I still think you're grossly mischaracterizing the inline option. Inlining does not force the browser to download the image, or to display it even if it does. The browser issues a HTTP GET and obtains the HTML of a page and renders that page. This may or may not involve the browser issuing more HTTP GETs in order to obtain the binary of the images referred to in IMG tags embedded in the HTML. It's entirely under the control of the browser and thus, subject to browser design and selection, under the control of the user. So users *already* have the option of downloading only those images that they want to see. In the worst case they can run lynx in a cmd window (Windows) or xterm (MacOS, BSD, Linux). In addition, users can control their stylesheets (either by the on-site override or using their browser controls) and use this to suppress image display. You may say "until", I say that what you propose is available to our users *now*. "Oh but it's too complicated", I hear you say. "Our users are Moms and Dads, not geeks." Well people will say precisely the same thing when you have your on-site image controls. Even a straightforward "reload without images" button on the top nav will be too complicated for some people; as for selecting whether they want to see objectionable images or not, it will have to be hidden away in some page in special: and it will be every bit as hideous as anything in the much maligned Internet Exploder. What's more, its behavior will change over time, probably at a fairly rapid rate as category hierarchies are changed and the feature evolves.
But how do I know there is an image there in advance? Strangly no matter how much I mess around with my browser it does not appear to have a setting for "don't display images of sex acts, shocksites and graphic injuries". I can't block an image until I see it. Btw it's not just mums and dads who will have I problem there are bits of your post I don't understand.