This repository contains the dataset and Jupyter Notebook with the training code for the OrnaNet project, as presented in the paper "OrnaNet: cGAN-Based Ornament Predictions for 3D Concrete Printing". OrnaNet is a machine learning framework that uses a conditional Generative Adversarial Network (cGAN) to predict the final geometry of ornamental designs in 3D Concrete Printing (3DCP). The dataset consists of the complete set of processed image pairs used to train the model. These images were generated from the physically printed Tor Alva columns by pairing the geometric features of the digital print paths (input images) with the corresponding high-resolution 3D scans of the final objects (target images). The accompanying Jupyter Notebook provides the complete Python code, implementing a pix2pix architecture in TensorFlow, to train the model on this image-to-image translation task. For a full description of the methodology, data processing, and results, please refer to the associated publication.