How are you going to do that, Stephen? An ICC color transform on an image is kind of all or nothing, isn't it?
Hi Rich, you are correct - however Photoshop gives the end user many flexible options. I was thinking of making an action to pin to this topic (and probably still will), however it is easy enough to explain the broad steps in text form...
1. Create a new layer, move it to the top of the layer stack.
2. Stamp all visible data as a merged copy into this new layer.
3. Copy layer data to clipboard, create a new temp document.
4. Paste data in to new temp doc, convert to the profile that reduces total ink.
5. Copy data to clipboard, close down the temp doc without saving.
6. Paste the converted data into the new top layer, preserving colour numbers/values (perhaps purge clipboard data).
7. Use layer option blend-if sliders and or a layer mask to limit the contribution of this layer to only the darkest areas of the image (probably also including some three quarter tone transitions for safety).
Of course, if any of the underlying layers change, you would have to run the action again.
There is no real need to convert the whole document's colour builds to a different CMYK recipe, when all that is required is the reduction of total ink (throwing the baby out with the bath water and all that). I have compared making masked edits to reduce total ink, however the results have never been as good as when using a profile conversion.
But, the assigned profile is not representative of the colorspace of the file.
Sure it is still representative, it is still the same colorimetric description, just a different profile created from the same characterisation data. As long as the two different profiles are created from the same measurement data, for all intents and purposes the colour is exactly the same. It does not matter that the ink reduced image was converted to the "ISO Coated VIGC 260" profile, and that the assigned profile was "Fogra 39 320", the L*a*b* values of the pixels all remain the same (sometimes for some colours, there may be an acceptable +/- 1 Lab value shift). A conversion out of CMYK will still provide exactly the same conversion, it does not matter than one profile is ISO Coated 320% and the other profile is ISO Coated 280%, or Fogra 39 300% etc.
Why embed a 30mb VIGC profile into a 2mb image, when one can embed a 2mb profile that describes exactly the same colorimetric intent? Depending on the workflow, one may not even wish to tag a profile to the image at all.
Stephen Marsh