, from which the model must select the correct option. This technical report introduces the challenge task, benchmark resources, and two official Codabench tracks. The Source-Limited Track restricts participants to the official baseline model and a small support set, whereas the Open-Source Track permits broader choices of models and training data under rules that prohibit the manual construction of target-domain training data. In total, the challenge received more than 1,500 submissions from over 130 participants, with 19 teams participating in the Open-Source Track and 38 teams in the Source-Limited Track. We further present the official leaderboard results and summarize the winning solutions from both tracks. We hope that this report will serve as a useful technical reference for advancing cross-domain egocentric video understanding. All resources, including the challenge data, baseline implementation, and code released by the winning teams, are made publicly available.

For practical use, diffusion- or flow-based generative models must be aligned with task-specific rewards, such as prompt fidelity or aesthetic preference. That alignment is challenging because the reward is defined for clean output images, but the alignment procedure requires value function estimates at noisy intermediate latents. Existing methods resort to Tweedie-style or Monte Carlo approximations, trading off estimator bias against computational cost: Tweedie estimates are efficient but biased, while Monte Carlo estimates are more accurate but require expensive rollouts. A natural alternative would be a learned value function, but it remains an open question how to effectively train a strong and general value model specifically for noisy latents. Here, we propose StitchVM, a model stitching framework that efficiently transfers reward models pretrained for clean images to the noisy latent regime. StitchVM starts from an existing, truncated pixel-space reward model and attaches a frozen diffusion backbone to it as its head. From the pixel-space model, the resulting hybrid retains a carefully pretrained, robust reward capability; from the diffusion backbone, it inherits its native ability to handle noisy latents. The stitching procedure is exceptionally lightweight, e.g., stitching and finetuning CLIP ViT-L and SD 3.5 Medium takes only 10 GPU-hours. By lifting powerful pixel-space reward models to latent space, StitchVM opens up a new style of diffusion alignment: instead of rough, yet costly per-sample approximation of the value function, the correct function for the actual, noisy latents is constructed once and then amortized over many samples and iterations. We show that this approach yields improvements across a broad range of downstream steering and post-training methods: DPS becomes 3.2× faster while halving peak GPU memory, and DiffusionNFT becomes 2.3× faster.

Self-supervised depth estimation from monocular sequences relies on the joint learning of a depth and a pose network. Despite abundant research done to improve the depth network, efforts on the pose remain limited. In this context, even when depth is estimated up to scale, we highlight the importance of the alignment between the scene scales estimated by the pose and depth nets. Then, we introduce SA4Depth, an approach to improve this alignment and boost the depth predictions while keeping the inference time unchanged. Our proposed method uses the depth estimated during training to reproject learnable visual features across consecutive frames and refine the pose estimates by reducing feature alignment residuals. With our method, the estimated scene scales by the separate depth and pose networks are aligned, and the prediction scale consistency is improved across different sequences. Our differentiable refinement integrates seamlessly into existing self-supervised pipelines and substantially improves their depth estimates. We demonstrate this with extensive experiments both outdoors and indoors on KITTI, Cityscapes, and NYUv2. Additionally, results on KITTI Odometry confirm the effectiveness of our pose refinement.

Recent advances in generative AI allow users to create 3D models from text or images. However, these models prioritize visual plausibility over geometric accuracy, often generating results with flaws that compromise their intended use post-fabrication. We present InstructMesh, an interactive post-generation refinement tool that enables selective repair of generative 3D models through region selection and targeted operations, such as opening or sealing voids, or adjusting local thickness. Users can invoke edit operations via natural language prompts or slider controls. By operating directly on the intermediate latent representation, InstructMesh allows users to apply robust geometric corrections without requiring expert modeling skills. To inform our design, we first analyze common fabrication-related failure modes in outputs from state-of-the-art generative tools. We then conduct two user studies, demonstrating that novices can identify and perform fabrication-relevant repairs on generative outputs using InstructMesh, and revealing user preference for hybrid interfaces that combine slider controls with natural language input.

Generative novel view synthesis from sparse input images is rarely all reconstruction or all generation: pixels visible in some source view have a unique correct value modulated only by view-dependent shading, while pixels in disocclusions or beyond the captured volume admit a distribution of plausible completions. Existing generative novel-view-synthesis methods conflate these regimes under a single uniform loss, blurring the line between geometric fidelity and creative hallucinations even when scene geometry is injected through warped point clouds or projected depth. We introduce GenRec, a multi-view flow matching model that builds the reconstruction–generation split directly into its architecture, supervision, and gradient flow. Guided by an observation mask derived from the source cameras and a monocular depth estimator, a flow matching backbone jointly denoises RGB and scene-coordinate maps across all target views, while a pixel-space refinement stage restores high-frequency detail on observed pixels; the same mask gates supervision so regression signals do not contaminate the generative prior. Across RealEstate10K, DL3DV-10K, and Mip-NeRF 360, in both single-view extrapolation and two-view interpolation, GenRec attains the best reconstruction fidelity in observed regions while also surpassing purely generative baselines on perceptual quality in unobserved ones, showing the effectiveness of our approach.

