I have wondered on this and I have thought it is dependent on whether a bot is doing the banning or a human. If the former then the bot wouldn't be able to tell if pics are watermarked but if human they would easily link them together. On the other hand a bot would easily be able to notice links. I did randomise links with url shorteners. I got banned within a week with 50 accounts recently with watermark in the pics but correlation != causation. So it could well have been something else. I just dread having another ban wave so soon and having to fill out those fucking multi tier picture captures again and again. Would love to learn how to automate that newschool captcha shit. Thoughts?