Distillation in AI extends beyond simple text imitation to more complex domains. Text distillation is straightforward: a large language model produces a distribution over the next token or a complete response, and a smaller model is trained to imitate that behavior, such as learning to say "Paris" or replicate the shape of an explanation. Diffusion distillation, however, is fundamentally different. A diffusion model does not emit an image in one clean forward pass; instead, it starts from noise and repeatedly edits that noise until a coherent sample appears. Generation is a trajectory, not an answer, making the model less like a database query and more like a sculptor taking dozens of tiny cuts. Consequently, diffusion and multimodal distillation must compress trajectories, distributions, motion, and the semantic geometry between different worlds, rather than merely imitating a single answer.