The rapid progress of large, pretrained models for both visual content generation and 3D reconstruction opens up new possibilities for text-to-3D generation. Intuitively, one could obtain a formidable 3D scene generator if one were able to combine the power of a modern latent text-to-video model as "generator" with the geometric abilities of a recent (feedforward) 3D reconstruction system as "decoder". We introduce VIST3A, a general framework that does just that, addressing two main challenges. First, the two components must be joined in a way that preserves the rich knowledge encoded in their weights. We revisit model stitching, i.e., we identify the layer in the 3D decoder that best matches the latent representation produced by the text-to-video generator and stitch the two parts together. That operation requires only a small dataset and no labels. Second, the text-to-video generator must be aligned with the stitched 3D decoder, to ensure that the generated latents are decodable into consistent, perceptually convincing 3D scene geometry. To that end, we adapt direct reward finetuning, a popular technique for human preference alignment. We evaluate the proposed VIST3A approach with different video generators and 3D reconstruction models. All tested pairings markedly improve over prior text-to-3D models that output Gaussian splats. Moreover, by choosing a suitable 3D base model, VIST3A also enables high-quality text-to-pointmap generation.

Large Vision-Language Models (LVLMs) map visual inputs into dense token sequences, imposing a quadratic computational bottleneck for inference. Elastic visual-token compression addresses this by training a single model that can run at multiple visual-token budgets. However, existing approaches struggle under aggressive compression. Spatial-only compression, as in nested pooling, behaves as an imperfect low-pass filter and induces spectral aliasing that obscures fine-grained detail. Query-only compression, as in nested query resampling, replaces explicit grid-aligned tokens with non-local summaries and substantially degrades spatial grounding. To resolve this representational conflict, we introduce PARCEL (Pool-Anchored Resampling with Conditioned Elastic Queries for Efficient Vision-Language Understanding), a visual tokenization architecture that dynamically partitions the labor of feature extraction. PARCEL establishes spatial pool tokens as low-frequency layout anchors and conditions elastic query tokens on these anchors through Pool-Conditioned Query Resampling. This encourages query tokens to focus on complementary visual features rather than redundant spatial mapping. Extensive evaluations across 27 benchmarks show that PARCEL improves the performance-efficiency Pareto frontier, consistently outperforming existing matryoshka baselines across visual-token budgets while preserving the "train once, deploy anywhere" paradigm.

Forecasting the evolution of dynamic environments is crucial for autonomous agents. While generative world models have recently achieved high photorealism in 2D video synthesis, by mixing within the image plane ego-motion and environmental dynamics, they exhibit physical inconsistencies, such as morphing or vanishing objects, especially over long time horizons. In this paper, we propose FR3D, a world model that predicts a persistent 3D latent representation for future dynamic 3D reconstruction. Unlike prior works that treat the world as a sequence of image-based features, FR3D explicitly decouples the 3D evolution of the scene from the agent's trajectory, treating the inferred ego-motion as a latent proxy for action. This disentanglement resolves the ambiguities between self-motion and world-motion, ensuring geometric consistency into the future. Furthermore, we introduce a teacher-student distillation strategy that leverages the spatial "common sense" of off-the-shelf foundation models, leading to robust zero-shot generalization. Extensive experiments demonstrate FR3D's strong performance for future dynamic 3D reconstruction from monocular observations across multiple datasets, even 2 seconds into the future.

Neural Radiance Fields and Gaussian Splatting have emerged as powerful representations for reconstructing complex dynamic urban scenes from multi-view driving logs. By producing photorealistic and geometrically consistent renderings, these methods offer a foundation for closed-loop simulation as well as scalable data augmentation engine, enabling the synthesis of diverse viewpoints and conditions without the need for costly additional data collection in the real world. However, existing approaches are almost exclusively tailored to well-lit, daytime environments and struggle with the challenges of nighttime settings, where noise, strong flares, and moving light sources dominate the visual signal. To address these limitations, we propose NightSplat, a Gaussian Splatting framework specifically designed for high-fidelity reconstruction and novel view synthesis of nighttime driving scenes. Through dedicated components, our method explicitly models sensor noise, lens flare effects, and introduces a lightweight module to represent dynamic vehicle lights. Extensive evaluations on nuScenes and Waymo Open Dataset demonstrate that NightSplat significantly outperforms prior state-of-the-art baselines, both quantitatively and qualitatively, thereby extending simulation and testing to nighttime scenes.

