The goal of this PhD project is to investigate the potential of generative models in 3D for creating high-quality, realistic objects suitable for various applications. While significant progress has been made in text-to-image synthesis using diffusion models trained on billions of image-text pairs, applying this technique to 3D synthesis requires extensive labeled 3D asset datasets and efficient denoising architectures for 3D data, which are currently lacking. To address this gap, the project will focus on developing techniques to transfer pretrained 2D image-text diffusion models to 3D object synthesis. The ultimate objective is to contribute to the advancement of generative models in 3D, capable of producing complex and lifelike objects with high fidelity.