> Signal can and should append some noise to the images when encrypted (or better yet, pad them to a set file size as suggested by paulryanrogers in a sibling comment) to mitigate the risks of this attack for those who do have threat models that require it
Adding padding to the image wouldn't do anything to stop this "attack". This is just watching which CF datacenters cache the attachment after it gets sent.
Right, my bad on the ambiguity—I was replying to the OP's concern about image sizes, not the attack in TFA:
> It really doesn't matter that they can't see an image or video, knowing its size upfront or later (for example in response to a law enforcement request) is enough