Visual geometry transformers have become powerful architectures for multi-view 3D reconstruction, enabling joint prediction of multiple 3D attributes in a feed-forward manner. However, their computational cost grows quadratically with the input sequence length due to the global attention layers inside these models. This limits both their scalability and efficiency. In this work, we address this challenge with a simple yet general strategy: restricting the number of key/value tokens that each query interacts with during global attention. To achieve effective token selection, we introduce a two-stage framework. First, an inter-frame selection step operates at the frame level to identify frames that should be preserved. Second, an intra-frame selection step further discards more redundant tokens within the selected frames. Our analysis highlights the advantage of a diversity-based strategy for inter-frame selection, which ensures broad coverage of the scene. For intra-frame selection, we show that layer-aware sparsification is necessary, with the selection process guided by the entropy of the global attention pattern. Our approach offers a superior speed-accuracy trade-off compared to existing solutions. Extensive experiments show that it accelerates visual geometry transformers by over 85

We address the problem of 4D reconstruction from partial point cloud sequences, where observations from depth sensors are incomplete, unordered, and lack explicit point correspondence over time. Recovering coherent 4D geometry in this geometry-only setting is challenging due to missing observations and ambiguous dynamics. While recent progress has largely been driven by image-based methods, existing point-based approaches typically focus on single-object scenarios, assume relatively complete inputs, and rely on explicit correspondence. To mitigate these limitations, we propose DynaTok, a point-based framework for correspondence-free 4D reconstruction from partial point cloud sequences that operates without images. DynaTok encodes each frame into compact latent tokens, aggregates incomplete observations over time with a Transformer-based spatiotemporal encoder, and decouples geometry and motion via a residual token design within a single unified model. Conditioned on the aggregated tokens, a point flow-matching decoder reconstructs complete and temporally consistent 4D point cloud sequences using only point cloud supervision. Experiments on object-level and scene-level benchmarks demonstrate improved reconstruction quality and temporal coherence under partial point cloud observations.

Object pose estimation using visual data is crucial for robotic interaction with the environment. Many existing instance-level methods are restricted by their requirements for 3D CAD models or multiple object views, which limits their flexibility and generalizability. Overcoming this limitation is critical to enhance the adaptability of pose estimation systems. In this work, a novel pipeline that leverages recent advances in reconstruction techniques is presented to address these challenges. To this end, Large Reconstruction Models (LRM) represent an advanced neural architecture capable of generating 3D object models from a limited set of views. Nevertheless, the resulting 3D models often lack relevant geometric and texture details due to insufficient input information. This research presents InstantPose, an innovative zero-shot instance-level pose estimation method that, building upon LRM, can determine the pose of unseen objects using as little as a single unposed RGB reference and RGB-D query images. Extensive experiments demonstrate that InstantPose achieves remarkable performance in object pose estimation on the YCB-V dataset, compared to methods conceived to rely on a geometrically perfect object's model. Furthermore, the 6D pose provided through the presented approach facilitates successful object grasping, highlighting its practical utility in robotic manipulation tasks.

State-of-the-art single-image 3D reconstruction methods often rely on complex hybrid architectures or necessitate compressing geometry into latent spaces to leverage pre-trained latent diffusion models. In this work, we demonstrate that such architectural overhead is unnecessary. We introduce a minimalist pixel-space Diffusion Transformer built on a plain ViT, which operates directly on raw point map patches and is conditioned on image tokens from a pre-trained DINOv3. Unlike existing latent diffusion-based approaches, we train our diffusion backbone entirely from scratch, eliminating the need for point map tokenizers. We show that this streamlined approach yields results superior to complex latent-based diffusion models while remaining significantly simpler than hybrid alternatives. Notably, our model produces sharper geometric structures and achieves significantly better results on highly ambiguous regions, such as transparent objects.

