Selection refinement API for photo editing plugins

The moment happens inside a developer's sprint: they're building a plugin or a web-based image editor and they need selection or background removal that actually works on complex edges, but every off-the-shelf API they test produces the same jagged results users complain about in the big editors — 'rugged edges and less pixel-aware than desired,' 'not sensitive enough to people/object outlines.'

The gap exists because the major photo editors have no incentive to expose their best segmentation quality via API — it would commoditize their core product. And the generic background-removal APIs that do exist are optimized for the easy case (person against plain background) and marketed to non-technical buyers, not developers who need fine-grained control over tolerance, edge feathering, and output format.

What developers specifically need and can't get today is: programmatic control over edge sensitivity (the complaint that tools aren't sensitive enough to color and tone changes), the ability to specify regions to include or exclude without a GUI, and deterministic output quality — not the 'inconsistent' behavior that users describe as maddening. Current developer-facing tools return a flat PNG with a baked-in hard edge; there's no way to retrieve a soft mask with transparency data that the downstream application can further refine.

This is a business because every image editing product, mobile app with a 'remove background' button, and e-commerce CMS eventually needs this, and they need it repeatedly as they add new content types. The developer pays once in integration time but pays ongoing in API calls — the more successful their product, the more they pay, which is the right alignment.

What to build

Build a REST API that accepts an image and optional hint regions (bounding box, color sample, or rough polygon), returns a 16-bit soft mask as a separate channel alongside the composited PNG, and exposes per-call parameters for edge feathering radius, gradient color tolerance, and hair/fur detection sensitivity.

Where to start

Target Shopify app developers first — there are hundreds of apps in the Shopify ecosystem that need product image cleanup, the developer audience is concentrated and reachable through a single marketplace, and the use case (product on white background with complex edges like fabric or jewelry) is well-defined enough to nail quality quickly.

The hard part

Developers will benchmark you immediately against existing background-removal APIs on a standard test set — you need to demonstrably outperform on the hard cases (hair, fur, gradients) before you can charge a premium, and that requires a training dataset and inference infrastructure that takes months to get right.

How it makes money

Usage-based pricing per API call, tiered by image resolution; volume discounts at 10k and 100k calls per month; a flat monthly minimum for teams that want SLA guarantees.