Simulation in the field of robotics is a less costly, safer, more scalable, and more versatile alternative to real-world scenarios. In this work, we present a pipeline for the digital reconstruction of environments using the panoptic segmentation technique . The overall aim is to automate the transcription of real geometries into the digital environment, creating a virtual scene where objects are meticulously segmented, making it possible to manipulate, replicate, remove, and practically move these objects. The methodology is based on Neural Radiance Field for segmentation and reconstruction of scenes, classical methods for manipulating structures, SDFormat as a scene description file, and Gazebo as a 3D robotic simulation platform . The results show that the method is capable of transcribing the scene into a virtual environment and using it as a space for testing, validating, and training robots.