OCR that reads the screen, not a photo of it
Local · On DeviceText locked in an image is text you cannot copy. Lumen extracts it locally with OCR, on your machine, so an error message in a canvas, a label in a video, a chart axis, or a line in a scanned document becomes selectable text you can paste.
Text extraction that runs on your device
The recognition happens locally rather than being sent to a server, which means the text is recognised in the same resolution you are looking at. That is why small type, dense tables, and code in a screenshot come through accurately.
It also means the capture never has to leave your machine to be read.
Where screen OCR actually saves time
A terminal error that has scrolled past. A dashboard you can only see. A label inside a video frame. A serial number on a part. A figure in a paper you cannot copy out of. A chart in a tool that renders to a canvas and offers no text layer at all.
In each case the text exists and is perfectly legible, but the interface will not give it to you. OCR gives it back.
Extracted text feeds straight into the other tools
Once the text is yours, it can be summarised, explained, or asked about directly. Select a region and Lumen can read it, recognise the text, and answer a question about it in one pass.