<ul data-eligibleForWebStory="false"><li>During inference, every user prompt is transformed into a six-slot schema for generating high-fidelity prompts for Anna Fantasy's video diffusion backbone.</li><li>The six slots include Shot, Lighting, Subject + Action, Background, Camera Move, Style, and Mood, with a total length of 25–35 tokens and the flexibility to omit unnecessary slots.</li><li>To address issues like blurry faces, choppy motion, over-exposed skin, and unexpected elements, specific instructions are provided such as adjusting camera angles, altering lighting, and redefining scene elements.</li><li>The prompt writing process emphasizes visual intent, directing users to consider camera placement, lighting conditions, and framing ownership to guide Anna's handling of the scene during diffusion-time alchemy.</li></ul>

Crafting High-Fidelity Prompts for Anna Fantasy

Discover more