Smartling Image Translation allows you to translate text in flattened images, even without the original design source files. Simply upload a PNG, JPEG, or WebP image, and Smartling automatically extracts the text. You can then request translation using any type of workflow, just like any other file type. Once translation is complete, Smartling produces a fully translated version of the image that closely mimics the original layout and style.
This is especially useful when you have stock images, screenshots, UI captures, or other flat-file images with text that needs to be localized, such as product screenshots in your help center or app store listings.
Image Translation is currently in beta. Results may vary. Your account is limited to 50 flat images, shared across all users in your account. Each translated image file generated counts toward this limit.
Supported file types
Image Translation supports the following file formats:
- PNG (.png) (transparent backgrounds are unsupported)
- JPEG (.jpg, .jpeg)
- WebP (.webp) (regular, with no transparent background and no animation)
If you have access to the original source file from a design tool, use the appropriate Smartling integration instead:
- For Figma files, use the Smartling Figma Plugin.
- For Adobe InDesign files (.indd, .idml), upload them directly. Smartling parses InDesign files natively, or use the Adobe InDesign (INDD) plugin.
- For PSD or other layered formats, export to a supported flat-file format (PNG, JPEG, or WebP) before uploading, or use the Adobe Photoshop Plugin.
Image Translation does not extract or translate images embedded within other file types such as PowerPoint, PDF, or Figma files. To translate images, upload them as standalone files.
The maximum supported file size is 5 MB, and the maximum resolution is 4000 x 4000 pixels.
How it works
When you upload an image for translation, Smartling processes it through three stages:
- Text extraction: Smartling automatically detects all text in the image, along with its position, font style, size, color, and alignment.
- Background reconstruction: The original text is removed from the image, and the background behind it is cleanly restored to create a blank canvas for the translated text.
- Translated text rendering: The translated text is placed back onto the image in the same locations, preserving the original formatting and layout as closely as possible.
The result is a translated image that closely matches the original, just in a different language.
Requesting images for translation
There are two ways to get an image into Smartling for translation, depending on whether you're adding it to a job that already exists or creating a new job for it.
Adding an image to an existing job
Use this method to upload an image and preview its extracted strings before deciding which job to add it to.
- From within your Smartling project, navigate to the Files tab.
- Click Image Translation, in the upper right of the file list.
- Click Upload images.
- Drag and drop your image file, or click Select File, then click Upload.
- Once Smartling finishes extracting the strings, you'll see the file name and string count with a green checkmark. From here you can:
- Click Add File to a Job, then select a job from the Job dropdown, to add the extracted strings to that existing job.
- Click Close to leave the image unassigned for now. The image is saved to the project's Images list, and you can open it any time and click Add to a Job to assign it later.
This upload flow can only add an image's strings to a job that already exists. It cannot create a new job at the same time. If you want the image in a brand new job, use the method below instead.
Adding an image to a new job
To include an image in a job you haven't created yet, create the job first, then upload the image to it directly:
- Navigate to the Jobs tab and click Request Translation, which prompts you to enter a job name and drag and drop your image file, or
- Upload the image file directly to the Files tab, then click Request Translation to add it to a new job.
For step-by-step instructions, see Create a Job / Request Translation.
Translating images
Once added to a job, the image strings behave like any other content in Smartling. This means you get the same capabilities you use for all other content types (for example, translation memory, glossaries, AI and machine translation, CAT Tool quality checks). You can authorize the content for translation in any workflow, set due dates, and download a translated version of the file once the job is complete.
Static visual context is automatically generated so that as translators work in the CAT Tool, the original image is available to help them understand where each string appears. A preview of the translated image is not available in the CAT Tool. To see the final translated image, you will need to download the completed file.
Downloading translated images
You can download a translated image from the job, the same way you would any other file type, or from the Image Translation page, which lets you preview exactly how the translated image will look before you download it.
To preview and download from the Image Translation page, navigate to your project's Files tab, click Image Translation, then click the image you want. Each target language the image has been translated into appears as its own card showing translation progress.
From a language card, you can:
- Click the download icon to download that translated image.
- Click the ••• (more) icon, then View Strings, to see the strings in the Strings View.
- Click the ••• (more) icon, then Generate, to regenerate the translated image (for example, if the result doesn't look right, you can try again).
The downloaded file is a flat-file image (the same format as the original) with the translated text in place, ready to use in your help center, app store listing, or other content. It is not an editable design file. If you need to adjust the translated image, download it and edit it in an external design tool.
For instructions on downloading files from the job, see How to download translated files.
Best practices
To get the best results from Image Translation, keep these tips in mind:
- Use clean, simple images. Screenshots with standard UI text, solid or simple backgrounds, and clear typography produce the best results.
- Avoid heavily stylized or custom fonts. Smartling detects the font family (sans-serif, serif, monospace, or cursive), size, and color from the original image and applies a matching font in the translated version. The exact font is not matched, so standard fonts translate more reliably than brand-specific or decorative typefaces. For non-Latin scripts, appropriate fallback fonts are applied to ensure the text renders correctly.
- Check images with dense text. Images with a lot of text content may need review after translation, since translated text can be longer than the source and may need to fit within the same layout space.
- Review translated images. Smartling preserves the layout and styling as closely as possible, but it's important to review the final output, especially for images with complex layouts or multiple text regions.
- Use source files when available. If you have access to the original Figma, InDesign, or PSD file, using a Smartling design application plugin will give you an easier way to incorporate a Desktop Publishing (DTP) process. Image Translation is designed for cases where you only have the flat-file image.
Current limitations
Image Translation works best with screenshot-style images that have clean layouts and standard typography. The following are not currently supported:
- Large, complex tables
- Complex or photographic backgrounds behind text
- Stylized text (for example, WordArt, multi-color fonts, text effects)
- Embedded charts, graphs, or diagrams
- Angled, curved, or rotated text
- Handwritten text or annotations
- Overlapping text and images
- Watermarks
If your image falls into one of these categories, consider using the original design source file with the appropriate Smartling plugin, or working with your localization team on a manual DTP process.
FAQ
Can Image Translation work with images embedded in other files?
No. Image Translation works with standalone image files (PNG, JPEG, or WebP) uploaded directly to Smartling. It does not extract or translate images embedded within other file types such as PowerPoint, PDF, or Figma files.
Can Image Translation translate images on a translated website (GDN) or images found in connector content?
Image Translation is available only for standalone image files uploaded directly to Smartling. It is not integrated with the GDN or connectors at this time, but the team is continuing to develop the feature's capabilities.
Is the downloaded translated image editable?
No. The downloaded file is a flat-file image in the same format as the original. It is not an editable design file.
How does Image Translation handle fonts?
Smartling detects the font family (sans-serif, serif, monospace, or cursive), size, and color of the original text and applies a matching font in the translated version. The exact font is not matched, so the translated image may use a different typeface within the same font family. For non-Latin scripts, appropriate fallback fonts are applied to ensure the text renders correctly.
Can the formatting of the translated image be adjusted in Smartling?
Not yet. The translated image is generated automatically and cannot be edited. This capability is coming soon. For more control over the translation output, use the original source file with a Smartling design application plugin.
Can translators preview the translated image in the CAT Tool?
Translators can see the original image as visual context while working on the extracted strings. A preview of the translated image is not yet available in the CAT Tool. To see the final result, you will need to download the completed file.