AI packaging photography for food, beverage and CPG: compliant labels, craveable scenes

In food, the pack is the promise — and the promise has legal ink on it. Nutrition panels, barcodes, certification marks and claim text must appear exactly as printed, while the serving suggestion next to them must look genuinely appetizing. Berrys splits the job the way a studio would: compliance surfaces come from your real artwork, appetite comes from generated scene and styling — and the two never bleed into each other.

Reviewed October 2026 · Prepared by Berrys · Technical specifications & Canvas node workflows

CPG visual compliance quick reference

Parameter / SlotMarketplace & Production Requirement
Never generatedNutrition panels, ingredient lists, barcodes, allergen callouts, certification marks — taken from real artwork only
Must match packServing-suggestion contents, stated weights, honest fill levels
Marketplace coversShopee: solid bg, ≥60% frame · Amazon: RGB 255 white, ≥85% fill · TikTok: real photo main, front view
Claims regimesHealth/nutrition claims carry substantiation duties in SG/MY/ID/EU/US — the medium (AI or studio) changes nothing

What breaks when food packaging meets a generation model

The nutrition panel problem

Dense, small-typography regulatory text is the single hardest thing for any generation model. One mis-rendered 'mg' is a listing violation. Rule: panels, barcodes and ingredient lists are never generated — they are preserved regions cropped from your real artwork inside every render.

Matte pouch physics

Foil-lined stand-up pouches combine crinkle, sheen banding and gusset geometry. Models average the crinkle into plastic-flatness. Reference geometry plus controlled-light recipes keep the material honest.

Condensation realism

Cold drinks live on droplet behavior — size distribution, gravity streaks, light through condense. Generic 'freshness' often reads as dirty glass until tuned against your real bottle.

Serving-truth boundaries

A cereal bowl overflowing with fruit the box doesn't include isn't styling, it's misrepresentation. Scenes may provide context; implied contents must match the pack's own serving suggestion.

How Berrys runs the CPG catalog

Preserved-region rendering

Compliance surfaces pass through untouched from your artwork files while scene and styling generate around them.

Flavor-family batching

One background + lighting graph, N flavor SKU nodes: the whole range renders with identical mood, only pack artwork swapping.

Channel crops from one master

Each render outputs the marketplace cover, the in-hand lifestyle crop and the banner crop — product pixels identical everywhere it appears.

Production workflows in Berrys Canvas

Workflow 1

Flavor-family batch (stand-up pouches & cans)

1. File Node: Upload approved label artwork dielines for all flavor SKUs.
2. Brand System Node: Apply matte pouch or aluminum can lighting preset.
3. Image Node (Nano Banana 2.1): Render 2048px packshots with preserved Nutrition Facts and barcode regions.
4. List Node: Fan out across 6 flavor variants maintaining identical lighting geometry.
Workflow 2

Serving-suggestion & appetite context

1. File Node: Connect approved packshot master.
2. Brand System Node: Define honest serving context (e.g. breakfast table or iced pour) matching stated package net weight.
3. Image Node (Nano Banana 2.1): Render mouth-watering scene without modifying product packaging pixels.
Workflow 3

Condensation & texture macro pass

1. File Node: Ingest beverage can or bottle reference.
2. Brand System Node: Enable micro-droplet condensation realism rules.
3. Image Node (Nano Banana 2.1): Render 2048px macro crop showing realistic gravity droplet streaks and cold beverage chill.
Berrys Plugins · Model Context Protocol (MCP)

Create in chat. Keep it on your canvas.

Already working in an AI agent? Connect Berrys to Claude, Cursor, ChatGPT, Codex, or Hermes. Your agent plans the food & beverage suite and runs the shots on your Berrys canvas, with saved Brand Memory and credit approval in chat.

Connector URL: https://app.berrys.io/api/mcpRemote MCP
Drop-in Agent Brief (Food & beverage)
Using Berrys, create an appetizing, compliance-true CPG packshot suite from my artwork in ./assets:
1. File node: Load packaging dieline artwork.
2. Brand System node: Lock Nutrition Facts panel and barcode regions as immutable preserved zones. Apply 'Matte Pouch Studio' preset.
3. Image node (Nano Banana 2.1): Render 2048px hero packshot on pure white.
4. Serving suggestion: Render an appetizing breakfast table scene matching physical pack contents.
5. Macro pass: Render 2048px close-up on packaging texture and seal finish.
Estimate credit cost and request approval before generating.

Compliance & production boundaries

Food is regulated representation: what the pack shows is what the buyer receives. Berrys treats compliance surfaces as immutable inputs and appetite as the generated variable. Where a platform or regulator requires actual photography of prepared contents, keep that one shot and let the canvas handle everything around it.

When NOT to use AI generation for this

  • Anything resembling a regulatory panel you haven't printed yet
  • Serving scenes implying included items that aren't in the box
  • Freshness states the shipped product can't survive (hot-steam scenes on ambient goods, etc.)

Frequently asked questions

Will the AI keep my nutrition panel and barcodes readable?

They don't go near it: compliance surfaces are preserved regions fed from your printed artwork files. Generation happens around them. That's the difference between a tool you can ship with and a toy that invents 'mg' glyphs.

Can it handle pouches, cans, glass and cartons together?

Yes — each format is its own SKU node with its own geometry and material recipe, under one shared lighting graph, so a mixed catalog reads as one brand.

Can it do Amazon/TikTok food content too?

Same engine, different export: compliant white-background masters for marketplaces and short pour/crumble motion clips for social, all locked to the identical pack.

How is serving-suggestion styling kept honest?

You own the composition rule; the canvas applies it consistently. Generated context (surface, lighting, garnish) can change; implied contents cannot exceed the pack's own claim.

In-Depth Guides & Research from Berrys Journal

Create on-brand assets without booking a studio

Start free on Berrys Canvas — 50 credits, no card required — or send Berrys Studio a brief and get finished campaign visuals with agency art direction.

Chat with us