Compared with SIFR, this suffers from some issues. Biggest one perhaps is that you can't select text since it's rendered as an image. Not so with SIFR. I wouldn't go using this method for body text, but if you're already considering doing image replacement for elements like headings, this would be a library to watch. There's bound to be accessibility improvements in the future using methods that put invisible text over the image, or something like that.