Before generation
An operator reviews uploaded photos, the site template and soundtrack, assigned roles, scene instructions and the permissions statement. Review considers the six prohibited categories, recognizable public figures, minors and rights concerns. A photo upload or consent checkbox alone is not content approval. Before any paid processing starts, the service submits the final generation instructions for each planned segment to Waffo’s prompt scanning API with semantic enforcement enabled. Only allow permits generation; review and block do not. Changed instructions require a new check.
Before delivery
Generated media remains private and held for output review. The operator checks the completed result for prohibited content and material identity, body, instrument or technical problems before delivery. Pending or rejected outputs are not opened for customer download.
What is automated and what is not
The site enforces file, setting, access and review-state checks. These are not equivalent to semantic image/video moderation. The server integrates Waffo prompt scanning for text instructions, alongside operator review and the generation provider’s safety controls. Text screening does not establish consent, copyright clearance, or photo/video safety. We do not claim that OpenAI moderation or a separate automated image/video moderation service is integrated.
Pending checks and service errors
A pending or degraded safety verdict pauses generation. Rate limits follow the provider’s Retry-After delay. Temporary service errors have bounded, delayed scan retries; other request errors require support attention. A safety retry never automatically repeats a paid generation. Blocked requests are not forwarded to an alternative model.
Rules and records
Input and output decisions, review reasons and timestamps are recorded. Prompt checks record the instruction hash, verdict, category/reason codes and Waffo check identifier without copying the prompt into the safety log. Operators apply the published Acceptable Use Policy, current provider restrictions and reported abuse patterns. Rules are revisited when templates, providers or safety risks change; a static keyword list alone is not the review process.
Enforcement
Requests can be refused, media access paused, prohibited files removed and further rendering restricted. A provider rejection is not a reason to bypass safety checks. For a paid result we cannot deliver, the original order is retained for support and refund handling.
Report a result
Contact luhuairong7@gmail.com with the relevant order number or page. Do not send payment-card details, passwords, API keys or identity documents. You may also use the Contact report form. Identify the page or order and describe the issue without resending harmful material. The report is saved for operator review. We do not operate a customer public gallery in the current build.
Appeal a decision
Explain why you believe a review or access decision is incorrect and include the reference. An operator reviews available evidence and may request limited additional information. Appeals do not guarantee approval and do not trigger a fresh paid generation without authorization.