Extracts the images embedded in a PDF, in the browser and at full resolution. JPEG streams are saved exactly as stored, so nothing is re-encoded; Flate, LZW and RunLength rasters are decoded and written as PNG with their transparency, covering DeviceGray/RGB/CMYK, Indexed, ICCBased and Separation colour spaces, 1–16 bit samples, Decode arrays and stencil masks. Several images are packaged in a ZIP. JPEG 2000, CCITT fax and JBIG2 streams are listed rather than dropped silently, and whole-page rasterising is not included — this pulls out the images a PDF contains.