Neural Radiance Fields (NeRF) is a method that solves the problem of depth and color estimation for the 3D reconstruction of objects by using different images. Using a neural network to sample and manipulate the 3D geometry of an object described as a spatial probability distribution instead of relying on traditional 3D models. The network acts as a volumetric function around a prior camera position and angle that predicts the density and color of each point in the scene. The model learns from images taken from different angles to produce increasingly realistic 3D reconstructions of the object that accurately represent its shape and surface texture. This paper explores the potential of NeRF for generating high-quality 3D models given a collection of images. The object is represented as a continuous signal and the network learns to predict the density and color of a point based on both its spatial coordinates and viewing angles. Here, we will follow the same approach as Matrix-NET based on a fully connected network that takes a 5D input consisting of the spatial coordinates (x, y, z) and the viewing angles \((\theta ,\phi )\) . The model is able to create very precise 3D models of the object that encompass its structure and surface features by training on images taken from multiple viewpoints.