Translating images with Lara (image-to-image)

Lara allows you to translate standalone image files by extracting and translating the text contained in the image and returning the translated content in the same image format.

This feature is useful when working with visual assets such as screenshots, graphics, or scanned images that are not part of a document.

Supported image formats

Lara supports the following image formats for image-to-image translation:

• AVIF

• BMP

• GIF

• HEIC

• HEIF

• PNG

• JPG

• JPEG

• TIF

• TIFF

• WEBP 

The same format uploaded is returned as output (for example, JPG → JPG, PNG → PNG).

How image-to-image translation works

When you upload an image for translation:

1. Lara extracts the text from the image using OCR

2. The extracted text is translated into the selected target language

3. The image is reconstructed with the translated text

4. The translated image is returned in the original format

The original layout and positioning of the content are preserved as much as possible.

Image translation quality

Image translation relies on OCR and image processing. Results may be affected by:

• image resolution and clarity

• text size and contrast

• image dimensions

Exact limits depend on the OCR engine and may vary by format.

How image-to-image translation affects usage

When translating standalone images, usage is calculated per image.

Each translated image consumes a fixed amount of 1,500 characters, regardless of the amount of text contained in the image.

This article is about: 

• translating standalone image files with Lara

• supported image formats and output behavior

• how image-to-image translation works

• limitations related to OCR and image processing

• how image translation billing works