In this article
For most Southeast Asian e-commerce brands, the answer is neither pure studio nor pure AI: traditional studios stay cost-effective for hero products and detail-critical SKUs, while AI pipelines win on long-tail catalogue coverage, regional resizes and campaign variants. The brands that cut creative spend the hardest in 2026 run a hybrid workflow — a small studio master shoot per product, then AI-generated variants from approved references.
Editorial review: Berrys Team · 2 November 2026. Berrys operates an AI creative studio and Canvas platform. Price ranges below are standardized to USD with regional Southeast Asian benchmarks (such as published Singapore studio rate guides) and should be re-verified for your market and brief.
What a Studio Shoot Actually Costs in SEA
Published Singapore rates are the highest benchmark in the region; Malaysia and Indonesia typically run below them for equivalent output, because the cost base (studio rental, crew day rates, retouching) is lower while camera, lighting and post-production steps remain identical.
The numbers that appear across commercial studio rate guides:
| Deliverable | Standard Rate (USD) | Notes & Regional Context |
|---|---|---|
| Standard white-background product shot | $25–$100 per product | For 1–10 item batches (approx. SGD 30–150 in Singapore) |
| Same, at volume (30+ products) | $20–$50 per product | Setup stays constant, throughput rises (approx. SGD 25–70) |
| Half-day session (10–25 products) | $600–$1,500 | Crew + studio time (approx. SGD 800–2,000) |
| Full-day session (25–50 products) | $1,200–$3,000 | The unit economics improve sharply (approx. SGD 1,500–4,000) |
| Catalogue set: 50–100 SKUs, 2–3 angles | $1,500–$5,000 | White-background only (approx. SGD 2,000–6,000) |
| Styled lifestyle session (homepage/ads) | Add $1,000–$3,000 | Separate day, props, sometimes model (approx. SGD 1,500–4,000) |
| One complex hero campaign shoot | $2,000–$8,000 | Model, set design, location sourcing |
Two structural facts drive every quote:
- Fixed setup cost per product. Rigging, lighting a transparent perfume bottle, and colour-managing a black handbag each take a real photographer's time regardless of catalogue size. Complexity multiplies this.
- Linear scaling. A 500-SKU catalogue does not get 10x cheaper than a 50-SKU one; it gets roughly 10x more expensive, minus volume discounts. Industry surveys place the traditional average near USD 100 per SKU for a complete image set once studio time, crew, props and retouching are included.
The hidden costs sit outside the invoice: shipping physical products to the studio, collecting them back, approving images, and reshoots when a variant colour arrives late. For a seller rotating 30–50 SKUs per season, that logistics loop often costs more than the retouching.
What AI Product Photography Changes (and What It Doesn't)
AI generation breaks the two cost curves: no per-product rigging and no marginal crew time. Published comparisons consistently put AI at 50–80% below equivalent studio output, and tool vendors quote under USD 5 per SKU for full sets at catalogue scale. The honest caveats:
- Source quality caps output quality. AI cannot recover product facts missing from the reference. A blurry phone photo of the back of a bag produces a confident, wrong back panel.
- Detail-critical surfaces still need review. Labels, serial text, reflective watch cases, sheer fabric and dense patterns are where generated geometry fails. Human review remains mandatory for the final representation.
- Brand drift is real without constraints. Each fresh prompt re-rolls lighting, crop and styling. Without a saved brand reference system, fifty "AI studio" images read as fifty different brands — the exact inconsistency that costs marketplace sellers conversion.
- The review pass is your new line item. You trade studio day-rates for QC time. Budget 1–2 minutes of review per generated image at pilot scale; collapse it with saved visual standards.
Where AI genuinely wins on cost: long-tail SKUs the studio day never covered, colour and size variants of one master product, regional aspect-ratio exports (Shopee 1:1, TikTok Shop 1:1, 3:4 or 9:16 story crops), A/B campaign backgrounds, and seasonal re-contextualisation of last year's hero shot.
The Hybrid Workflow: Studio Master, AI Variants
This is the workflow that actually scales for SEA brands running both a flagship and a deep catalogue:
Step 1 — Shoot masters, not scenes. Commission one studio session per product family covering the accuracy-critical angles: front, back, side, one macro detail, one flat-lay. White background, even light, no styling drama. You are buying ground truth, not finished ads. Typical cost: the $20–$50 per-product volume tier (approx. SGD 25–70).
Step 2 — Lock a written visual standard. Before any generation, document the fixed elements: approved background tones, light direction, shadow softness, camera height, crop, permitted props. "Warm neutral seamless, soft key from left, 15° above product, product fills 60–70%, no text" — not "premium look".
