Creation Guide

Choose the right starting point

Use text to image or text to video to create a scene from an idea. Use image editing or image to video when an existing picture should guide the result. Upload limits are shown in the studio.

Build a prompt from five parts

Subject, action, setting, light and camera. Make these agree before adding a style or constraints. A short clip benefits from one clear action rather than several competing scenes.

A white ceramic teacup on a wooden table, steam slowly rising. Soft morning window light from the left, a slow camera push forward, quiet natural colors, no text.

The MIT cookbook distinguishes action choreography from visual style descriptions. Start with a single action and a clear endpoint. Prompt constraints help communicate intent but do not guarantee adherence.

Prepare a reference image

Choose a sharp image with an unobstructed subject and even lighting. Include the body or object area needed for the action, remove distracting borders, and use material you are authorized to upload. Photo counts in source dance templates do not describe this studio's upload capabilities.

Check settings before submitting

Review the selected model, ratio, duration and estimated credits. The connected video tools currently use Seedance 1.0 Pro; references to Seedance 2.5 are not evidence that it is connected here. Review one output and change one variable at a time. A technically successful task with an unwanted composition is different from a failed task.

Finish your creation

Download a result and use your preferred editor for cuts, titles and music. The generation studio is not a full editing timeline, and describing sound does not make an unsupported audio feature available.

Further reading: MIT cookbook prompt patterns and photo preparation.