What’s cool about playing with CSS3 is you can radically muck around with appearance without resorting to graphic editors tweaks, which has always been painfully slow. And partly because of the editors themselves, which make it painful to keep editing-and-saving. ie. every time you save a jpeg, most editors will keep coming up with the same dialog to set quality level, progressiveness, etc.; and then they’ll get you to confirm you want to override the previous file. Oh…and they separate “exporting” to jpeg vs saving in their native format, instead of just doing both.
But mostly it’s because all the mouse-driven tweaking on a GUI graphical editor is just plain slow. And dealing with the file system - saving to the right place and linking to it from the HTML/stylesheet. Lots of overhead.
Some things can be automated by server-side image generation scripts. Tweaking the URL tweaks the gradient, or rounded corners, or angle, whatever. But flexibility lost, not to mention a massive performance and bandwidth drain.
CSS3 does away with all that. You can make all kinds of radical changes in one or two lines. Tweaking from the comfort of your favourite text editor (==vim). Some would say it’s “too easy”. (They are the same people who think spreadsheets shouldn’t include macros because non-programmers might be able to do useful things with them.) Not every project can justify a dedicated visual designer. Most can’t. So now that we have all this power, there’s a need for simple guidance on how to design visually with all these elements. To create passable designs, not masterpieces, which is good enough for most sites. Let’s see that kind of advice from the visual design world!