For an indie brand, a DTC startup, or an Etsy/Shopify store, AI image models can now cover most catalog and ad work that used to need a studio booking. But the defaults look “stocky” or “AI render-y.” The fix is the same as for portraits: structured prompts with explicit lighting, surface, and camera language, generated on a model that actually suits the shot. Below are ten copy-ready templates, a June 2026 model comparison, and a consistency workflow.
TL;DR
- Best all-round product model (June 2026): Flux 2 Pro for photorealistic materials and lighting; Midjourney v7 for the most “art-directed” commercial look.
- If the package has readable text: generate on Ideogram v3, GPT Image 2, or Nano Banana Pro (Gemini 3 Pro Image) — they render legible type; everything else still garbles labels.
- Three things every prompt must name: surface/material, light direction + quality, and a real camera/lens. Drop
4k,8k, andhigh quality— diffusion models don’t parse them. - Commercial rights: Midjourney paid plans ($10–$120/mo) grant full commercial use; companies over $1M annual revenue must be on Pro ($60) or Mega ($120). Always recheck the current terms before a client job.
What separates “AI render” from “real product photo”
Three layers everyone underspecifies:
- Surface and material:
matte ceramic plate,polished walnut tabletop,crumpled linen background - Light direction and quality:
soft north-facing daylight from upper-left, not juststudio lighting - Camera spec:
85mm f/2.8 on Hasselblad, nothigh resolution
Also: name the deliverable type so the model composes for it:
- Hero shot (full product center, room to add headline)
- Lifestyle (product in use, in scene)
- Packshot (product on plain background, no shadow)
- Detail crop (zoomed on texture / closure / pour / fabric)
- Group / family shot (multiple SKUs aligned)
Which model for which shot (June 2026)
No single model wins everything. The split that matters in practice:
| Model (June 2026) | Best for | Plan / cost | Notes |
|---|---|---|---|
| Flux 2 Pro | Photorealistic materials, lighting, lifestyle | API / partner apps | The default to try first for catalog realism |
| Midjourney v7 | Art-directed hero shots, moody campaigns | Basic $10 → Mega $120/mo | --sref codes + Omni Reference (--oref) for catalog consistency |
| Imagen 4 Ultra | Hardest-to-fake photorealism (fabric, skin, reflections) | Google AI Pro $19.99/mo | Strong on texture fidelity |
| Ideogram v3 | Readable on-pack text and typography | Free tier + paid | The reliable choice when the label must be legible |
| GPT Image 2 | In-image text, multi-language labels | Inside ChatGPT Plus $20/mo | ~95% text accuracy; good for mockups with copy |
| Nano Banana Pro (Gemini 3 Pro Image) | Accurate text + subject consistency, fast edits | Gemini app / API | Pulls real-world references; strong for localized label text |
Practical rule: shoot the picture on Flux 2 Pro or Midjourney v7, and if the package carries text you need readable, either regenerate the label step on Ideogram v3 / GPT Image 2 / Nano Banana Pro or leave the label blank and composite real type in post.
Pricing above is as of June 2026 and changes often. Check the official pages before buying: Midjourney plans, Ideogram pricing, Google Gemini image.
10 copy-ready prompt templates
1. Skincare hero shot
Product photography of a frosted glass skincare serum bottle with minimal silver cap, sitting on a polished travertine block, surrounded by a single fresh eucalyptus stem, soft north-facing daylight from upper-left, gentle shadow to right, plain cream linen backdrop, 85mm f/2.8 macro, Hasselblad look, --ar 4:52. Coffee lifestyle shot
Lifestyle shot of a small-batch coffee bag on a warm wooden tabletop next to a steaming ceramic cup, hands of a woman lifting the cup, morning kitchen scene, soft window light from left, slight steam, warm color palette, 50mm f/1.8, --ar 4:53. Sneaker packshot
White sneaker on a plain white seamless backdrop, three-quarter front angle, soft overhead key light with subtle fill from below, no harsh shadow, clean catalog packshot for e-commerce, 100mm macro, sharp focus throughout, --ar 1:14. Beverage splash
A glass bottle of sparkling lemon soda with a frozen wave of liquid splashing around it, bright studio key light from above with hard shadow, vibrant yellow background, water droplets on glass, high-speed photography look, 85mm f/8, --ar 4:55. Jewelry detail crop
Macro shot of a gold signet ring with a small green tourmaline, resting on crumpled black silk, soft directional light revealing texture of fabric and metal, slight catch-light on stone, 100mm macro f/5.6, ultra-sharp detail, --ar 1:16. Tech gadget hero
Hero shot of a matte black wireless earbuds case open, single earbud floating slightly above with subtle motion blur, dark grey gradient background, hard rim light from behind, soft fill from front, modern Apple-style product photography, 85mm f/4, --ar 4:57. Skincare lifestyle (bathroom)
