By intelligent perceptual rendering, I meant a rendering which takes the gamut of the colors that are actually in the image into account before arbitrarily desaturating colors when there are no out of gamut colors in the smaller space, as explained by Mike Chaney in the referenced review.
BTW, I can’t agree with everything here (which I glanced at quickly), for example:
Perceptual Intent: Use this method for most of your work especially if you intend to just set it and leave it alone. Perceptual intent will produce prints with accurate hue and while overall saturation levels may be a bit less, you are unlikely no notice this by just examining the print by itself. In addition, this method reduces artifacts like banding in blue skies.
Couple things to keep in mind. One is, there’s no standard way to produce a perceptual rendering intent. Its like a picture look between a Canon and a Nikon or how Fuji feels Velvia should render a scene versus Kodak using Ektachrome. Its really very subjective. One profile maker might do a completely different job than another when ink hits paper.
Second, images are far too complex to lump discussions of best rendering intents. Profiles don’t know squat about images. They see everything as solid color patches which when we view in context, and appear to us as an image. One profile with one rendering intent will treat a black dog on coal identically to a white cat on snow and every other images imaginable the same way. The reason we soft proof and toggle between intents is because the technology we have today simply can’t pick the correct or ideal rendering intent. Its very subjective. This is like most other “auto” anything we have, be it in software or in processing equipment. They can do a reasonable job most of the time, but few would admit they do an ideal job all the time. Its why we make our own prints, either in the darkroom or lightroom, because we have to make subjective decisions about our images.
The best rendering intent to use is the one that produces the appearance you desire. So when you consider that all perceptual rendering intents are different, and how complex images are, and that modern ICC color management is based on something as simple as how solid color patches match, and not on any color appearance models, saying one should use a particular rendering intent all the time sends the wrong message. It depends, YMMV. And to say that a Perceptual rendering intent will in any way produce a more accurate print (without defining accuracy and ignoring what the RelCol intent is supposed to do) is not something I’d agree with. Accurate and preferable often don’t equate! As for banding, I can assure you that if I build a profile using a tiny set of patches, lower number of nodes or just poor processing of the profile, the perceptual mapping will not produce less banding than using a RelCol intent with a superior built profile.
I understand the article is aimed at the entry level. But making hard and fast rules like the above isn’t very useful because such readers take them as gospel. YMMV is much a safer and truer statement.