Trusted by 1,000+ enterprise brands. Rated 4.6/5 on G2.
To translate a picture, upload the file to Smartcat, choose English or any of 280+ target languages, and download the picture back with the text translated in place. Nothing is retyped, and the layout stays as it was.
It works in three stages. First, optical character recognition (OCR) detects every text region in the file — headlines, labels, captions, and text inside diagrams or screenshots.
Second, each region is translated into any of 280+ languages, applying your glossary and translation memory so product names and recurring terms come out the same way every time.
Third, the background behind the original text is removed and the translation is set back into the original layout, in one of Smartcat's supported fonts, matched to the original as closely as that set allows.
Pictures, images, and photos are the same input here — a file with text in it. What changes the result is the file: a designed graphic translates most cleanly, while a camera shot depends on capture quality.
How to translate a picture in five steps
1
Upload your picture — drag it into the browser, one or many at a time. JPG/JPEG/JFIF, PNG, TIF/TIFF, BMP, GIF, DCX/PCX, JP2/JPC, DJVU/DJV, JB2, image-only PDF and AI files are read by OCR, up to 30 MB per picture.
2
Pick your languages — the source is detected for you; choose English or any of 280+ targets, as many at once as you need.
3
Check what it read — the recognized source text is yours to review before you commit. Fix a misread word once and the fix carries into every target language.
4
Adjust in the editor — every text region is an editable box: reword it, resize a line, or select something like a logo and use Hide to keep it out of the translation.
5
Download the translated picture — same file format, same layout, new language. There is no cap on how many pictures go into one project, and it does not change by plan.
We've been able to increase our productivity by 50% while reducing our outsourcing costs by 50%. This has had a huge impact on our bottom line.
”Read the case study →
280+
Languages
Any supported language pairs with any other, so a picture with Japanese, Arabic or Hebrew text translates to English the same way English translates out.
30 MB
Per picture for OCR
The constraint is per-file size, not file count — there is no cap on how many pictures go into one project, and it does not vary by plan. Scans need at least 300 DPI.
1,000
Smartwords per picture
Picture translation is metered at a flat 1,000 Smartwords per file rather than per word, so a dense infographic costs the same as a headline banner.
If you are standing in front of a menu or a road sign, your phone is the right tool. Google Lens and camera apps read a picture and show you the meaning instantly, and this page is not trying to replace that.
The job changes the moment you need the picture itself. An overlay on your screen is not a file you can publish, send to a client, or archive. Translated text runs longer than the original and overflows the button it lived in. And when different people translate different pictures, one product name comes back three different ways.
for ease of setup
for ease of use
enterprise brands
average rating on G2
Free for 15 days with 15,000 Smartwords and full access to translation capabilities, no credit card — then a paid plan, with no free-forever tier. Details on the pricing page. If your need is one picture, once, a free consumer app will serve you fine.
+50% / −50%
Productivity and Outsourcing Cost
expondo raised productivity by 50% while cutting outsourcing costs by 50% across 18 countries.
+30%
Project Output, Same Resources
Wunderman Thompson increased project output by 30% using the same resources.
90 min → under 5
Content Turnaround
Kids2 cut ecommerce content turnaround from 90 minutes to under five minutes per upload.
This page is for the everyday job: one picture, translated properly. If yours is bigger — product catalogs, campaign kits, whole folders of visuals in many languages — the AI image translator is the same engine with batch uploads, glossaries and review built around it.
Yes. Upload the picture in your browser, choose English as the target, and download the result — the layout stays as it was, with the English text set where the original appeared. It is free for 15 days with 15,000 Smartwords and no credit card; picture translation is metered at a flat 1,000 Smartwords per file, however much text it holds.
You do not retype anything — recognition is the first step of the product. Smartcat detects every text region in the picture, shows you what it read, and translates from there. The one manual step worth taking is a quick look at the recognized text, because a misread character carries into every target language.
JPG/JPEG/JFIF, PNG, TIF/TIFF, BMP, GIF, DCX/PCX, JP2/JPC, DJVU/DJV, JB2, image-only PDF and AI (Adobe Illustrator) are all read by OCR, up to 30 MB per picture. HEIC and WebP are not on the supported list. There is no cap on how many pictures go into one project, and that does not change by plan. If your pictures sit inside a PDF, use translate PDFs with pictures instead.
Same engine, different jobs. This page covers the everyday task — one picture, translated properly. The image translator is the business-scale version: batch uploads, glossaries and brand consistency across hundreds of files. The photo translator handles camera-specific problems such as glare, skew and handwriting in phone photos.
For scans: 300 DPI or better, straight, evenly lit, with clear contrast. For screenshots, designed graphics and camera shots there is no DPI to aim at — the practical bar is that a person can read the text at 100% zoom. Blur, heavy compression and low-contrast text degrade recognition, and recognition errors carry into the translation.
Yes. Each detected text region becomes a box in the browser editor — change the wording, font size, colour or alignment, or select something like a logo and use Hide to keep it out of the translation. The translation is set in one of Smartcat's supported fonts, matched to the original as closely as that set allows.
Usually, with one honest caveat: translations change length. The translation is re-set into the original layout automatically, and where a longer language crowds the space you adjust the line in the editor before downloading. You see the result before you use it — nothing is finalised behind your back.
Yes — the flow runs in both directions and into several languages at once. Any of 280+ supported languages pairs with any other, so one English source picture can come back as German, French and Japanese versions in a single pass.
Clean print recognises well; handwriting is genuinely harder and results vary with legibility. Check the recognised source text before trusting the output. For handwritten notes photographed on a phone, the photo translator page covers what to expect in more detail.
Uploaded files are processed inside your Smartcat workspace and are not shared outside your account. Smartcat is SOC 2 compliant; the security page carries the current certification details.
1. Smartcat customer case study, expondo. https://www.smartcat.com/cases/expondo/.
2. Smartcat customer case study, Wunderman Thompson. https://www.smartcat.com/cases/wunderman-thompson/.
3. Smartcat customer case study, Kids2. https://www.smartcat.com/cases/kids2/.
4. Smartcat Help Center, Understanding Smartwords. https://help.smartcat.com/understanding-smartwords/.
5. Smartcat pricing and free trial terms. https://www.smartcat.com/pricing/.