A skincare bottle on a marble bathroom shelf next to a folded white towel and a small green plant, soft morning bathroom light, faint steam in air, calm and clean Scandinavian aesthetic, 50mm f/2.8, --ar 4:58. Apparel flat lay
Top-down flat lay of a cream knit sweater folded neatly, paired with a brass watch, a leather notebook, and a single dried wheat stem, on a warm oak tabletop, soft overhead daylight, autumn editorial aesthetic, 35mm directly overhead, --ar 4:59. Food packaging on-shelf
A premium chocolate bar in matte dark green packaging with gold foil typography, standing upright on a polished walnut counter, blurred kitchen background with copper utensils, warm window light from right, 85mm f/2, --ar 4:510. Holiday / seasonal campaign
A bottle of fragrance on a dark moody background with scattered dried rose petals and a brass candleholder, single warm key light from upper-right, soft red and gold color palette, luxury winter holiday campaign aesthetic, 85mm f/2.8, --ar 4:5Per-deliverable tuning
- Hero shot for headline space: leave one third of the frame negative
- Packshot: plain white seamless + no shadow + sharp focus throughout
- Lifestyle: include a human element (hand, arm, partial face) — boosts conversion
- Detail crop: 100mm macro + close focus + texture description
- Flat lay:
top-down 35mm, neutral wood / linen / marble, complementary props
Common mistakes
- Asking for “professional studio lighting” — too vague, says nothing
- Forgetting the surface — model invents a random one
- Saying “high resolution” or “8k” — diffusion models don’t actually parse this
- Specifying brand names (Apple, Nike) — usually blocked, replace with style words
- Cluttered scene — model puts random props everywhere; specify exactly which props
How to keep style consistent across a catalog
Inconsistent lighting across SKUs is what makes an AI catalog look amateur. What actually holds a series together:
- Reuse one lighting line verbatim in every prompt (the single biggest lever).
- Same surface / background description across the series, and the same aspect ratio.
- Midjourney v7: apply a
--sref <code>(or--sref <url>) to lock palette, medium, and lighting; note v7 reinterprets old codes, so re-pick yours on v7. Add Omni Reference (--oref <url>with--owto set strength) to carry the actual product or prop across shots. - Flux / Nano Banana Pro: feed the same reference image and keep the prompt’s lighting and surface clauses identical between SKUs.
- A fixed seed only loosely steadies composition on current models; the lighting/surface text and a style reference matter more.
See How to Keep AI Image Style Consistent for the full workflow.
FAQ
Q: Can I sell images made with these models? A: On paid Midjourney plans (Basic $10 to Mega $120/mo, as of June 2026) you own the output and can use it commercially; free-trial images can’t be sold. If your company made over $1M in gross revenue last year, Midjourney requires the Pro ($60) or Mega ($120) plan for commercial rights. Flux, Ideogram, and Gemini/Nano Banana each have their own terms, so confirm the current license before a client job.
Q: Can I just upload my real product photo and ask AI to “make it nicer”? A: Yes — use image-to-image (or an edit model like Nano Banana Pro) at roughly 30–50% strength. The output keeps your product geometry but cleans up lighting and background. Lower strength preserves the product more faithfully.
Q: How do I avoid the “AI render” plastic look?
A: Specify a real lens (85mm f/2.8), real surfaces (linen, marble, walnut), and natural lighting (north-facing daylight). Drop 4k, 8k, and high quality — they do nothing. Imagen 4 Ultra and Flux 2 Pro produce the most convincing materials in June 2026.
Q: Can I get the AI to reproduce my exact bottle / package?
A: Not from a plain text prompt. Use image-to-image with your product photo as input, Midjourney v7 Omni Reference (--oref), or ControlNet keyed off the product silhouette.
Q: Best aspect ratio for product photography?
A: 4:5 for social feeds and Shopify product pages, 1:1 for catalog grids, 9:16 for Stories/Reels and App Store shots, 16:9 for hero banners.
Q: Why do labels and text on the package always look wrong? A: Most diffusion models still garble text. Either accept blurred type and composite real text in post (Figma / Photoshop), or generate the label step on a model built for text — Ideogram v3, GPT Image 2 (~95% text accuracy), or Nano Banana Pro.
Related articles
- Best Anime Character Image Prompts
- Realistic Beautiful Woman Portrait Prompts
- Cinematic AI Video Prompts
- How to Keep AI Image Style Consistent
- How to Write a Product Image Prompt: Lighting, Background, and Brand Consistency
- AI Sets Product Visual Direction
- Kitchenware Product Image Prompts: 10 Cookware Hero Templates