Problem with that thinking is that most of the time you aren’t getting architects plans – you’re getting a oil painting of a house by someone who’s never learned architecture. It looks like a house, doesn’t meant it will actually stand upright.
Even if working from a PSD when writing the HTML you should pretend it doesn’t even exist, and if you hit up against limitations of the medium you need to be ready to pitch that pretty painting in the trash. To be brutally frank I consider the entire current approach being taught/advocated in the industry of sleazing out a picture in photoshop, working up a template and then shoe-horning the content into it to be the leading cause of broken layouts… and as a large font/120dpi user and a netbook users, all these fancy fixed width often with fixed height element garbage sites you see vomited up for small businesses and cruddy little personal websites are exactly that – BROKEN, inaccessible rubbish. It’s putting the cart before the horse.
Semantic markup of the content (or a reasonable facsimile of your content) first, bend that markup to your will to make the layout with CSS… then and only then do you start up the goofy paint program to hang graphics on it. The result may not be as artistically pleasing, but its’ a HELL of a lot more useful.
Because at the end of the day, people do NOT visit websites for the goofy graphics you hang on the content, they visit FOR the content… so putting the content as the first priority usually means a more successful site instead of the megabyte-plus sized monstrosities people are vomiting up and calling websites of late. Having worked in art/print before doing websites that was a hard pill to swallow – but my code and practicality of design benefited from taking all the artsy preconceptions and throwing them out the window.
Which is why the majority of successful websites aren’t exactly a graphical tour-de-force. eBay, Amazon, Google, Slashdot, the majority of forum skins… Just enough graphics to be pretty, not so much it gets in the way and certainly not wasting time vomiting up a PSD.
It’s also why I often refer to some things people try to do in HTML/CSS as “But I can do it in photoshop” idiocy. The web is not a static width delivery system, a desktop resolution screen is not/should not be your only target device, the web is not print… With the entire purpose/point of HTML being to deliver content to an endless combination of widths, heights, fonts, and device capabilities – ranging from teletype to print, from braile to screen reader – drawing a pretty picture for just one target (or even multiple targets) before you have semantic markup is just idiotic at best, completely crippling one’s ability to use HTML for what it’s actually FOR.
NOT that the majority of people sleazing out HTML 3.2 and slapping a 4 tranny or 5 lip service on it have any clue what the point/purpose of HTML is, and just keep their heads firmly wedged up 1998’s backside. Concepts like multiple media targets, separation of presentation from content, caching models going Pow-zoom-phwee right over their heads.
Part of my increasing disgust with the industry as a whole, part of why I’m visiting less and less websites as they are increasingly USELESS to me as a user, and part of why I can’t believe ANYONE is DUMB ENOUGH to be drinking the HTML 5 kool aid.
Don’t care how pretty the picture is, if the end result is bloated, slow and inaccessible, what in blazes good is it? Unless of course you’re “polishing a turd” by hanging pretty graphics on something without actual content of value – in which case it’s just a drain on everyone’s resources.