HSHiroyuki Sakai

Hiroyuki Sakai

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hiroyuki Sakai has a total of 9 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Metalog Materials

    by Hiroyuki Sakai, Saip-Can Hasbay, Christian Freude, Michael Wimmer, David Hahn

    A central aspect of modern physically based rendering is the practical yet realistic modeling of material appearance, which requires efficient and accurate representations of bidirectional reflectance distribution functions (BRDFs). In this paper, we present a unified material reflectance representation based on metalog distributions. Introducing quantile-based metalog interpolation and metalog mixture models, our representation captures a wide range of complex reflectance behaviors of both analytic models and real-world materials, while remaining interpretable and efficient. Our approach achieves competitive runtime performance and a small memory footprint, while enabling continuous interpolation between fundamentally different reflectance behaviors. Furthermore, our model is fully differentiable with respect to the coefficients defining the distributions, thus easily integrates into existing differentiable rendering pipelines. In summary, metalog materials can effectively represent real-world reflectance data, interpolate between multiple materials within a unified representation, and be acquired from real-world images via differentiable rendering.

  • Open Access

    Rendering Synthetic Defects for Learning‐Based Industrial Inspection

    by Runzhou Mao, Hiroyuki Sakai, Christian Freude, Christoph Garth, Petra Gospodnetić, Juraj Fulir

    Computer vision increasingly uses synthetic data from physically based rendering to supplement limited real‐world datasets. In industrial inspection, defect data is scarce and the rendering pipeline is explicitly controlled, and synthetic defect generation therefore becomes a dataset design problem. In this setting, building a dataset means choosing points in a rendering parameter space: defect shape, material, illumination, viewpoint, and sampling define the training distribution, but their effect on downstream learning is often hard to judge from images alone. We therefore study how these factors change defect features, where their relation to downstream learning is easier to inspect. Our results show that the rendering factors do not matter equally: defect shape, material, illumination, and viewpoint often affect downstream behavior much more than the number of samples per pixel. Synthetic subsets that transfer better downstream tend to stay close to real defect features, cover the observed defect modes, and stay separated from the defect‐free (OK) region. Based on these observations, we build a simple feature‐space screening heuristic for selecting subsets from large candidate pools. The selected subsets often outperform matched random selection for downstream segmentation on real data.

  • Open Access

    Statistical Error Reduction for Monte Carlo Rendering

    by Hiroyuki Sakai, Christian Freude, Michael Wimmer, David Hahn


  • Open Access

    A Statistical Approach to Monte Carlo Denoising

    by Hiroyuki Sakai, Christian Freude, Thomas Auzinger, David Hahn, Michael Wimmer