Tech & Appsbreaking news
Google Translate's camera has run for over a decade, and it still stumbles on menus
The camera mode is easy to use and works live, but diacritics and menus trip it up, and AI apps and built-in phone tools offer other routes.

A decade-old camera mode
The Google Translate app has offered a dedicated camera mode for over a decade, and it is available to anyone with an Android phone or an iPhone [1]. It opens a viewfinder that lets you point the phone's camera at anything in the real world [1]. It translates text in real time, so you do not need to capture an image [1].
How to use it
Install the app from the Google Play Store on Android or the App Store on iOS, open it, and tap the Camera icon at the bottom of the screen [1]. You can select any two of Google Translate's hundreds of languages for a translation [1].
If you cannot identify the language you have come across, the default Detect language option in the source field helps [1]. You can also download frequently used languages to translate text offline [1].
Where it falls short
The author found that Google Translate's camera struggled to identify diacritics in printed text while traveling in Vietnam [1]. Diacritics matter because a word can have two completely different meanings depending on whether its letters carry the correct diacritics [1].
The author also finds the app stumbles on restaurant menus [1]. Menus often use informal shorthand or a single word, which can leave the app without enough context for an accurate translation [1].
What AI apps do differently
The author found the output quality of AI apps for translation to be far and away better than Google Translate [1]. ChatGPT, Gemini, and Claude are named as AI apps that can translate text from images [1]. Modern AI models understand context, so sending a picture of a menu will typically yield food-related translations [1].
AI models also often understand the intended meaning of blurry or hard-to-read text based on surrounding text instead of replying with gibberish [1]. To use one, you upload a picture of the text and type a short instruction, such as a request to translate to English, before sending the message [1].
The trade-offs of AI translation
AI apps reply with translated text but cannot overlay the translation over the image as Google Translate does [1]. They often cannot translate in real time and require an internet connection [1]. ChatGPT's free tier limits users to a few image uploads every day, and processing can take a few seconds [1].
The author uses AI translation as a last resort when traditional translation methods fail [1].
Built-in options on your phone
Google Lens on Android can be used for translation without a dedicated translation app, and many phones include it out of the box [1]. Google Lens and Google Translate can recognize text inside an image and translate it [2]. Google says its camera translation works with both live camera views and previously captured images [2].
Recent iOS versions include a translation feature built into the camera app through the Live Text feature [1]. When the iPhone camera detects text, a yellow frame appears around it and a grid icon appears near the bottom of the viewfinder [1]. Tapping the grid icon and selecting Translate performs the translation [1].
Apple says Visual Intelligence on supported Apple Intelligence-enabled iPhones can translate or summarize text [2]. It requires a supported Apple Intelligence-capable iPhone and is not available on every model [2].
What this means for you
If you want live translation without taking a photo, Google Translate's camera mode does that, and downloaded languages work offline [1]. If the language is unknown to you, the Detect language default can help [1].
If you are reading menus or text with diacritics, the author's experience suggests checking the result, because the app can lack context for menus and can miss diacritics [1]. An AI app is an option when other methods fail, though it needs an internet connection and may limit free image uploads [1].
Translation may not always be perfect, particularly with tiny, stylized or unclear text, so common sense is needed when the meaning matters [2].
Sources
- How To Use Your Phone's Camera To Translate Anything — Engadget
- How to Identify Almost Anything With Your Phone — Gotta Be Mobile
Written from the sources above and checked twice before publishing. How we work · Report an error