← Nexus Automation

Product studio pipeline

One product photo. Ready for its close-up.

Turn a phone photo into six consistent studio angles, with product copy ready for your store. Your next collection, without another photoshoot.

6 studio angles. Under 2 minutes per product.

Same product. A whole new first impression.

The project

Product Studio Pipeline

The problem

The client had 1,200+ raw, unedited cap photos and wanted every one turned into professional studio photography. Hiring it out the traditional way, with a product photographer, a studio rental, and a graphic designer per angle, runs ~$60-80 and a 2-3 day turnaround per product. At 1,200+ products, that's six figures and months of backlog before a single listing goes live.

The build

A fully custom n8n pipeline turns a single raw product photo into a complete, store-ready e-commerce listing in under 2 minutes: 6 consistent studio angles generated with Nano Banana 2, AI vision extracting exact colors and materials, and a full SEO package written automatically, all landing as one structured Airtable record. The prompts, angle set, and background style are fully customizable to the brand's own look, and it's a system he now owns and can run himself on every future drop, with no photographer, designer, or multi-day turnaround required.

The outcome

6 studio angles per product in under 2 minutes, at $0.26 per product versus ~$70 and 2-3 days manually. A 99% cost cut and same-day turnaround, on a fully customizable system he can run himself indefinitely.

6Studio angles from 1 raw photo
<2mPer product vs 2+ hrs manually
$0.26Cost per product vs ~$70 manual
99%Reduction in production cost

How it runs

  1. Drop one product photo

    A single reference photo lands in a Google Drive folder-per-product convention, and the folder name itself drives the product naming and SEO downstream.

  2. Generate 6 studio angles

    Kie AI (Nano Banana 2) generates Angled, Front, Side, Back, Underpeak, and Overhead angles with uniform framing, lighting, and a pure white background, via an async webhook callback architecture that tracks each angle's task state independently.

  3. Analyze with AI vision

    GPT-4.1 vision inspects the product and extracts precise colors (with hex codes), materials, stitching, and construction details. That same analysis drives both image accuracy and the SEO copy.

  4. Write the SEO package

    Meta title, meta description, URL slug, per-image alt text, and keyword tags are generated automatically for every product, ready to push into Shopify.

  5. Store, dedupe, and ship

    Every product lands in Airtable as one record with all 6 images and metadata. Registry-based deduplication makes it safe to run on a CRON schedule over a live folder without ever reprocessing a product.

Proof, not promises

The actual Airtable staging table mid-run: each angle tracked independently (pending → done) until all 6 finish, then the SEO package (title, meta description, slug, alt text, keywords, tags) is written automatically per product. Nothing here is mocked up.
The actual Airtable staging table mid-run: each angle tracked independently (pending → done) until all 6 finish, then the SEO package (title, meta description, slug, alt text, keywords, tags) is written automatically per product. Nothing here is mocked up.
The input: one raw, unedited phone photo.
The input: one raw, unedited phone photo.
Generated studio angle — Angled. Uniform framing, lighting, pure white background.
Generated studio angle — Angled. Uniform framing, lighting, pure white background.
Generated studio angle — Front.
Generated studio angle — Front.
Generated studio angle — Side.
Generated studio angle — Side.
Generated studio angle — Back.
Generated studio angle — Back.
Generated studio angle — Overhead.
Generated studio angle — Overhead.
Generated studio angle — Underpeak.
Generated studio angle — Underpeak.

In short

2,000+ photos? No problem. It works while you sleep.

n8nKie AI (Nano Banana 2)OpenRouter (GPT-4.1 Vision)AirtableGoogle DriveShopify
Have a process worth automating? →