This project develops high-quality 3D scientific visualizations of molecular and biomedical mechanisms for research communication, scientific presentations and dissemination. The visualizations are designed to translate complex molecular processes into scientifically accurate dynamic representations, including molecular recognition, targeted delivery, catalytic activation, biodistribution and cellular responses. The workflow combines Python-based procedural scene generation with Blender for 3D modelling, animation and physically based rendering. Scientific constraints are incorporated directly into the construction and animation of molecular structures to maintain consistency between the underlying biological mechanism and its visual representation. Development and validation are performed locally using low-resolution preview rendering. Once geometry, materials, animation and scientific accuracy have been validated, high-quality frames are rendered using Blender Cycles. These final renders are computationally demanding because of complex molecular and biological geometries, volumetric and optical effects, physically based materials, depth of field and high sampling requirements. The requested GPU resources will therefore be used primarily for final rendering and selected high-quality validation renders. Individual animation frames are independent and can be distributed efficiently across GPU jobs, making the workload particularly suitable for parallel execution on the Arrhenius GPU infrastructure.