Rivya Video Workflow: Plan, Test, and Finish
Plan a video project in Rivya around source material, shot requirements, sound and review. Test a focused clip, control retries, and prepare files for editing.
Last reviewed on November 11, 2026
Use this workflow to plan a video project, test candidate clips and prepare a finished edit. For the upload controls and submission steps, see Video Studio. A project with several scenes needs planning, selection and assembly beyond generating each clip.
Separate generation, editing and upscaling
Gemini Omni 1.2 Flash has distinct reference, first/last-frame and video-edit modes. Keep frame input separate from references or video. Editing follows the server-verified source duration and does not send the generation duration field.
Grok Imagine Video 1.5 Preview now includes a priced 1080p option while retaining its one-image input limit. Rivya Video Upscale processes an existing clip within a selected output box; it does not turn a text prompt into a new scene. Review source duration and the current reserve before submitting either video operation.
Define the deliverable before choosing a model
Write down the destination, required framing, subject, movement, sound and acceptance criteria. For a bottle promotion, you might need a wide product shot for a website and a vertical version for a social post. Review each separately; changing the ratio can change the composition and product details.
Keep the brief tied to the real product. A requested camera move is different from asking the bottle to demonstrate a mechanism that has not been verified. Decide which shape, color and label details must remain accurate.
Use the video overview to explore the category, then compare compatible options in the model directory. Check availability, supported inputs and the current form settings for the operation you need. A model’s name or public page alone does not establish that it supports that operation.
Choose the starting material and its role
| Starting point | Decision to make |
|---|---|
| Text | Describe a scene and a movement that the selected text-to-video model can attempt |
| Approved still image | Use an image-compatible model; state which product or layout details should remain |
| First and last frames | Select a mode with those inputs; two reference images are not automatically endpoints |
| Existing footage | Confirm the model’s actual operation and limits; an accepted video is not a promise to preserve all cuts, sound or subtitles |
| Speech or sound that drives the scene | Choose a model with the required audio input or generation capability before the visual test |
If the visual direction is unresolved, prepare a still in Image Studio. Check video reference roles and the model-specific count, format, size, dimensions and duration limits in references and uploads. Do not assume inputs from one model remain valid after a switch.
Test one question in four steps
Choose one representative shot and write its opening, main movement and ending. Define what would make the result usable.
Sign in before uploading or submitting. Select the model and mode, wait for uploads to finish, and check which files were successfully attached, including any dedicated frame fields.
Set the available duration, ratio, resolution and audio options for the shot. Check the credit estimate. A shorter test is useful only if it can answer the question; it does not guarantee a cheaper project overall.
When the task completes, watch the full clip and listen to any audio. Save a usable result locally before changing the next request.
For a model accepting an image, try a brief such as:
Use the attached bottle photo for its appearance. Begin with the whole bottle on a plain desk. Move the camera slowly closer while the bottle stays still. End with the whole bottle and label visible. Keep its color unchanged and add no objects.
This is a writing example, not a tested result or an exact-reproduction promise. Match the real controls using quality, duration and aspect ratio; descriptive words do not select unavailable settings.
Decide whether to revise, change models, or stop
| What the review shows | Next action |
|---|---|
| The movement or framing is wrong | Clarify that instruction and check the relevant setting before another attempt |
| A reference conflicts with the intended product | Resolve its role or replace the conflicting input within the model’s limits |
| The required input or operation is unsupported | Choose a compatible model and recheck every setting and attachment |
| The clip meets the agreed criteria | Save it and move to the next shot or finishing stage |
Keep a short record of the change and the observed difference. Changing the model, prompt and references together makes the cause harder to identify. Model changes can reset settings and remove unsupported or excess inputs.
Set a retry budget and a review point before generating more. Rivya reserves credits when it submits a generation; the estimate can depend on duration, quality, audio or reference inputs, and the final credit status depends on the task and reported usage. A more expensive model does not guarantee a better outcome. See credits and billing and failed tasks and refunds: a technically successful clip you dislike is not automatically refundable.
If the page disconnects after submission, check History before sending again. Waiting or losing a browser connection is not proof of task failure. See queue and waiting guidance; the Notifications Center also helps you check updates.
Plan sound and finish outside the generation form
When speech timing, lip movement or generated sound is central, test that capability early with a compatible model. For separate narration or sound effects, check the available models in Audio Studio. Suno Music supports new custom-mode generation; Music extension, WAV conversion and vocal separation from Music results remain unavailable. Assemble the chosen audio and video files in a suitable editor; Rivya does not automatically mix separate audio into an existing clip or create every language version.
Where available, use multiple-shot controls to prepare a new generation request from shot prompts and any supported reference inputs. Check the selected model's requirements and credit estimate before submitting. Veo extension and higher-resolution actions require an eligible successful result from your account, with a matching source model; check the source-result requirements and separate estimate for the action. These options do not replace an editing timeline or ensure continuity between clips generated in separate requests.
Review each shot and the assembled edit for product accuracy, unwanted objects, readable text, crop, motion and sound timing. Compare product details with the original material, not only with the previous generated clip. Use the commercial review checklist, and add exact interface copy, subtitles or logos through editing where needed.
Save the handoff and continue with a specific use case
Download approved outputs while they are available, and keep your source files, accepted prompts, model names, settings and final edit together in your own project folder. Saved sessions are not a guarantee of a complete editable project or permanent file access.
For the first generation, read how to generate video with Rivya. For a campaign or product brief, continue with marketing videos or product demo videos.
