matrix_jan - 7:18 pm on Feb 3, 2013 (gmt 0)
Here's what I did and so far it looks good.
1. Only few user agents are allowed to crawl my images (googlebot, msnbot, facebook, pinterest, baidu and few more). This is done because of the blank referrer used in https. The rest get redirected to the parent page (which contains the image).
2. If there is a referrer I serve watermarked(a very large one, other than that all images are watermarked with the company name by default) and scaled image. This is for forums and websites that hotlink, so Bing and Google are excluded from this option.
Possible answers :)
Why allow facebook and pinterest?
Because if there is a need for pinning or sharing an image, then I'd like to be the lucky one who's serving it... I look at it as a free ad.
Why allow google and bing bots?
The web is huge, and if the main content (at least what people mainly come for) of your website is images, then blocking search engines is like shooting yourself in the foot.
After upgrading my server I use less than 20% of the bandwidth limit. So blocking images has nothing to do with the usage. It's just that people tend to click on bad quality images in order to see the image in normal quality they were looking for.
There are only few powers that can change google. Governments, competitors, and webmasters. I don't think the first two have any chance(perhaps besides Fr and De), so don't underestimate the power of unity.