Synthetic Data News
The latest Synthetic Data coverage curated for game developers.
arXiv cs.GR details NaRPA, a ray-tracing pipeline built to generate virtual space imagery for navigation and sensor testing. The framework models passive and active vision sensors, plus a velocimeter LiDAR, …
SyntheticDoc brings 1,000,000 procedurally generated, high-resolution training images to document unwarping and illumination correction. The dataset pairs each sample with pixel-perfect UV, normal, albedo, and shading maps, aiming to replace …
Researchers have built an Unreal Engine pipeline that generates action-conditioned video for world-model pretraining, pairing control inputs with rendered scene changes. The system produced 2,691 hours of 1080p footage and …
A new framework cuts the cost of building urban segmentation training data by turning rough synthetic scenes into target-aligned images. It adapts an off-the-shelf diffusion model with only imperfect pseudo-labels, …
WorldRover turns long-horizon world exploration into a synthetic data problem, using Unreal Engine to render minute-scale traversals with full trajectories and scene geometry. The resulting 10M-sequence dataset pairs RGB with …
Researchers have proposed a meshless domain-randomization pipeline built on 3D Gaussian Splatting, aimed at closing the sim-to-real gap for complex organic assets. Instead of extracting difficult textured meshes, the method …
TexSketch is a procedural pipeline for generating colored sketch datasets with texture-aware stylization, aimed at training sketch colorization models without hand-annotated pairs. It combines region extraction, semantic color prediction, and …
A diffusion pipeline is being used to synthesize sand-boil inspection imagery for earthen levees when real defect annotations are scarce. The system fine-tunes Stable Diffusion XL with DreamBooth, then steers …
Researchers synthesized 48,000 vision-language-kinematics trajectories from reconstructed indoor scenes to train a humanoid policy without any human demonstration data. The interesting part for developers is the sim-to-real pipeline: 3D Gaussian …
arXiv cs.GR details a stereo-rendering quirk that can leak a learnable cue into disparity training. In path-traced synthetic pairs, variance fields line up across views far more than expected, and …
A new hyperspectral-multispectral fusion model, SCALMU, blends classical CNMF structure with learned multiplicative updates to recover high-res hyperspectral images. The big practical angle for graphics and imaging teams is that …
A new data-engineering pipeline aims to make synthetic augmentation more reliable in low-data vision tasks. The system pairs controllable diffusion generation with staged filtering, optional uncertainty-based selection, and human verification. …
Unreal Engine 5 is emerging as the preferred platform for robotics simulation, transforming how robotics teams develop and test their systems. This shift allows developers to leverage UE5 not only …
A new generative framework allows for large-scale synthetic household data generation, crucial for developing interactive AI agents. This tool enables developers to create diverse datasets that reflect real-world interactions, significantly …
A new pipeline for single-view 3D reconstruction of ships could revolutionize maritime monitoring. By leveraging synthetic data, developers can achieve real-time visualization without the need for extensive multi-view setups. This …
A new unsupervised training strategy for hyperspectral image super-resolution could revolutionize how developers handle remote sensing data. By utilizing synthetic abundance data, this method eliminates the need for ground truth …
A new system for hand-object tracking leverages synthetic data, eliminating the need for human demonstrations. This advancement is particularly relevant for programmers and designers focused on manipulation mechanics, as it …
A new deep learning approach for camera calibration using synthetic data could significantly streamline workflows for graphics programmers. By leveraging a comprehensive dataset, this method allows for accurate parameter prediction …