Generative AI (GenAI) has transformed visual effect design. Complex motion and depth cues effects that are traditionally difficult to produce can now be created with just text prompts. However, most GenAI tools focus on generating finished pixels, often leaving designers with static outputs that are difficult to modify or repurpose.AI-powered effect design tools should not just generate pixels, they should generate possibilities and unlock iterative explorations. In this talk, I will share some recent research on empowering 2D graphics designers to craft advanced motion and depth effects. I will show how AI-driven automation of key visual tasks can streamline effect authoring workflows without sacrificing creative control.