Text Lens is a mobile translation app designed to make translating real-world text faster and easier. Users can translate using their camera, images, voice, or typed input — all in one app.
Most translation apps work well for typed text, but struggle with real-world use cases like:
- translating text from images
- switching between multiple apps
- handling complex layouts (like manga)
This makes translation slower and more frustrating than it should be.
Text Lens combines multiple input methods and translation tools into a single, simple interface:
- Camera translation (live + photo capture)
- Image translation with OCR
- Voice and text input
- Multiple translation engines (Google, DeepL, offline)
- Save and organize translations
During testing, we found that vertical Japanese text (like manga) was translated very poorly by standard OCR systems.
To address this, we implemented Manga Mode, which improves detection and grouping of vertical text, especially inside speech bubbles.
- 12 users opted in for testing
- Peak of 9 active users
- Users prefer fast, simple workflows
- Camera-based translation is the most used feature
- Specialized cases (like manga) require custom solutions
Feedback helped us improve accuracy and simplify the interface.
Text Lens aims to make translation:
- faster
- more accessible
- easier to use in real-world situations
- Improve camera translation UI
- Increase translation accuracy
- Enhance OCR for complex layouts
- Filter raw text noise for better translations
https://youtube.com/shorts/m6dlDMFUsig
This project highlighted the importance of:
- user testing early and often
- identifying real-world edge cases
- iterating based on feedback
It also showed that small, focused features (like Manga Mode) can significantly improve user experience.