How to create on-brand AI product images with a human model: a step-by-step Lamina workflow for turning packshots into ecommerce lifestyle, catalog, and social assets while preserving logos, labels, packaging, and product proportions
A practical Lamina workflow for turning a locked SKU packshot into human-model catalog, lifestyle, and social images without treating packaging accuracy as optional.
Lamina Team
Product Team @ Lamina
How can you create on-brand AI product images with a human model while keeping the product unchanged?
Build the scene around a locked SKU reference; do not ask for a fresh version of the product. The packshot, supporting angle views, label artwork, dimensions, and non-changeable copy set the fixed parts. Lamina can vary the model, pose, scene, light, props, and crop.
That split heads off the usual mess: a slick lifestyle shot with the wrong bottle cap, rewritten label, or a product that looks absurdly small in someone’s hand. Lamina is built to produce brand content from a brief and brand kit—product photography, model try-ons, lifestyle scenes, reels, and campaign banners. Treat the brand kit as production control. It is not a mood board.
| Metric | Value | Source |
|---|---|---|
| Shared generation cost across the three tested variants | $0.040 per asset | uselamina.aias of 2026-08-04 |
| Controlled catalog-hero generation time | ~49 seconds | uselamina.aias of 2026-08-04 |
| Model-led lifestyle ecommerce generation time | ~40 seconds | uselamina.aias of 2026-08-04 |
| Social-first campaign-asset generation time | ~26 seconds | uselamina.aias of 2026-08-04 |
| Why visible package copy requires a separate approval pass | AI-generated text can be garbled, making altered close-up listing images unusable | pebblely.comas of 2026-03-20 |
What did the Lamina workflow experiment actually show?
The reported experiment shows generation speed and cost across three locked-product-reference variants, not a quality winner. The catalog hero took about 49 seconds; the model-led lifestyle image took about 40 seconds; the social-first asset took about 26 seconds. Each carried the same reported per-asset generation cost.
The test used the original packshot as its product source, a separate human-model reference, explicit preservation constraints, and staged work: composition first, detail verification second. Sensible production setup. Still, the supplied results offer no scored comparison of logo legibility, label fidelity, package geometry, realistic hand contact, approval rate, retouch time, or generations per approved asset. Treat the reported figure as generation cost alone; human review, revisions, and distribution work sit outside it.
A reported Lamina experiment tested a staged, locked-product-reference workflow across a catalog hero, a model-led ecommerce lifestyle image, and a social-first campaign asset.
Workflow coverage
over Reported 2026-08-04
Generation economics
over Reported 2026-08-04
Product-fidelity evidence
over Reported 2026-08-04
What should go into a SKU truth pack before generation?
A SKU truth pack needs enough proof to identify the exact item and reject every accidental substitution. Start with a current, clean packshot. Add front, side, and back angles where relevant, logo and label close-ups, print-ready artwork, exact dimensions, material and finish references, the current size or variant, and a written list of details that cannot change.
Keep packaging revisions, regional labels, colorways, and sizes out of the same set. For branded products, retain print-ready label art, a complete angle set, dimensions, material references, and non-changeable claims; if a generated scene damages a small area, restore exact packaging through perspective-aware compositing. Lamina’s brand-kit approach supports locked product references and reference imagery, so keep SKU truth separate from campaign direction.
How do you turn a packshot into approved human-model assets in Lamina?
Define the deliverables and their hard requirements
List the catalog or PDP image, lifestyle image, and social crops before you generate anything. For each asset, spell out the destination, aspect ratio, product prominence, crop-safe area, and copy shoppers need to read. A catalog hero and an expressive social crop need separate briefs, not one fuzzy instruction.
Load the SKU truth pack, then update the brand kit
Add the approved packshot and supporting SKU evidence, followed by the approved palette, typography, voice, visual do and do-not rules, and campaign references. Keep identity rules apart from styling: SKU, label layout, logo placement, color, finish, dimensions, and pack size stay fixed. The setting and styling can move.
Specify a model interaction that could actually happen
Use a licensed or otherwise permitted model reference if you need a specific likeness. For handheld work, state the hand, grip, product orientation, finger placement, whether the front label faces camera, and whether the item is held, opened, poured, used, or displayed. That gives reviewers a physical standard to check instead of an aesthetic hunch.
Write a production brief that locks the product
Name the exact SKU and direct Lamina to preserve its silhouette, dimensions, closure, label layout, visible text, logo placement, color, finish, and pack size. Then set the model action, environment, props, camera angle, light, negative space, and destination format. Example: Use the supplied SKU as the locked product reference. Preserve its exact packaging. Create a 4:5 PDP lifestyle image of a model holding it upright in her right hand, front label facing camera, with natural finger contact and soft window light. Change only the environment, model, pose, and lighting.
