Automated Testing of a Digital Image Processing Software

Summary

A global digital imaging software company was keeping a poorly documented legacy image-editing tool marketable while building a brand-new version on modern technologies with CI/CD. It needed a reliable QA partner to ensure quality across both. INNERLUXES took over testing for the legacy and the modernized products, combining manual and automated testing — automating regression and cross-platform checks for the new version under a unified Windows/macOS interface, mocking the image engine, and wiring everything into the CI/CD pipeline, while handling what could not be automated manually.

About the Customer

The Customer is a global company in digital image editing and quality measurement for mobile phones, tablets, and digital cameras, with a rich portfolio of patented intellectual property. Its products enhance the user experience across all aspects of digital imaging, including processing, storage, printing, and display.

Two Product Versions to Keep at High Quality

The Customer had a legacy tool for professional digital image editing that runs on both Windows and macOS, working as a standalone application and as a plugin for Adobe Photoshop, Adobe Lightroom, Adobe Aperture, and similar apps. The tool was poorly documented and based on old technologies, yet the Customer wanted to keep it marketable by fixing bugs and adding features — while simultaneously writing a completely new UI on contemporary technologies, with CI/CD for sustainable, on-demand releases. It needed a reliable contractor to own the QA for both the legacy and the new versions.

An Integrated Manual-and-Automated Testing Approach

INNERLUXES took over the smooth, consistent testing of both versions using an integrated manual-and-automated testing approach, applying automation where it paid off and manual testing where it did not:

• Legacy version — because the legacy app used old technologies, most tests could not be automated, so automation covered only basic functionality checks and the rest were run manually.
• Modernized version: regression — most of the functional regression tests were automated.
• Cross-platform automation — Windows (PyWinAuto) and macOS (ATOMac) automation frameworks were elegantly combined under a single universal interface to cut testing time and volume.
• Integration tests — tests checking the connection to the image processing engine were automated; with no direct access to the engine, the team used a pseudo-interface to mimic its behavior.
• CI/CD — automated testing was integrated into the CI/CD pipeline to support continuous delivery.
• Manual where needed — integrations with Adobe Photoshop, Lightroom, Aperture, and others were tested manually, as each tool's specifics made universal, reusable tests impractical, and initial, one-time, and hotfix checks were also run manually.

Loyalty Retained and a New Audience Won

• By ensuring a seamless image fine-tuning experience, the Customer retained the loyalty of the legacy version's users.
• The new product quickly attracted a significant portion of a new audience.

Technologies and Tools

Spix, PyWinAuto, ATOMac, Behave (BDD approach), Python.