[argyllcms] Re: Perceptual intent

  • From: <robert@xxxxxxxxxxxxxxxxxx>
  • To: <argyllcms@xxxxxxxxxxxxx>
  • Date: Sat, 20 Dec 2014 09:49:50 -0000

Graeme wrote:

> When the encoding space has a very large gamut (i.e. L*a*b*, ProPhoto,
> BruceRGB to some degree, etc.), then you almost certainly don't want your
> image to be rendered to fill that space, since you will be making it look
> terrible, and throws away the actual look of the image.

Well of course not!  The large working space is to give elbow-room for the
few colors that are very saturated and which one does not want to lose (and
which our modern printers can print).  Presumably our intent is to preserve
the image, distorting it as little as possible throughout our workflow
(except for artistic reasons of course).  

Small workspaces like sRGB are too small, AdobeRGB is OK and has the
advantage of being within wide-gamut displays and modern inkjet printers
(more or less), Beta RGB does give some additional elbow room, Prophoto is
much too large IMO.  

Compressing an image into a smaller destination space is unavoidable,
unfortunately, but surely no-one would consider EXPANDING an image to a
larger encoding space!

Ben Wrote:

>There's another danger of using large working spaces.

>If you're fiddling around with the various knobs in an image editing
>application, you may well be tempted to fiddle them in such a way that
>gives the image lots of "pop" on the screen...and, in so doing, expand the
>encoded gamut waaaaay beyond not only what your printer can print but also
>what you're actually seeing on the display. And, unless you explicitly
>check for that, you'll never know...and you'll also never know why you
>always get strange artifacts any time you try to do anything with the image
>other than look at it in the original editing application.

>Yes, I type from experience....

I couldn't agree more Ben.  When I started off with Photoshop I used Deke
McClelland's Photoshop Bible, and, as you may know, he is/was a big advocate
of editing in Lab.  But he wasn't a big advocate of explaining the dangers!
So I quite merrily edited away, curves and levels flying in all sorts of
directions (not realising that I couldn't see what I was doing, literally),
and that what I subsequently printed would be nothing like what my sRGB
display could show.  Needless to say my results were not entirely
satisfactory - but I had to put that down to bad paper, poor printers ...

If you followed some of my posts on Luminous Landscape you will know that I
very much disagree with the use of very large working spaces like ProPhoto
(at least without great care and knowing what one is doing).

But there is a tradeoff ... modern printers do have very wide gamuts and
trying to accommodate them in a 'triangular' workspace pretty much requires
the workspace to be very large.  Beta RGB, for example, cannot fully
accommodate my iPF6400 on papers like the Canson Baryta.  AdobeRGB
definitely cannot.  So what can we do?  Use a different working space for
each image based on its gamut?  Use a larger working space and use it with
caution (and possibly with Argyll's image-specific mapping for perceptual

Probably what we need is a working space with a more complex shape (Graeme
mentioned a cylindrical working space I think) ... but that's not available
now, so what's your recommendation?


Other related posts: