LANCropper

Inspiration

I am an independent fashion designer running my own brand. Creating ecommerce product images is a major part of my work, but the traditional workflow is slow and fragmented: crop and edit on a computer, export files, transfer them to a phone, preview them, then return to the computer if the composition does not feel right.

I wanted to create a tool that lets creators build product images directly on the screen where customers will actually see them.

What it does

LANCropper is a phone-first photo studio for ecommerce image creation.

Users can import and crop a photo, isolate people into editable layers, replace backgrounds, arrange multiple models, adjust color, create shadows, make local corrections, and export a finished composition. The result is a direct, visual workflow for creating product images in a mobile browser.

How we built it

LANCropper is a local web application built with Python, Flask, JavaScript, HTML, and CSS.

It uses AI-powered portrait segmentation to create editable cutout layers, while keeping high-resolution source assets on the computer for final export. The phone receives lightweight previews for responsive editing over a local network.

I used Codex throughout the project to turn my real creative workflow into features, debug image-processing issues, improve mobile performance, and iterate on the interface.

Challenges we ran into

The hardest challenge was balancing three things at once: accurate cutouts, high-resolution exports, and smooth phone performance.

High-resolution images can quickly consume mobile memory, especially when multiple editable layers, masks, and shadows are involved. We separated lightweight mobile previews from original-resolution export processing so that editing stays responsive without sacrificing final image quality.

Another challenge was edge quality: hair, translucent fabric, dark shadows, and light-colored backgrounds can all create visible halos. This required repeated iteration on masks, edge refinement, and manual correction tools.

Accomplishments that we're proud of

We transformed a repetitive desktop-to-phone workflow into a practical phone-first editing experience.

The project supports multi-person compositions, editable layers, background replacement, local retouching, color controls, shadow controls, and high-resolution export. Most importantly, it was shaped by a real working fashion-design process rather than an abstract editing demo.

What we learned

We learned that “mobile-first” is not only about fitting controls onto a smaller screen. It means designing around the moment when a creator can actually judge the final image.

We also learned that image-editing tools need to communicate visually and immediately. Every control must feel predictable, responsive, and connected to the final output.

What's next for LANCropper

Next, I want to improve the precision of AI cutouts, especially around hair and difficult edges, and make edge refinement more reliable in the live preview.

I also plan to add reusable product-image templates, smarter composition suggestions, and a simpler workflow for creators who need to produce consistent ecommerce images quickly.

Built With

• birefnet
• codex
• gpt5.6
• javascrip
• python