Open-vocabulary 3D segmentation enables the exploration of 3D spaces using free-form text descriptions. Existing methods for open-vocabulary 3D instance segmentation primarily focus on identifying object-level instances in a scene. However, they face challenges when it comes to understanding more fine-grained scene entities such as object parts, or regions described by generic attributes. In this work, we introduce Search3D, an approach that builds a hierarchical open-vocabulary 3D scene representation, enabling the search for entities at varying levels of granularity: fine-grained object parts, entire objects, or regions described by attributes like materials. Our method aims to expand the capabilities of open vocabulary instance-level 3D segmentation by shifting towards a more flexible open-vocabulary 3D search setting less anchored to explicit object-centric queries, compared to prior work. To ensure a systematic evaluation, we also contribute a scene-scale open-vocabulary 3D part segmentation benchmark based on MultiScan, along with a set of open-vocabulary fine-grained part annotations on ScanNet++. We verify the effectiveness of Search3D across several tasks, demonstrating that our approach outperforms baselines in scene-scale open-vocabulary 3D part segmentation, while maintaining strong performance in segmenting 3D objects and materials.

Several disruptive research directions have recently emerged in computer vision, including foundation models achieving previously unseen zero-shot performance in scene understanding, even interactively, and generative models that synthesize extremely realistic images. The latter have also been shown to be highly effective in scene understanding tasks thanks to their rich priors. However, for promptable segmentation, foundation models struggle with accurately segmenting an object's region, leading to false positives and over-segmentation. Notably, early attempts that leverage generative priors use prompts only during post-processing, yielding suboptimal segments because the process is agnostic to the user input. In this paper, we target these limitations with Prompt2Seg, a spatial conditioning framework for diffusion-based segmentation. Prompt2Seg augments a frozen diffusion segmentation model with a conditioning branch. Our approach takes spatial prompts, represented as 2D Gaussians or confidence maps, as explicit input signals, training the model to respond directly to user intent. Fine-tuned on a deliberately constrained set of object categories drawn from Hypersim and Virtual KITTI 2, Prompt2Seg generalizes zero-shot to a wide range of unseen object types and visual domains. We evaluate on seven datasets ranging from standard benchmarks to more challenging domains, including paintings, egocentric views, and X-ray data. Furthermore, we demonstrate that Prompt2Seg consistently outperforms the underlying diffusion segmentation backbone across all benchmarks. Our results suggest that the rich priors encoded in generative pretraining, combined with principled spatial conditioning, offer a compelling path toward broadly generalizing interactive segmentation without large-scale mask supervision.

Current Large Reasoning Models (LRMs) exhibit remarkable general capabilities but significantly underperform in spatial reasoning tasks. Existing approaches treat this gap as a knowledge deficit, relying on supervised fine-tuning (SFT) to ingest labeled data from external vision sources or synthetic engines. In contrast, we argue that for many tasks, spatial capabilities are already present in pre-trained LRMs but require alignment through principles of internal logical coherence. In this work, we propose a self-supervised reinforcement learning (RL) framework that targets the internal Chain-of-Thought (CoT) process without requiring ground-truth annotations. By formalizing the notion of consistency verifiers—reward functions that check for geometric and semantic consistency under transformations like flipping or swapping the order of objects in the question—and optimizing them via our new OT-GRPO strategy, a minimal-consistency matching variant of group relative policy optimization, we demonstrate that models can self-correct their spatial logic. Our results show that this label-free consistency training approaches the accuracy of models trained with ground-truth supervision and achieves similar generalization across diverse tasks and domains.

Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric experiments to improve VLM training. As part of DCVLM, we collect 160 datasets spanning four data types – image-caption pairs, multimodal interleaved documents, text-only, and instruction-tuning data – into a corpus of 6T multimodal tokens. DCVLM allows participants to test curation strategies (filtering, mixing, formatting, sampling) across 1B-8B models and 6.25B-200B token budgets. Models are then evaluated on a carefully selected suite of up to 52 downstream benchmarks across 9 domains. We conduct extensive experiments on DCVLM and find that data mixing, not filtering, is key to a high-quality training dataset: instruction-heavy mixtures scale better than caption-heavy ones, with gains widening at larger scales. The resulting dataset, DCVLM-Baseline, enables training an 8B VLM to 63.6

We present Layered Ray Intersections (LaRI), a fully supervised method for occluded geometry reasoning from a single image. Unlike conventional depth estimation, which is limited to visible surfaces, LaRI predicts multiple surfaces intersected by the camera rays using layered point maps. Compared to the existing approaches that leverage neural implicit representations or iterative refinement, LaRI achieves complete scene reconstruction in one feed-forward pass, enabling efficient and view-aligned geometric reasoning to underpin both object-level and scene-level tasks. We further propose to predict the ray stopping index, which identifies valid intersecting pixels and layers from LaRI's output. To better underpin and evaluate this task, we build an annotation pipeline using rendering engines, construct annotations for five public datasets, including synthetic and real-world data covering 3D objects and scenes. As a generic method, LaRI's performance is validated in object-level and scene-level reconstruction tasks.