Lidar point cloud synthesis based on generative models offers a promising solution to augment deep learning pipelines, particularly when real-world data is scarce or lacks diversity. By enabling flexible object manipulation, this synthesis approach can significantly enrich training datasets and enhance discriminative models. However, existing methods focus on unconditional lidar point cloud generation, overlooking their potential for real-world applications. In this paper, we propose SG-LDM, a Semantic-Guided Lidar Diffusion Model that employs latent alignment to enable robust semantic-to-lidar synthesis. By directly operating in the native lidar space and leveraging explicit semantic conditioning, SG-LDM achieves state-of-the-art performance in generating high-fidelity lidar point clouds guided by semantic labels. Moreover, we propose the first diffusion-based lidar translation framework based on SG-LDM, which enables cross-domain translation as a domain adaptation strategy to enhance downstream perception performance. Systematic experiments demonstrate that SG-LDM significantly outperforms existing lidar diffusion models and the proposed lidar translation framework further improves data augmentation performance in the downstream lidar segmentation task.

Recent progress in novel view synthesis of indoor scenes using diffusion models has attracted significant attention, particularly for generating desired poses from a source image. Existing methods can produce plausible views near the input view, but they often fail to extrapolate views far beyond the input perspective. Additionally, achieving a multiview consistent diffusion model typically requires training of computationally intensive 3D priors, limiting scalability to long-range generation. In this paper, we present a transformer-based latent diffusion model that leverages view geometry constraints, including explicitly warped feature maps of the input view as the denoised target view and a conditioning combination of epipolar-weighted source image feature map, Plücker raymap, and camera poses. This approach allows for the extrapolation of consistent novel views, both semantically and geometrically, over long-range trajectories in a single-shot manner. Our model is evaluated on two indoor datasets, ScanNet and RealEstate10K, using a diverse set of metrics for view quality and consistency evaluation. Experimental results demonstrate the superiority of our approach over existing models, showcasing its potential for semantically and geometrically consistent novel view synthesis, scalable in video generation applications.

Novel view synthesis (NVS) from a single image is highly ill-posed due to large unobserved regions, especially for views that deviate significantly from the input. While existing methods focus on consistency between the source and generated views, they often fail to maintain coherence and correct view alignment across long-range or looped trajectories. We propose a model that addresses this by decomposing single-view NVS into a 360-degree scene extrapolation followed by novel view interpolation. This design ensures long-term view and scene consistency by conditioning on keyframes extracted and warped from a generated panoramic representation. In the first stage, a panorama diffusion model learns the scene prior from the input perspective image. Perspective keyframes are then sampled and warped from the panorama and used as anchor frames in a pre-trained video diffusion model, which generates novel views through a proposed spatial noise diffusion process. Compared to the prior work, our method produces globally consistent novel views-even in loop-closure scenarios, while enabling flexible camera control. Experiments on diverse scene datasets demonstrate that our approach outperforms existing methods in generating coherent views along user-defined trajectories. Our implementation is available at https://github.com/YiGuYT/LookBeyond.

The imbalanced distribution of different object categories poses a challenge for training accurate object recognition models in driving scenes. Supervised machine learning models trained on imbalanced data are biased and easily overfit the majority classes, such as vehicles and pedestrians, which appear more frequently in driving scenes. We propose a novel data augmentation approach for object recognition in lidar point cloud of driving scenes, which leverages probabilistic generative models to produce synthetic point clouds for the minority classes and complement the original imbalanced dataset. We evaluate five generative models based on different statistical principles, including Gaussian mixture model, variational autoencoder, generative adversarial network, adversarial autoencoder and the diffusion model. Experiments with a real-world autonomous driving dataset show that the synthetic point clouds generated for the minority classes by the Latent Generative Adversarial Network result in significant improvement of object recognition performance for both minority and majority classes. The codes are available at https://github.com/AAAALEX-XIANG/Synthetic-Lidar-Generation.