A peacock from a pigeon

In short: Casual photos taken by phones are good for perceptual rendering intent (RI). In other cases, use colorimetric RI and edit the colors by your hand.

In long:

In Japan (I live in Japan), convenience stores are everywhere, and they provide MFP as a service. The MFP can print photos from phones. I studied the MFPs, and I found they ignore embedded ICC profiles. They print just the same image from different embedded ICC profiles.

But it isn’t a serious flaw for those MFPs, because they are not concerned about color reproducibility at all. They print gorgeous colors far beyond the source image allowed. They print a peacock from a photo of a pigeon.

But this is life. We are the rest of most. The only two exceptions which even most people care about color reproducibility are:

• When gray is printed greenish or brownish or something.
• When human skin is printed greenish or bluish.

Oh yes, Zygomatic Color helps you facing the cases, of course. But you are the rest of most, right? Yet to make sure, perceptual RI is written in ICC profile, and it is generated by ICC profile builders like Zygomatic Color, OK?

General characteristics of perceptual RI

There is no best practice for gamut mapping algorithms (GMA, generates perceptual RI), so different ICC profile builders generate different results. But they play by the same rule. The better the GMA, the more similar the result, just like marathoners’ physique.

A good GMA inevitably distorts contrast. To explain it as Photoshop’s level adjustment, the midtone slider goes left (becomes a large number). The distortion isn’t very noticeable for unsaturated color, but deep saturated color makes much difference. Look at the violet square near the center of the images below:

Is this a good behaviour? It depends on what you need. In my eyes, the violet square isn’t very good. But look at Macbeth charts. I feel perceptual RI plays better than colorimetric RI for this object. Your eyes may have a different opinion. It also depends on the objects and the eyes, not only the colors.

The comparison image above is Zygomatic Color’s, but all GMAs play by the same rule. Some GMAs may play noticeably better in some cases. But I believe such GMAs will play worse in most practical cases. The better the GMA, the more similar the result.

Conclusion: If you take a photo with great care, you shall have what to indicate by the photo. Use colorimetric RI, edit the colors with great care, and indicate it. If not, perceptual RI is good enough. In this world, a peacock from a pigeon makes most people happy.