Trusted by 1,000+ enterprise brands. Rated 4.6/5 on G2.
An image translator is software that finds the text inside an image file, translates it, and renders the translated text back into the image in place of the original.
Smartcat's image translator works in three stages. First, optical character recognition (OCR) detects each text region in the file — headlines, labels, captions, and text inside diagrams.
Second, each region is translated into any of 280+ languages, applying your translation memory and glossary so brand names and recurring terms stay consistent across every image.
Third, the background is removed and the translated text is re-inserted into the original layout, set in Smartcat's documented set of supported fonts and matched to the original as closely as that set allows.
Because translations often run longer than the source, every text box stays editable in the browser: adjust a line break or font size before you download the finished image in its original format.
How to translate an image in five steps
1
Upload your images — drag them 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 image.
2
Pick your languages — one source, as many targets as you need, from 280+.
3
Translate — Smartcat detects every text region, removes the background behind it, and re-inserts the translated text into your original layout as editable text blocks.
4
Review and adjust — click any text box to reword or resize it, or hand the file to an internal expert or a Marketplace reviewer if the content warrants one.
5
Download — the same image, same layout, new language. There is no limit on how many files go into one project: batch uploads of hundreds of images work the same way on every plan.
We love having total visibility over our entire translation lifecycle on our Smartcat workspace due to the centralized nature of the platform and being able to gather all the linguistic assets we need, including translation memories and glossaries.
”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 image for OCR
The constraint is per-file size, not file count — there is no cap on how many images go into one project, and it does not vary by plan. Scans need at least 300 DPI.
1,000
Smartwords per image
Image translation is metered at a flat 1,000 Smartwords per image rather than per word, so a dense infographic costs the same as a headline banner.
Text in an image can't be selected, so most people fall into the same loop: retype the words out of the JPG into a translator, paste the result into a design tool, and rebuild the graphic by hand — once per language. Annoying for one banner. For forty product images in six languages it is a week of work that produces its own errors.
The other failure modes are quieter. A screenshot “translated” by a free app gives you an overlay you can read but no file you can publish. Translated text runs longer than the original and overflows the button it lived in. And when different people translate different images, the same product name comes back three different ways across one catalog.
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 image, once, a free consumer app will serve you fine.
10 days → 2–3 days
Faster Turnaround
Smith+Nephew cut eLearning translation turnaround from an average of 10 days with their previous providers to two to three days for the same course length.
Up to 70%
Cost Savings
See how Stanley Black & Decker reduced spend while raising quality.
31 hours
Saved Monthly
Babbel's marketing and L&D teams reclaimed 31 hours a month.
Upload the images, pick the languages, download files you can publish — same format, layout intact, every text box still editable. Free for 15 days with 15,000 Smartwords, no credit card.
JPG/JPEG/JFIF, PNG, TIF/TIFF, BMP, GIF, DCX/PCX, JP2/JPC, DJVU/DJV, JB2, image-only PDF, and AI (Adobe Illustrator) are all processed via OCR. Animated GIFs are read first frame only. Each image can be up to 30 MB. HEIC and WebP are not on the supported list. If your text lives in a PDF — even a scanned one — use the PDF translator instead; it keeps multi-page structure.
Yes — any supported language pairs with any other, 280+ in total, so a picture with Japanese, Arabic, or Hebrew text translates to English the same way English translates out.
Yes, on the OCR pipeline. Each detected text region becomes a bounding box in the browser editor, and the formatting toolbar changes font type, size, colour, and alignment; you can also select text such as a logo and use Hide to keep it out of the translation.
One limit: the generative pipeline has no layer-editing tools (Move, Resize, Font Select), so choose the OCR pipeline when you need to move things.
Yes, and screenshots are a strong use case: add your UI terms to a glossary so button labels and feature names translate identically across every screenshot in your docs.
OCR accuracy falls with resolution. Sharp, straight-on images translate reliably; blurry or heavily compressed ones lose characters. Scans should be at least 300 DPI. See translate scanned documents for what scan quality changes.
No limit on how many files go into one project. Batch uploads of hundreds of images are supported; the constraint is per-file size (30 MB per image for OCR), not file count, and it does not vary by plan. See translate multiple images at once.
Image translation is metered at a flat 1,000 Smartwords per image, not per word — so a dense infographic and a short headline banner cost the same. Full details on the pricing page.
For 15 days, yes — 15,000 Smartwords with full access to translation capabilities, no credit card, and no free-forever plan afterwards. After the trial it is a paid plan.
Not always on the first pass — many languages run longer than English, and a text box sized for the source can overflow. That is why the output stays editable: adjust the line break or font size in the browser rather than reopening a design tool.
Translated text is re-inserted using Smartcat's documented set of supported fonts, matched to the original as closely as that set allows. The list lives in the product documentation and can change, so check there if a specific typeface matters. Custom font uploads are not part of the documented set today.
Smartcat is SOC 2 Type II compliant. Files are encrypted in transit and at rest, workspaces are isolated, and access is role-based. Details on the security page.
Two honest limits.
Only the first frame is processed — an animated GIF is handled as a still image, so you get one translated frame rather than a rebuilt animation. See the GIF translator page for what that means in practice.
They run through the same OCR pipeline, and results track image quality: hold the camera straight on, fill the frame with the text, and avoid glare. The photo translator page covers camera photos specifically.
If you need to read one sign, menu, or screenshot right now, a free app such as Google Lens is faster and it is the right tool.
Use Smartcat when you need the translated image file back. Lens gives you an overlay, one image at a time, with no glossary and no reviewer. Smartcat returns the same format with the layout intact, runs a batch as one job, enforces your terminology, and keeps review in the same workspace.
Book a demo — a 1:1 walkthrough with a specialist, no commitment.
1. Smartcat customer case study, Smith+Nephew. https://www.smartcat.com/cases/smith-nephew/.
2. Smartcat customer case study, Stanley Black & Decker. https://www.smartcat.com/cases/stanley-black-and-decker/.
3. Smartcat customer case study, Babbel. https://www.smartcat.com/cases/babbel/.