Robotic simulators play a crucial role in agricultural robotics research, enabling preliminary validations, digital twin experiments, and AI-based sim-to-real methodologies. However, creating realistic farming environments in simulators is challenging and often requires significant manual effort. This paper addresses this challenge by proposing a semi-automated method for creating realistic tomato farm environments compatible with two popular simulators Gazebo and Unity. We conduct a comparative analysis of these simulators in terms of visual fidelity and integration effort, with a focus on their compatibility with the ROS 2 framework. Our contributions include the development of a realistic and parameterisable digital twin of a tomato farm, a detailed comparative analysis of Gazebo and Unity frameworks, and the publication of open-source code integrated into ROS 2. Through our study, we identify limitations in current digital twins, such as their static nature and limited realism, and suggest future enhancements, including the integration of physics for improved fidelity and the deployment of robots for practical simulated tasks. Our findings provide valuable insights for researchers and practitioners in the field of agricultural robotics.