How to put multiple products in one AI photo with Lamina

For a multi-product AI photo in Lamina, first approve one accurate group composition, then treat it as the fixed merchandise reference for every on-brand variant. Don’t ask the generator to build a bundle from a loose stack of similar SKU images. A bundle image makes an offer claim: the items, variants, and count all have to be right.

Lamina’s product workflow is built around taking one accurate product image and producing many on-brand assets from it. The rules are firm for a reason: every SKU or variant gets its own reference pack, product truth clears approval before creative selection, and channel versions come from approved images only. That discipline gets even more critical with a three-piece gift set, a six-color collection ad, or a subscription box with tightly specified contents.

What is the safest Lamina workflow for a multi-product AI photo?

Start with one approved group composition as the product reference, then build scenes and channel treatments around it in Lamina. Lamina’s published seven-step process begins with references and rules, runs through cleanup and hero creation, and requires fidelity QA before batch export. For bundles, that group composition is the extra checkpoint ahead of the hero-image stage.

Lamina’s published materials cover brand-kit-led generation from a brief and accurate base imagery; they do not describe a multi-SKU placement canvas that independently positions separate product files in one frame. Fine. Build verified cutouts into one master group image before the Lamina generation step instead of trusting a text request to assemble the merchandise.

Claid documents the same fallback for a single-product-style generator: place products together on a transparent background, upload the result, and treat that ensemble as one product. Nightjar describes the other interface pattern, where up to eight product assets can be named and positioned separately in one prompt. They are different setups. For Lamina bundle production, use the pre-composed master until you have confirmed the workspace’s current input options.

What prompt should you use for a three-product gift set?

A useful prompt tells Lamina what merchandise is fixed and what may change in the scene. Begin with the approved group master, then give the model only the composition and brand constraints required for the campaign treatment.

Use this template: “Use the attached approved three-product gift-set reference as the exact merchandise. Show exactly three included products: [hero SKU] in the foreground center, [SKU two] rear left, and [SKU three] rear right. Preserve label text placement, package colors, proportions, and visible cap shapes. Place the set on [surface] with [lighting direction] and [shadow style]. Add only [permitted non-saleable props]. Do not add, remove, duplicate, substitute, or obscure products. Leave [left/right] negative space for headline copy. Output [aspect ratio] for [channel].”

“Exact merchandise” does real work here. It frames the brief as controlled staging, not product invention. Rivya’s reference guidance likewise flags small text, precise logos, tiny packaging details, and strict geometry for close review. A persuasive lifestyle surface cannot cover up a doubtful label or altered carton edge.

How should PDP, collection-ad, and lifestyle bundle images differ?

A PDP bundle image must prove the offer. A collection ad should establish hierarchy, while a lifestyle image creates context without changing that offer. Use the same approved group master across all three, then adjust the crop and prop tolerance for the job.

The PDP version stays literal. Keep every included item visible, make packaging readable where feasible, avoid props that could look like included merchandise, and retain manifest order if the listing copy names it. This is what a shopper uses to verify what will arrive.

Collection ads can push harder. Give the hero SKU the greatest visual weight, hold copy-safe space, and use a tighter crop when the group still reads clearly. For a lifestyle or seasonal gift-set image, change the surface, light, and non-saleable styling around the approved group master; leave the product count and packaging alone.

Where should the human approval gate sit?

Put product-truth approval ahead of every channel adaptation, then run art-direction approval afterward. Lamina’s benchmark separates product truth from later creative selection and recommends deriving channel variants from approved images. That sequence stops a polished 4:5 ad crop from becoming the first place anyone spots a missing travel size or swapped colorway.

The product-truth reviewer works from the manifest and source pack, never memory. Inspect the final file at the size required to see label edges, closures, logo forms, carton corners, and overlaps. If the master is wrong, fix the cutout or group composition and regenerate. A handsome background, shadow, or styling does not earn approval for a bad master.

The creative reviewer has another brief: copy space, hierarchy, crop, channel format, palette, lighting, surface, and permitted props. Folding both reviews into one vague “looks good” pass is how bundle imagery becomes a merchandising error.

How should you plan generation time for bundle variants?

Treat generation as one production step, not the whole production clock. Lamina platform telemetry recorded a median asset-generation time of 225s and a 90th-percentile time of 472s on August 20, 2026. That works out to roughly four minutes for a typical generated asset and nearly eight minutes at the slower end—before human review, corrective composition work, revision cycles, approvals, or media trafficking.

For a small campaign, approve the group master first, then generate only formats with a real buyer. A sensible batch could be one literal PDP proof image, one 4:5 collection-ad composition, one 9:16 social version, and one email crop. Reusing a locked merchandise reference and template saves time; you still need to inspect final package details.

Claire Xue, a creative leader, describes the practical reason commercial teams care about this workflow: traditional production can involve props, locations, studios, and models. AI-assisted generation changes that production path. It does not make a weak bundle definition safe to publish.

What must be checked before a multi-product AI image goes live?

Publish only after the final exported file passes a full-resolution manifest check and a separate channel check. The release call is straightforward: reject it if a buyer could infer the wrong contents, color, quantity, or package from the image.

Use this final checklist: every included SKU appears once; no SKU is duplicated; each variant and color matches the manifest; labels, logos, caps, seals, and carton shapes remain credible; relative scale does not imply a false size; props do not appear included; the hero item matches the campaign brief; headline space stays open; the crop fits its destination; and the approved master, prompt, manifest, and export are stored together.

That final record makes a holiday gift set repeatable for Valentine’s Day, a new marketplace crop, or a collection refresh. Lamina can generate new surroundings. The approved group composition keeps the merchandise honest.