Can you find the img src?

Well, it does prevent the "average" internet user from doing that.
But that only proves, how low the average is nowadays.

I think the motivation for image stealing is in most cases, that it shouldn't take to long. Of course this kind of processing is no massive obstacle for a thief.

curl -O "http://intermedia.pixelboxx.com/demo/metaimage/_8BE5SrQJKr3_Vv1mfOGaIHd5d/f,p/plastic_kiss_on_a_car.png"

That doesn't take much longer than right click and save and it's internet 101 level stuff.

Using a command-line tool like cURL is definitely something for more experienced web user and neither faster if you don't navigate with a GUI web browser (open the shell, knowledge of the tool, parameters, ...). The "101 level stuff" doesn't work here every time without some modifications, because of "http-accept".

Well, I even resolve to "deliver the png" when a "*/*" is preferred via accept.
So that's why CURL works without further options.
Enhancing this method by browser identification, whitelisting and such would elevate effectiveness to about 8 to 9 out of 10. Which is farely good.
And you can still src the image from another site.