Step 3 — Feed the brand system, generate variants. Load masters plus the visual standard into a brand-memory-equipped pipeline so every scene inherits the same lighting grammar. Batch-generate lifestyle scenes, campaign backgrounds and seasonal variants. One master, fifty contexts.
Step 4 — Review against the physical product. Keep the unedited master beside every generated export. Check silhouette, logo, label wording, colour under both D65 and warm light, component count. Reject and re-roll rather than patch — patched images teach your brand system to drift.
Step 5 — Version masters so formats are cheap. When a new channel appears (say, Shopee's optional 3:4 display), you regenerate exports from versioned masters instead of re-shooting the product. The studio invoice never comes back for a resize.
When Studio Still Wins
Be honest about the failure modes and use a real camera when:
- The product is optically hostile. Glass, mirror-finish metal, sheer textiles, dense repeats and fine embossed logos. Studio macro + focus stacking beats generation every time on these surfaces.
- One hero must be flawless. A launch campaign's centre image deserves a physical shoot; AI is excellent at volume, risky at 1-of-1 stakes.
- You need a model wearing a brand-new cut. Garment drape physics on unseen fabric still produce artefacts at sleeves, cuffs and hems — on-model studio for the flagship, AI on-model for size and colour variants once the master fits. If you are generating on-model variations, pair flatlays with a standardized reference pose using our free AI Model Pose Reference Library (100 commercial poses with ControlNet depth maps across 500 angles) to keep posture, camera elevation, and garment drape stable across the catalogue.
- Client or platform demands provenance. Some tender packs and insurance-grade catalogues require EXIF-captured assets; generated files don't qualify.
Cost Math on One Realistic Catalogue
A Jakarta-based accessories brand, 120 SKUs, 60 need lifestyle scenes, 120 need clean white-background covers, 30 need on-model shots:
- Full-studio route: ~$2,700–$5,400 (approx. SGD 3,600–7,200 equivalent regional pricing), plus a 4–6 week logistics loop, plus every future variant re-opened on a re-shoot quote.
- Hybrid route: one week of masters (120 product front/back/detail, negotiated volume tier), then AI generation for variants, scenes, and the 30 on-model shots from a single approved garment master. Typical total: 40–60% below the full-studio route with the same or broader asset coverage — and the marginal cost of the next campaign drops to near zero.
- Full-AI route: cheapest, but without studio masters the accuracy debt shows up later as inconsistent catalogue images, listing rejections and return-rate complaints. The real cost of inconsistency has its own article; read it before deciding.
Decision Checklist
Before your next creative budget cycle, score each SKU set:
- Is the surface optically hostile (glass, sheers, mirrors)? → studio.
- Is it a single 1-of-1 hero? → studio or studio-master + AI polish with review.
- Is it long-tail, variant-heavy, or resize-hungry? → AI from locked master.
- Will this product get 2+ campaign refreshes per year? → hybrid; the masters pay back by variant three.
- Can your team not review images at all? → buy the done-for-you studio pass instead of a raw tool subscription; unreviewed AI output scales brand damage faster than it scales content.
Frequently Asked Questions
Is AI product photography cheaper than a studio shoot?
Yes, on cost per usable image. Commercial studio rates in Southeast Asia typically run $25–$100 per SKU, plus reshoots when a detail is wrong. AI production cuts that by 50–80% for variants that do not need new physical photography — but only when the master shot already exists and packaging truth is preserved.
Can AI product photography replace a studio entirely?
No. AI generates variants around an existing product, so it cannot invent packaging that does not exist yet. You still need one real master shoot per SKU to establish accurate label text, container geometry, and true color. Everything after that master can be AI-generated.
What is the hybrid workflow?
Shoot each SKU once in studio, or on a smartphone rig for smaller catalogs. Lock that image as an immutable master node. Then generate background scenes, lighting changes, angle variants, and channel-specific crops in AI around that master, so the product itself never gets re-drawn.
When does a studio still win?
When the shot depends on real physical conditions: fabric drape on a model, liquid refraction in glass, intricate hardware that models smear, or a launch that needs art direction on set. Also when the product itself is not yet manufactured.
How do I keep AI-generated product photos accurate?
Treat the master packshot as ground truth and never let the model redraw it. Generate the environment around the product instead of prompting the product itself. Check label text, barcode, and color against the physical item before publishing.
Put the ideas to work.
Brief our studio. Get back finished, on-brand creative.
Talk to us ↗