查看文章

The crucial step in the creation of every new product is the design process which is highly subjective and time-consuming. Artificially generated design suggestions can help to accelerate this process. We propose a method which can automatically generate 2D images with different design ideas for the specific design nodes. Images are generated by the DCGAN network. The DCGAN model is trained using real data in combination with synthetically generated images. Synthetic data is created using the generative design module of PTC Creo Parametric. In the paper, we train the DCGAN model to generate bicycle frame images. After some improvements to the model, it can become a helpful tool in the product design process.