Generate a small composition batch, then choose on fidelity
Judge candidates at full resolution. Thumbnails lie. Choose the frame with believable grip and occlusion, contact shadows that belong in the scene, package-consistent material reflections, credible scale against the body, and an unchanged, readable package wherever the channel requires one.
Run visual QA separately from product and claims QA
Visual QA covers silhouette, cap or pump, material, reflection, perspective, label placement, hand anatomy, contact, and crop. Product and claims QA checks every visible brand name, product name, variant, dosage, size, certification, and claim against the source SKU. A persuasive image fails the moment it changes what ships.
Repair the failed detail only, then review again
If a nearly approved scene has one wrong logo, word, or small packaging feature, mask and repair that area using clear reference shots rather than regenerating the whole frame. Where exact artwork is required, composite the approved artwork or product cut-out in perspective. Then check curvature, occlusion, reflections, shadows, and edges again.
Record approval, then distribute the approved asset
Keep the source packshot, truth-pack files, brief, brand-kit version, seed where used, output ID, and approval record with the final asset. Lamina documents a create, track, evaluate, and distribute workflow. Delivery options include S3, Google Drive, Sanity, Shopify, and webhooks.
How should you check logos, labels, proportions, and human contact before publishing?
Approve the product as though this were a product listing. Shoppers will read it that way. Compare all visible package text—including brand name, ingredients, dosage, size, and claims—against the source item, correct the affected area, then review the finished image again.
Use a second pass for geometry and interaction. Check package silhouette, closure, relative size, perspective, label position, surface finish, finger overlap, and cast shadow. Generative systems make pixels holistically; they do not reliably read a label. Brand-critical text is therefore an approval requirement, never an assumed output capability.
When should you repair or composite rather than regenerate the whole image?
Repair locally if the composition, model pose, lighting, and product placement are approved and only one detail is off. A focused mask-and-regenerate workflow can fix a logo, text fragment, or small surface detail while keeping the rest of the frame intact. Clear reference shots give the repair a better target.
Use perspective-aware compositing for exact label artwork or a verified product cut-out where packaging fidelity is non-negotiable. That does not let you skip review. Check the corrected product one final time for edge integration, curvature, lighting direction, shadows, reflections, and natural hand occlusion.
What is the practical Lamina workflow for ecommerce teams?
Use Lamina to build channel-specific presentations around a verified SKU, with product QA as the gate between creation and distribution. The working order is plain: define the asset set, lock the truth pack, set brand rules, direct the model interaction, generate compositions, approve at full resolution, repair narrow defects, and keep an audit trail.
Human art direction still matters, especially for hero images and regulated or claim-heavy packaging. Generation earns its keep by giving you a repeatable way to make catalog, lifestyle, and social variations without letting a handsome scene quietly swap out the product you sell.
Methodology
Original Lamina experiment run 2026-08-04. Hypothesis: A locked-product-reference Lamina workflow—using the original packshot as the product source, a separate human-model reference, explicit preservation constraints, and staged generation (composition first, detail verification second)—will produce more on-brand, ecommerce-ready lifestyle, catalog, and social images than a single all-in-one generation prompt, while retaining logo legibility, label accuracy, package geometry, and realistic human-product interaction.. Measured 3 variant(s) for cost and latency on the Lamina image engine; numbers cited here are our own measurements.
Continue reading
Data report: What is the best workflow for creating on-brand AI product images with Lamina? A benchmark of brand-kit setup, reference-guided generation, product-accuracy checks, and image-to-campaign variations for ecommerce brands.
The fullest Lamina workflow took about 52 seconds per measured run at the same $0.040 asset cost, but the supplied data does not yet prove a quality winner.
Lamina Team
Product Team @ Lamina
How to edit ecommerce product images with AI without changing the SKU: a Lamina workflow for accurate, on-brand backgrounds, retouching, channel crops, and campaign creative
A SKU-locked Lamina workflow for editing ecommerce imagery while preserving packaging, labels, color, proportions, and every product fact that customers use to buy.
Lamina Team
Product Team @ Lamina
On-brand AI product images: a practical workflow for generating ecommerce visuals that match a brand’s look
Build on-brand AI product images with fixed SKU references, reusable art-direction controls, batch production, and a review process that protects product accuracy.
Lamina Team
Product Team @ Lamina