Orisod
A PDF page is a list of drawing instructions, not a layout file
A PDF content stream contains operators for drawing text, paths and image resources. An extractor follows the image references and decodes the associated object. Images may also appear inside forms, masks or other nested structures, so coverage depends on how thoroughly the tool walks the document.
The original upload is not always recoverable
PDF supports several image encodings, including JPEG, JPEG 2000, CCITT, indexed color and raw sample data. Some embedded JPEG streams can be copied without decoding, but other cases must be interpreted into pixels and color space first. Once that happens, the extractor is producing a new image file rather than restoring the exact original bytes.
Why PNG specifically, and not always JPEG
Once an image is decoded into plain pixels, re-saving it needs a format that can hold exactly that data without a second round of lossy compression on top of whatever loss already happened during the PDF's own original encoding. PNG is lossless by design; it stores pixels as-is, however many of them there are, without re-guessing or discarding detail. Re-exporting as JPEG at that point would mean compressing already-decoded data a second time, adding new artifacts that weren't in the source. PNG being larger in file size is the trade-off for not doing that.
One object may be reused across several pages
A logo can be embedded once and referenced repeatedly, reducing the PDF size. An extractor may list that object once and report every page where it appears. Other tools may export each occurrence, so duplicate behavior is an implementation choice rather than a property readers can assume.
Limits worth knowing
Why didn't any images come out of my PDF? If the PDF was built entirely from text and vector graphics; shapes, lines, and gradients described as paths rather than pictures; there's no raster image data anywhere in it to extract, even if the page visually looks like it contains graphics.
Can extraction recover a higher-resolution version of the image than what's shown on the page? No; it can only recover the resolution that was actually embedded in the PDF. If the source image was already downscaled or heavily compressed before being placed into the PDF, that's the ceiling on what comes back out.
A scan is often one large image per page, so extraction may return the complete scanned page rather than individual photographs visible inside it.
Pull the images out of a PDF
Orisod’s Extract Images from PDF finds supported embedded pictures and exports them as PNG files, individually or in a ZIP, without uploading the document.
Extract images →The short version: a PDF doesn't contain a folder of image files waiting to be copied out; it contains instructions for painting pixels, and extracting an image means actually running enough of those instructions to get real pixel data back, then saving that data in a format built to hold it without loss. PNG isn't a limitation of any particular tool; it's the natural output of that process.