Image Describer

Type "warm, moody, cinematic" into an image generator and you get a sunset from somebody else's holiday. Image Describer goes the other way: it reads an uploaded picture or a pasted image link and writes the prompt from what the picture shows. Four prompt modes are on offer, one of them shaped to the format Midjourney expects.

Besides prompts, an upload can come back as alt text for screen readers or as printed text lifted from a scan. Other modes write style notes, object lists, social captions and search-friendly text, in an output language you choose.

The batch page runs up to 100 images with the same settings, and the output downloads as a file for a spreadsheet or a website's content system. Finished results stay in your own browser for seven days, and the privacy policy says uploaded images are not used to train public models.

A separate practice page, which states that it is independent of Pearson, is built for the Describe Image task in the PTE English test. It turns a chart or a photo into one sample answer of 70 to 90 words, with a word count and a timer.