The generator uses a focused two-photo workflow with fixed identity instructions, a ten-second duration, and 720p widescreen output. Review both source photos before spending a video from your pack because clear, similarly framed portraits produce the most stable pair.
Videos are added after a confirmed payment and never expire. You can render as soon as the pack is active, and each render uses one video from the pack.
A failed task does not consume a video. The task record links to the exact reservation, and any failure - including a refusal by the provider's content-safety or copyright check - restores the balance automatically. Only completed renders are charged.