General AI image generators produce pretty pictures with unreadable, invented labels. That is the gap PaperFig tries to close for people who tried general image generators. It is an AI scientific figure maker for researchers. Labels stay editable: the built-in OCR reads them back and lets you correct them in place. The focus is research figures such as mechanisms, workflows and graphical abstracts. Core downloads are PNG; a purchase removes the watermark, and paid plans add TIFF/PDF export and a 4K upscale for final images. It is not limited to one field; the gallery runs from biomedical and materials work to energy, environmental, bioengineering, CS/AI and geoscience. The typical case is getting garbled text and decorative images from general AI art tools. The free plan needs no card and includes 150 credits at signup, plus a free figure each day for the first week. Visit PaperFig.