Deep Spherical Gaussian Illumination Estimation for Indoor Scene

Image synthesis in visual machine learning

“…Illumination maps by Gardner et al [3] are over-simplified with a limited number of light sources, and the simplification loses accurate frequency information which results in unrealistic shadow and shading in rendering. Garon et al [2] and Li et al [63] regress illumination parameters but are often constrained by the order of representative functions (spherical harmonic and spherical Gaussian). As a result, they predict illumination of low frequency and produce renderings with very weak shade and shadow in rendering as illustrated in Fig.…”

“…Gardner et al [3] estimate the positions, intensities, and colours of light sources and reconstructs illumination maps with a spherical Gaussian function. On top of it, Li et al [63] represent illumination maps with multiple spherical Gaussian functions and regresses the corresponding Gaussian parameters for lighting estimation. Gardner et al [61] generate illumination maps directly with a two-steps training strategy.…”

“…Quantitative Evaluation: We compare NeedleLight with a number of stateof-the-art lighting estimation methods including Garon et al [2] that estimates lighting in frequency domain and Gardner et al [3,61] and Li et al [63] that estimate lighting in spatial domain. To perform quantitative evaluations, we create three spheres with gray diffuse, matte silver and mirror silver materials for evaluation as shown in Fig.…”

“…Thus Garon et al [2] struggles to regress accurate lighting position and recover high frequency information. Li et al [63] adopts spherical…”

“…In addition, we can observe that the performance of the state-of-the-art methods is not consistent under different evaluation metrics. For example, Li et al [63] outper- Qualitative Evaluation: We also compare NeedleLight with four state-of-the-art lighting estimation methods qualitatively. we design 25 3D scenes with objects for insertion and render them with the predicted illumination maps.…”