How to generate colorway product photos for every SKU without reshooting

Colourway reshoots are the silent calendar-killer of catalog teams. Twelve colours of the same shirt is twelve shoot days at agency rates. With one source photograph and a hex map, AI compresses that work into one upload session — and the colour output is closer to reality than most retouchers ship.

This is the workflow we run for clients who launch six to thirty variants per SKU. It works for fashion, accessories, electronics with painted or anodised housings, home goods with finish variants, and any category where the same form exists in many colours. It does not work for products where the colour also changes the geometry — embroidery patterns, intentionally different prints, structural finishes. That distinction matters.

Why reshooting variants is the wrong unit of work

A photographer's day rate covers a finite number of hero-quality frames — fifty, generously. A twelve-colour drop with eight angles per SKU is 96 frames. Even at agency pace, that is two full days plus a retouching week. The economics broke years ago, but the workflow stayed because nothing else was reliable enough. AI colourway generation crossed that reliability threshold in late 2025.

How AI colorway generation actually works

There are two production-grade methods, and AngleForge uses a hybrid. Method one — hex mapping — identifies the colour-bearing surfaces of the product in the source photograph and re-tints them to the target hex value while preserving the source's lighting, texture, and specular highlights. Method two — material-aware rendering — uses category-specific models that know how silk reflects differently from denim, how anodised aluminium catches edge light, how glass tinting changes transparency. The hybrid uses hex mapping for surfaces the model is confident about and material-aware rendering for the rest.

Where AI gets it right

• Solid-colour woven fabrics — cotton tees, linen shirts, crew sweatshirts. Hex mapping handles these almost perfectly.
• Anodised or painted electronics housings — iPhone-style finishes, controller shells, headphone cups.
• Plastic and resin homeware — water bottles, kitchenware, organisers.
• Smooth leather goods — wallets, card holders, simple bags. Texture survives the recolour because grain is locked to lighting, not pigment.
• Coated metal furniture and hardware — chair frames, light fittings.

Where AI fails (and what to do)

• Knit garments where the colour also implies a different yarn — heathered marl vs solid. The AI tints the colour but the heather structure goes flat. Workaround: shoot a heather source separately.
• Print fabrics — the AI treats the print as colour-bearing surface and re-tints the pattern as a whole. For prints, shoot each colourway as its own source.
• Iridescent or duochrome finishes — the AI cannot infer the second colour shift if it isn't visible in the source. Shoot two-angle sources.
• Translucent materials — frosted glass, tinted plastics, smoked acrylics. Material-aware rendering is improving here but not consistent. Manual check required.
• Footwear with structural colour blocks — the panels are individual surfaces; the AI sometimes bleeds colour across panels. Use a generator with panel-mask awareness (AngleForge's fashion preset has it).

The twelve-colourway workflow, step by step

1. 01Shoot one hero source. Flat, dead-front, even diffused light. The lightest colourway you sell is the ideal source — recolouring is more accurate going darker than going lighter.
2. 02Define your hex map. List every variant as a hex code (`#0F1A2E navy`, `#A82C2C wine`, `#E6D5B8 oat`). Source these from your actual fabric / paint specs, not from your website palette.
3. 03Upload one source to AngleForge, select the 'Colourway batch' preset, paste the hex map, choose the angle set.
4. 04The forge renders every angle × every colourway in parallel. For a six-angle × twelve-colourway listing, expect roughly 90 seconds.
5. 05Pull the contact sheet. Run the four-frame audit (next section) on every variant. Re-roll any frame that fails.
6. 06Export marketplace-organised bundles — Shopify variant images, Amazon variant set, Etsy individual SKUs.

The four-frame variant audit

1. 01Pure-colour spot check — does the rendered colour match your physical sample side-by-side under daylight? Tolerance: ΔE under 3 to the eye.
2. 02Highlight integrity — does the brightest point of the variant still look like the same product photographed under the same light? If specular hotspots vanish, the AI lost the lighting grammar.
3. 03Shadow integrity — does the deepest shadow read as form, not as colour bleed? Watch for muddy 'middle tones' where the AI didn't commit to a colour.
4. 04Edge crispness — examine seams, panel joins, and edges at 100% zoom. AI sometimes softens edges on darker variants; re-render with a higher fidelity preset if so.

Shopify variant image assignment

Once you have the bundle, assign each variant image to its Shopify variant in admin. Two non-obvious tips: tick 'Use as featured image for variant' for the front-facing angle only (Shopify uses this in collection thumbnails); and tag your variants with the actual hex value or material name in the variant title — Google Shopping reads this for filter facets.

Where colorway AI fits in PIM workflows

If you run a PIM (Akeneo, Pimcore, Plytix, Salsify), AngleForge's batch export drops into the standard 'variant image' field for each SKU automatically when you upload via the bulk endpoint. The bulk endpoint accepts a CSV of SKU × hex × angle, and returns a CSV of generated asset URLs. Most catalog teams wire this into a nightly job: photoshoot weekly, variant fan-out nightly.

Cost arithmetic

Traditional reshoot for one twelve-colourway × six-angle SKU: roughly $1,800 (studio day, stylist, retoucher) at agency rates. AngleForge equivalent: 72 credits, ~$15 on the Pro plan, generated in ninety seconds. The arithmetic gets better with scale; for a brand running fifty SKUs through a launch, the saving compounds to mid-five-figures per drop.