The analysis of granulometry of substances is relevant in a great variety of the research and industrial applications. Unfortunately, image-based algorithms for granulometry analysis are difficult to tune and validate. In a typical setup, one or more cameras acquire images of a scene with a great number of objects or particles that must be measured. The distribution of the sizes/shape characteristics of the elements are the common output of such systems. The creation of supervised image database where the coordinates of all elements are known is very important and it allows for testing in a suitable manner the final size/shape distribution of particles produced by the image processing system with respect to the real distribution. Due to the great amount of objects or particles in the images, it is not often convenient, or even impossible, to individually measure each single objects in order to test the capability of the image processing system to locate and measure the elements (creation of a supervised dataset). One possible solution encompasses the creation of a synthetic image dataset where the position of each elements is known a priori. In this paper we propose a virtual environment to create and test an image-based granulometric system based on the 3D engine Blender. Results are encouraging and show the effectiveness of the proposed method.