-
1
Start with an adult-only fictional image or motion brief that avoids age ambiguity, real-person likeness, non-consensual framing, and unsafe context
-
2
Choose whether the project is text to image, image to image, text to video, or image to video before selecting a Wan model label
-
3
Keep public Wan references to verified route labels only: `wan2.7-image-pro`, `wan2.7-image`, `wan2.6-image`, and `wan2.6-i2v`
-
4
Describe subject, visual style, camera movement, duration, lighting, aspect ratio, continuity, and final use case in one focused prompt
-
5
Review X posts as market evidence only, then keep local generated assets and do not copy third-party media, checkpoint links, explicit prompts, or policy claims
-
6
Use prompt review to remove policy-circumvention wording, identity risk, explicit guarantees, age ambiguity, and policy-unclear details before saving variants