Start from text, one or two guiding frames, or a set of image, video, and audio references. Standard Wan 3.0 supports 2–30 second output at 480P, 720P, or 1080P with optional generated audio. The generator shows the current controls and credit estimate before submission.

Use a clear first frame when identity or product geometry matters, or assign dedicated image references in Reference to Video. Name the features that must remain stable, keep the first motion test controlled, and compare the opening, midpoint, and ending before adding more camera or styling changes.

Open Text to Video when the whole scene can come from a written direction. Choose Image to Video when one opening frame or a first-and-last-frame pair must control the look. Choose Reference to Video when images, motion clips, and audio cues need different jobs in the same Wan 3.0 brief. All three routes keep the model picker available.

Write the visible sequence in order: opening state, primary action, camera path, development, and ending state. Add lighting and style only after the subject and action are clear. For a clip near the 30-second limit, use a few connected beats rather than several unrelated locations or story lines, and place dialogue or sound cues beside the moment they should occur.

Use one frame when the subject, product angle, composition, or art direction must define the opening. Add a last frame when the shot also needs a specific destination, such as a transformation, reveal, transition, or planned edit point. Keep using text alone when Wan 3.0 is free to invent both the opening composition and the final state.

Upload only assets with a distinct purpose, then connect each one to the prompt using its @ token. For example, let @image1 define identity, @video1 demonstrate camera rhythm, and @audio1 set the timing of a visible action. Wan 3.0 accepts up to 10 images, 5 videos, and 5 audio clips, but a smaller set with explicit roles is easier to direct and review.

Standard Wan 3.0 supports 2–30 seconds and 480P, 720P, or 1080P. Use a short duration to validate one action or camera idea, then extend the clip when the brief genuinely needs development. Choose the delivery frame before generating: vertical layouts need tighter central staging, while widescreen layouts can support more environment and lateral movement.

Confirm the input mode, selected Wan model, duration, resolution, aspect ratio, and audio setting. Check that every frame or multimodal reference has an explicit role and that the prompt describes a sequence the chosen duration can hold. The generator displays the current credit estimate before submission, so verify the setup before spending the task.

Watch once for whether the intended scene reads, then inspect subject identity, hands or product geometry, camera path, background stability, lighting continuity, ending state, and audio timing. Compare the result with the original prompt and references. If one defect blocks use, revise the instruction or source most likely to cause it instead of replacing the whole brief.

Open My Creations to review task status, revisit completed videos, compare useful versions, and download a selected result. For production work, also keep the prompt, selected model, duration, resolution, aspect ratio, audio choice, source filenames, and the single change tested in each version. That record makes a successful direction easier to reproduce.