This research explores the vast possibilities emerging from Machine Learning (ML) diffusion models and Text-to-Image processes in the fields of fabrication and material science. Pixel-based images can serve as a medium for information, drawing from large datasets on existing materials, their properties, and hybrid combinations. This enables the creation of novel material hybrids with diverse properties, leading to a new realm of material interfaces. The various stages and challenges of transitioning from 2D to 3D and 4D are examined through experiments, culminating in initial 3d prints used as moulds, interfaces, or substrates.