I’ve been using mat2, which is the first recommended option on PrivacyGuides.com, to remove metadata. Is there any drawback to this compared to ExifTool, which is at the bottom of the page? mat2 relies on ExifTool, but only for removing metadata that isn’t a PDF, audio file, image, svg, or video. I usually use mat2 for videos, which utilizes FFmpeg to remove metadata. I don’t think ExifTool uses FFppeg, but I could be wrong.
One weird thing, I probably shouldn’t, but I still use ExifCleaner, which is an outdated Electron-based drag-and-drop metadata removal tool. After I run a video through mat2 and then through ExifCleaner, ExifCleaner says there were 60 metadata entries on the video file, but it reduces it to 59. This means mat2 seems to not remove all possible metadata.
I haven’t used either tool but some metadata isn’t privacy related. Properties such as video size, video codec, audio codec, etc are important for playback.
For my personal use, the only drawback of mat2 is that for PDFs it turns pages into PNGs (https://0xacab.org/jvoisin/mat2/-/blob/master/libmat2/pdf.py), so you lose the OCR layer/searchable text from the original file. Since ExifTool changes to PDF metadata are reversible if you don’t linearize them (https://exiftool.org/TagNames/PDF.html), now I just use this script to safely clean and keep the output file searchable: https://gist.github.com/sneakers-the-rat/172e8679b824a3871decd262ed3f59c6.
I guess you could compare the output files from mat2 and ExifTool using the fc or diff commands, to find out what’s the difference
Exiftool will only process images, so not things like PDF.
