Rivya Image Workflows: Plan, Compare, and Finish a Set
Plan a consistent set of AI images in Rivya. Choose inputs, test one composition, review changes, budget retries, and save approved files for the next step.
Last reviewed on November 11, 2026
Use this guide to move an image set from a brief to files you have reviewed and accepted. For uploads, generation controls, and result previews, use the Image Studio guide.
Choose the image operation before uploading
Use Seedream 5.5 Text to Image for a new image from text, or Image to Image for reference-led edits. Each submitted reference adds to the edit budget.
Layer Decomposition is a separate operation: its output limit includes the base image, and the reserve covers all potential outputs. Keep every returned layer and its JSON layout details when continuing in an editor. Rivya Image Upscale enhances an existing image; its price tier depends on verified output pixels, not the scale factor alone. Check the selected page for formats, size limits and output settings before uploading.
Define the set before generating
Write down the placements, facts that must remain accurate, and what may change between images. For example, a bottle campaign might need a square product image, a wide website image with space for a heading, and a vertical social image. These are three deliverables to review separately, not a batch that Rivya automatically assembles.
Keep the bottle's real shape, color, and label as acceptance criteria. Background, lighting, and framing may vary. Do not add unverified material, capacity, or performance claims to make the brief sound complete.
Choose inputs and an available model
Compare the image overview and model directory for supported inputs, output settings, and estimated credits. Check current availability before using a model's form; a familiar name does not establish suitability. Sign in before uploading or submitting work.
If the result must depict a real product, choose an available model that accepts the required references. Decide what each image should guide, such as product appearance or composition, and explain that role where the operation accepts instructions. References do not guarantee exact geometry, lettering, or consistency across a set. See image references and the model-specific upload limits.
Test one composition, then make the remaining images
Pick one representative placement and define its acceptance criteria before spending credits.
Write a focused brief, such as: “Center the bottle from the product reference on a light gray background. Leave space above for a heading. Add no props.” Enter it only where the selected operation accepts prompt instructions. This is an illustrative brief, not a tested output or a promise of exact reproduction. Use prompt-writing basics to clarify ambiguous instructions.
Check the supported settings, including aspect ratio and resolution where provided, along with the attached references and estimated credits. Submit one task. Choose a lower-cost setting only if it can answer the question you are testing. Lower cost does not guarantee useful results or lower total spending.
Review the image before adapting the approved direction to the other placements. Check each new form and result; a new ratio or reference can change the composition and product details.
Decide what to change and when to stop
| What you see | Next action |
|---|---|
| Wrong layout or unclear reference use | Revise the affected part of the brief, or clarify the reference's role where instructions are supported. |
| Good composition but visible defects | Identify the specific defect before another attempt or an available finishing operation suited to it. |
| Image meets the agreed criteria | Save it and move to the next placement. |
Keep a short note of what changed between attempts. Changing the model, prompt, and references together makes the cause of a difference harder to assess. Higher resolution alone does not correct false product details or an unsuitable composition.
Set a credit budget and stop to reassess when it is reached. A new generation or finishing task can incur a separate charge; review the estimate after choosing its settings, then check reserved or final credits afterward. See credit billing. A completed image that you dislike is not automatically a failed task or a credit return; see failures and refunds.
Finish and keep the approved files
Choose an operation for background removal, upscaling, or layer separation that is currently available, suits the remaining task, and accepts your image. Review edges, text, shadows, and any changed details at the intended display size. If you separate layers, inspect each result and assemble the files in an external editor. These operations do not verify product claims.
Find submitted tasks in History, then download accepted results. Keep the approved brief, model, settings, references, and final files yourself. Saved sessions are not a complete editable project archive or a guarantee of permanent file access.
For motion, continue with an accepted still and an available video model that supports the image input using video workflows. Review the destination's input limits, mode, settings, and credit estimate before submitting. The video request is a separate task; it does not automatically animate or assemble the rest of your image set.
Apply the workflow to a specific placement
Use the product-photo prompt guide for the product brief, the landing-page image guide for layout and text space, and the advertising image guide for an ad-specific review. Keep the same verified product facts across all three.
