# Video creation & downloads

## Create a video

1. Open [Create](/en/app#create), select the model and supported generation mode.
2. Enter the prompt or upload a Markdown prompt document. Review the imported text before submitting; uploaded Markdown is prompt content, not an executable configuration.
3. For image-to-video, provide a reference image using the input method supported by that model. Confirm its preview, format and size.
4. Open Advanced settings. Select supported duration, aspect ratio, resolution and other controls exposed for that model. Not every model accepts every parameter; changing a model can change the available controls and price.
5. Review the displayed estimate and available credit, then submit once. Keep the task/request ID.

## Where to find the result

The result appears under the creation form. A task ID or “submitted” status is not the finished file. Wait for completion, then open the output preview or download link. Reopen Request history and its task details to recover previous results. Download important resources promptly; provider URLs may expire.

If a task says completed but shows no resource, do not immediately generate again. Open the raw result details, keep the task ID and submit a [support ticket](/en/support). A completed status without an output URL needs investigation; it is not a usable download.

## Control quality and spending

Describe subject, action, camera, timing and exclusions clearly. Use a short low-cost draft before increasing resolution or duration. Parameter effects are model-specific; prompt wording cannot guarantee an exact video duration or physical consistency. Use [cost analysis](/en/docs/billing) to compare the final charged cost and known video seconds.