Text to Video: The Full Workflow
Last updated: November 2026 · 4 min read
Text-to-video generation turns a written description into a generated clip. The available models, audio options, durations, output formats, and controls can vary; use the application to see what is currently supported for your account.
Review available video options
Depending on availability, video workflows may produce a clip with audio or a silent clip:
- Video with audio where this option is available, the generated output may include audio as part of the result. Review its settings and cost before submitting.
- Silent video where available, a video can be generated without audio so you can work with sound separately using an available audio or music workflow.
Credit cost depends on the selected workflow and parameters. The applicable cost is shown or confirmed when you submit a request.
Write a one-line storyboard
Video prompts fail most often because they are static image prompts with "moving" pasted in. Instead, describe what happens in one sentence with a clear subject, action, and camera behavior.
Slow dolly-in on a lighthouse keeper climbing the spiral stairs,
lantern light flickering across weathered walls, dust motes drifting.Reserve the image-style blocks (medium, lighting, mood) for a second sentence. Leading with motion keeps the model anchored on the action.
Motion scale and duration
If motion controls are available for the selected model, use them to explore how much movement appears in the clip. This is general prompting guidance, not a guarantee of output:
| Motion scale | Character | Best for |
|---|---|---|
| Low | Subtle drift, gentle parallax | Product shots, portraits, atmospheric loops |
| Medium | Natural movement, slow camera | Narrative scenes, b-roll, explainer visuals |
| High | Dynamic action, fast cuts | Sports, action, transitions, trailers |
Clip duration and generation limits vary by model and account. For a longer story, you can plan separate shots and edit them together using your preferred video editor.
Character and scene consistency
Keeping a subject consistent between separate generated clips can be difficult. If the current workflow supports seeds or reference media, these may help, but results can still differ:
- Keep descriptions consistent. Reuse the same clear details for a subject, setting, and visual style.
- Use reference options when available. If a selected workflow accepts reference media, follow its in-app instructions and only upload material you have the right to use.
Syncing audio
If you want to add sound to a silent clip, use an available audio or music tool or an editor of your choice. See timing audio to video for general editing guidance. In-app editing options depend on the current service.
Exporting and publishing
After generation, review the result and use the download or sharing actions available for that creation. Before publishing, check the visibility setting and make sure you have the rights and permissions needed for the content.
Troubleshooting
- Morphing subject: lower the motion scale, shorten the clip, or anchor with an image start frame.
- Physics glitches: simplify the action sentence; one verb per clause reads better to video models.
- Rejected prompts: review the Terms and Help Center. Any credit return is handled under the in-product policy and is not a cash refund.