Arben Apura

Arben Apura’s public GitHub profile identifies him as a freelance full-stack developer in the Philippines, working on web and Android applications. His XianScan project applies that software-building background to a specific creative problem: translating comics while preserving the page as a readable visual composition.

XianScan combines speech-bubble detection, optical character recognition, language-model translation, image repair, and typesetting. Its local-first design puts several stages of a comic translation workflow in one studio, rather than asking the reader to move between unrelated tools.

A translator or comic enthusiast could use it to process a page, inspect extracted text, revise a translation, and rebuild the lettering. The project documentation explains its desktop setup and integrations. Local processing and cloud-backed translation options have different requirements, so the selected configuration determines where the work happens.

Apura is worth following for the way he assembles models into a usable end-to-end workflow. The most useful developments to watch are improvements in text placement, editing control, and language handling. A polished page still benefits from a human translator’s judgment about meaning, tone, and cultural context.