Reflectance Transformation Imaging (RTI) is a technique that provides an enhanced visualization experience. The current acquisition methods for Reflectance Transformation Imaging (RTI) are time consuming and computationally expensive. This work investigates the idea of getting best light positions for RTI acquisition using surface topography. We propose automating the RTI acquisition by estimating the surface topography using deep learning method followed by estimating light positions using unsupervised clustering method. This is one shot method which only needs one image. We also created RTI Synthetic dataset in order to carry out experiments. We found that surface topography alone is not sufficient to estimate best light positions for RTI without putting constraints.
Reflectance Transformation Imaging (RTI) is a non-invasive technique that enables the analysis of materials. Recent advancements in this technology, along with the availability of software for surface analysis through relighting, have improved the restoration and conservation of cultural heritage objects. However, there is a lack of appropriate benchmark data and reference light configurations, which makes it difficult to quantitatively compare and evaluate RTI data acquisitions. To address this, we have developed a dataset that can be used to assess the effectiveness of different surface light configurations for RTI acquisition. Additionally, we introduce methods to derive an ideal reference light configuration for a surface from its dense RTI acquisition. This dataset provides a standardized set of dense RTI acquisitions, accompanied by their corresponding reference light configurations that were obtained using our methods. This dataset can help researchers and developers to compare the performance of their approaches in solving the "Next Best Light Position" problem in RTI acquisition, which can ultimately improve the accuracy and efficiency of RTI acquisition and broaden its applicability in various fields.