Online AI Image Translator

Three steps: OCR finds every line of text and its position, a large language model translates each line in e-commerce tone, and an image-editing model replaces the original text with the translation. The replacement keeps the original font style, size, color and alignment and rebuilds the background behind the old text, so the result looks as if it had been designed in the target language from the start.

16 languages: English, Japanese, Korean, German, French, Spanish, Portuguese, Brazilian Portuguese, Russian, Arabic, Thai, Indonesian, Vietnamese, Malay, Traditional Chinese and Simplified Chinese. That covers Amazon in the US, Europe and Japan, Shopee and Lazada in Southeast Asia, TikTok Shop and Mercado Libre.

No. Pure numbers, prices, model numbers and brand names are kept as they are; only the text that needs translating is translated. Units are written the way the target market expects, for example a decimal comma in German. If you need unit conversion such as centimeters to inches, use the text-editing skill from the "keep editing" row after translation.

Text-heavy images that are painful to re-layout: feature callouts on detail pages, spec comparison tables, size charts, step-by-step instructions, promo posters and A+ infographics. Main images normally must not contain text, so they do not need translation.

The translation is asked to stay close to the original length, and any overflow is wrapped or slightly resized to fit the original area, so spill-over is rare. If you want to reword a line, send the image to the text-editing skill from the generation page and change just that line.

One task translates into one language. For several marketplaces, run the same source image again with a different language from the result page. Batch multilingual export is in development.