Rivya Video Studio: Generate and Review a Clip
Create a video task in Rivya, choose compatible references and motion settings, check audio and credit costs, and review the complete clip before download.
Last reviewed on September 10, 2026
Video Studio is Rivya’s signed-in workspace for submitting video tasks and reviewing their results. Start with a specific deliverable, such as a product shot with a slow camera move. A generated clip still needs review and may need editing before publication.
Choose a model for your starting material
Check the model directory and availability guide. A public model page does not mean every input mode is available in the workspace.
| Starting material | What to check before choosing |
|---|---|
| Text describing a scene | The model supports text-to-video and the required settings |
| An approved still image | The model accepts an image and its role matches your intent |
| First and last frames | The selected mode provides those frame inputs; two references are not automatically endpoints |
| A reference video or motion example | The model supports that video input and the intended operation |
| A portrait or video with recorded speech | The model supports the required image/audio or video/audio combination |
If the product appearance is still undecided, prepare and review a still in Image Studio. For supported reference roles, see video references. These are different model capabilities, not modes every video model includes.
Submit a focused first task
Sign in and select the model and input mode. Define one scene and the movement you need to evaluate.
Upload the required files and wait for completion. Inspect the attachments actually shown, including dedicated frame or reference fields. Check the model’s count, format, file-size, pixel-size and duration limits in references and uploads.
Where the model accepts a prompt, describe the opening, main motion and ending. Identify what should remain unchanged in the subject.
Choose the available duration, aspect ratio, resolution and audio settings for the intended placement. Check the displayed credit estimate before submitting.
Send the task, follow its status and play the complete result. Save a useful result before starting a revised attempt.
For an image-compatible model, an example prompt is:
Use the attached bottle photo as the visual reference. Begin with the bottle centered on a plain desk. Slowly move the camera closer while the bottle stays still. End with the whole bottle and label in view. Do not add objects or change the bottle’s color.
This illustrates a reviewable request; it is not a tested output or a promise of exact product fidelity. See prompt-writing basics for clearer instructions.
Check settings and audio separately
Choose vertical, square or horizontal only where the selected model offers that ratio. Use quality, duration and aspect ratio to match the output to its destination. Available durations and resolutions vary; writing “4K” in the prompt does not select a resolution. A shorter test is useful only if it can answer your motion question, and it does not guarantee a lower total project cost.
Some models generate audio; others use uploaded speech or other audio references. Those inputs are not interchangeable. Inspect the selected model’s audio setting and play back the result to check what it actually produced. See audio uploads for input constraints.
You can prepare separate narration or music in Audio Studio. Adding it to a finished edit requires a suitable editor; generating audio does not automatically attach it to an existing video, mix a soundtrack or create localized versions.
Changing the video model can reset parameters and remove unsupported or excess attachments. Recheck the entire form, including frame inputs, before sending another task.
Review the whole clip and save the usable files
Watch from beginning to end at the intended display size. Check subject shape, product color and label, hands, edges, reflections, unwanted objects, camera movement, opening and ending, and any audio timing or unexpected speech. A plausible motion is not evidence that the real product works that way. Use the commercial review checklist before publishing.
Exact interface text, subtitles, logos and dense on-screen copy may need separate editing. A multi-scene campaign also needs asset selection and assembly. The workspace is not a complete editing timeline.
Find submitted tasks in History, use the Notifications Center for updates, and download accepted outputs. Keep your own source files, approved prompt, model and settings. Saved sessions do not guarantee a complete editable project or permanent file access.
Understand retries and result operations
Each new generation has a cost. The estimate can depend on the model, duration, quality, audio or reference inputs. A reservation is not the final settled charge; see credits and billing. A technically successful clip you dislike is not automatically a failed task eligible for a credit return; see failed tasks and refunds.
Some eligible results offer extension or higher-resolution operations. Availability depends on the source task and model. Review the operation and its cost before sending it; these are separate tasks, not a universal editing control or automatic continuation of every clip.
If the page loses its connection after submission, check History before resubmitting. Waiting or a disconnected page alone does not prove that generation failed. See queue and waiting guidance. For an unresolved issue, retain the model name, task ID, approximate time and visible error for troubleshooting.
Build a larger video workflow
Use video workflows to plan multiple assets, review checkpoints and finishing work. For product-specific examples, read product demo videos and prompts for product demos.