PWZ AI Video Creation Guide

Open PWZ Studio, choose Video, and select a model. Switch to Text to Video to begin with a written prompt.

Include the subject, action, setting, and one clear camera direction. Add dialogue or environmental sound when it matters to the scene.

PROMPT /A cyclist stops beside a quiet coastal road at sunrise. The camera slowly moves closer as she looks toward the ocean. Soft wind and distant waves, natural cinematic lighting.

Choose the available duration, aspect ratio, and resolution. Check the credit amount shown on the generation button, then generate and review the result.

Select your model and choose Image to Video, or First / Last Frame in Wan 3.0. Choose the frame-control option available for that model.

Upload the opening image. To guide both ends of the shot, enable or select the first-and-last-frame option and add the ending image. Keep the subject’s appearance and scene layout compatible across both frames.

Explain how the subject and camera move between the images. Describe the visible change rather than repeating two static image descriptions.

PROMPT /Begin with the closed perfume bottle in the first image. The camera slowly arcs around it as warm light moves across the glass. End in the composition shown in the last image. Preserve the bottle shape and label.

Set the available duration and resolution, then generate. Seedance 2.0 supports video output up to 4K on PWZ. Select 4K when you need that output size and review the updated credit amount before generating.

Select a model and open Multi-Reference, Reference Media, or All-in-one Reference, depending on the generator.

Add the images, video clips, and audio your scene needs. Seedance generators also provide Virtual Portrait templates as a starting point for character references.

Use the reference names shown in the prompt field. Where available, type @ to select an uploaded file, then describe its purpose.

PROMPT /Use the selected portrait for the character’s appearance and the reference video for walking pace and camera movement. Place the character in a sunlit gallery. Keep the outfit consistent.

Add an audio reference only when it contributes to the scene, and explain whether it should guide a voice, rhythm, or sound quality.

Confirm that uploads are complete and within the model’s limits. Set the output options, review the credits, and generate.

Choose Seedance 2.5, open the multi-reference workspace, and upload the video you want to edit.

Turn on Video Edit. The output duration and aspect ratio adapt to the input video.

Describe the requested edit and the details to retain. Add relevant supporting references when needed, then generate.

PROMPT /Change the scene to warm evening lighting. Preserve the character’s appearance, walking action, camera movement, and scene layout.

Review both the requested change and the details that should remain stable. Results still need checking; a preservation instruction does not guarantee every detail stays identical.

Select Seedance 2.5, add the source video in the multi-reference workspace, and turn on Video Extend.

Continue from the action already underway. State the next movement, character reaction, or camera direction.

PROMPT /Continue as the character reaches the doorway, pauses, and looks back. Maintain the same clothing, warm lighting, and slow camera movement.

Select the available output duration, review the credit amount, and generate. Check the join for changes in appearance, movement, lighting, and sound.