Manufacturer workflow
Shortlist fabric options for your furniture range
Start with one identified SKU, try three preset sofa colors, then prepare a supplier swatch request for the fabrics you want to check.
By XinVise TeamPublished Updated


Built for this workflow
- Keep SKU, finish, source, and fabric ID together
- Review structure before judging the upholstery direction
- Hand visual candidates to product and textile reviewers with open questions
Start with one traceable SKU baseline
Record the SKU or style identity, finish, intended use, and the features that must not change. Use one complete furniture source photo as the baseline so every candidate answers the same variant question. If a manufacturer-owned swatch or collection is not available in the current public library flow, treat it as a scope question for evaluation rather than implying arbitrary upload support.
- Keep silhouette, arms, back, cushions, legs, seams, hardware, text, and logos visible
- Record the owner-supplied fabric name or ID
- Keep the unchanged source available for every comparison
Generate a controlled variant comparison
Prepare the furniture image and textile input using the relevant photo and swatch guidance, then use the current fabric-preview flow with an available fabric selection. Compare a single candidate or a small controlled set on the same baseline. The result is a visual direction, not a measurement, color-management proof, or production drawing.
- Hold the product image and review size constant
- Check coverage, direction, repeat, sheen, and apparent texture
- Do not infer hand feel, durability, construction, or exact scale from pixels
Deliver identified candidates, not an implied production pipeline
A product owner should check identity, outline, proportions, seams, hardware, text, and branding. A textile or upholstery reviewer should check coverage, pattern direction, repeat, and specification boundaries. Content or channel owners decide whether a candidate fits a named proposal or launch-draft use. Record pass, hold, or reject with the reason and unresolved questions.
For a factory, keep one supplied SKU identity and unchanged furniture source with every candidate and review decision. For a textile brand, retain the supplied collection or article reference, image version, rights status and physical sample. A small, manually reviewed comparison is not proof of automatic batch capacity or API availability. Agree the intended deliverable and unresolved questions before scaling or making a material claim.
Factory note: distinguish a visual shortlist from a cutting instruction or approved specification. Give the sampling team the source/candidate references, textile identity, reviewer role, date, intended use and open issues; attach the actual material and relevant product records for physical decisions.
| Handoff record | Minimum human record | Capability or approval not established |
|---|---|---|
| Authorized inputs | Supplied SKU/finish, source version and allowed use; minimal references rather than customer catalog contents | No identity recognition, asset-upload permission or unrestricted rights |
| Textile reference | Available catalog identity or separately scoped owned reference, known orientation/repeat unit and physical sample link | No arbitrary public collection import or inferred fiber/weight/performance |
| Candidate lineage | Source and output references, actual workflow, known mode/date and changed-source notes | No automatic export, stored ledger, version impact detection or PIM/DAM connector |
| Review decision | Product identity/count, seams, text/logos, intended use, reviewer role, date and open issues | Visual pass is not material, manufacture or channel approval |
- Choose a bounded use and identify the real SKU and textile references. Public generation still requires sign-in and available credits; do not assume free retries or credits from a tool-discovery visit.
- Compare a small set manually against the unchanged source. Route defects and missing physical facts before handing off. Keep rejected candidates and unanswered questions rather than counting every image as usable.
- Send a minimal review packet through an approved process. A public text contact form is not an authorization to send customer personal data, credentials or an unapproved collection.
- If batch/API is relevant, obtain specific product and commercial scope plus representative evaluated outputs, failure handling and agreed responsibilities. Neither a demo nor this checklist proves that XinVise has a batch generation API.
Recommended process
- 01
Lock the SKU baseline
Record the product identity and non-negotiable structure before preparing any variant.
- 02
Prepare the inputs
Use a complete furniture photo and an identified available fabric direction; ask for scope evaluation when an owned library is required.
- 03
Compare candidates
Generate a small comparable set and inspect product identity, upholstery coverage, direction, repeat, seams, and branding.
- 04
Hand off with a decision
Record pass, hold, or reject for the stated use and send open questions plus physical specifications to the responsible owner.
Sources and method
These primary sources support method and platform constraints; they are not third-party endorsements of XinVise product outcomes.
- GS1 Product Image Specification
GS1
Provides a reference for product views, identifiers, naming, and asset governance; it does not prove AI structure or production outcomes.
- About color profiles
Adobe
Explains screen, profile, and device differences so visual candidates remain separate from physical color approval.
Frequently asked questions
Does the source need a white background?
No. A clean, non-distracting background can help, but the important requirements are a complete, identifiable product and visible details. Use the furniture-photo guidance for capture decisions.
How should several fabrics be compared?
Keep one representative SKU image, crop, and review size fixed, then compare identified fabric directions one at a time or as a small controlled set. Preserve the source with every candidate.
What if seams or pattern scale change?
Put the candidate on hold or reject it for the defined use when seams, coverage, direction, repeat, structure, text, or branding change beyond the review question. Route it to retouching or another method when needed.
When is a candidate delivered and usable rather than production-approved?
A usable candidate meets the predeclared visual purpose and responsible review, not merely an accepted upload or successful request. Keep evidence of the actual handoff and unresolved issues; physical, technical and production approval remain separate. A GA4 identifier, widget or simulated event proves none of those outcomes.
Discuss a manufacturer workflow evaluation
Share representative SKUs, fabric combinations, intended use, and acceptance questions so the scope can be reviewed without promising a pilot result or integration.
Submit representative SKUs