GenX.art DokumentationText to Video
Generate video from a written scene description
Create a scene from text
Text to Video starts without source media. You describe the subject, action, environment, camera behavior, lighting, and constraints; the selected model interprets that brief into a video.
Workflow
Choose a model
Compare the available provider, supported duration, quality, aspect ratio, audio behavior, and estimated credits.
Write the prompt
Describe one coherent shot first. A practical order is: subject → action → environment → camera → light and mood → exclusions.
Adjust compatible controls
Only use the duration, aspect ratio, quality, or audio controls shown for the selected model. Switching models can reset incompatible values.
Review and submit
Confirm the model, prompt, settings, and credit estimate. Keep the task in creation history while it processes.
Prompt example
A ceramic cup on a dark wooden table, steam curling upward as morning light crosses the frame, slow push-in camera, quiet editorial mood, no text or logos.
Improve consistency
- Prefer one subject and one primary action per short shot.
- State camera movement explicitly only when it matters.
- Remove contradictory style, lighting, or motion instructions.
- Treat generated text and logos as unreliable; add exact typography in post-production.
- Compare models with the same brief before changing several settings at once.
Troubleshooting
Unexpected objects: simplify the prompt and add a concise exclusion.
Weak motion: describe the subject action and camera motion separately.
Composition changes: specify framing, subject position, and shot size.
Task delay: check creation history. Processing time varies with provider queues and selected settings.
Start with the active controls
Supported duration, quality, format, and cost are model-specific. Use the creator—not a fixed documentation number—as the source of truth.