The choice is about what information the model needs. For a clear text question, OCR can produce a compact text prompt. For a chart, diagram, equation layout, or visual puzzle, the image itself may be essential.
Use OCR for clear text
LuciQA uses a local Tesseract worker with English and Turkish recognition data. It reads the selected region on your device and sends the recognized text to your chosen provider. This can avoid sending image pixels, but recognition errors are possible. Check negation, numbers, symbols, and answer choices before trusting an explanation. The current OCR configuration should not be described as full Japanese, Korean, or Russian OCR support.
Use Image for visual structure
Image mode sends the selected PNG crop directly to configured vision-capable providers. It preserves diagrams and spatial relationships that plain text may lose. Image input can have different costs or limits; check your provider’s current terms. Some LuciQA providers automatically switch to their image-capable model for that request.
In Pro Multi
Multi suggests regions using local OCR. Review and adjust each box and choose OCR or Image for each question. It can process up to 12 questions, two at a time, with separate requests. Several questions multiplied by several providers can use substantially more quota.
Whichever mode you use, AI answers can be wrong. Verify important claims against a reliable source and respect rules for the material you capture.