Pages: [1]   Go Down

Author Topic: Compress RGB 1-255 when Printing?  (Read 2689 times)

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Compress RGB 1-255 when Printing?
« on: December 02, 2017, 07:28:51 pm »

Continually trying to improve my printing, I'm reading two different books on the subject "Fine Art Printing for Photographers" by Uwe Steinmueller and Juergen Gulbins and "The Digital Print" by Jeff Schewe.  They appear to diverge on one step that I find curious.  Steinmueller and Gulbins advocate determining the native black point and white point of particular paper/printer combinations, then using Levels to compress the white and black point of image files that exceed that range prior to soft proofing and printing.  Schewe doesn't appear to advocate that step (or I may have missed it) and instead appears to rely on black point compensation, paper color profiles and rendering intents to make acceptable adjustments.  This raises a couple questions.

1. Does the automated black point adjustment work the same way as compressing the blacks using levels? 

2. What about the white point?  Do both Relative and Perceptual rendering intents compress that assuming I have a good quality paper profile?

3. I suppose there might be some advantage to even further adjustments.  For example, looking at my color profiles with ColorSync, I can see that the blacks actually terminate several points darker in the blue space.  If I were to adjust the RGB of my files to 6-6-6, for example, that doesn't appear like it would get me the darkest tone I can reach appears to be something more like R4 G4 B8.  That presumably could also have follow on affects achieving more accurate colors using whatever math Perceptual uses to render its print files.  Or maybe I'm over thinking that for tone - maybe it could just result in a slight unwanted color shifts and I should stay with the pure black tone.  Dunno and am just starting to test this out but interested if there were any thoughts out there on this.

Probably differences without much practical affect if any, but as I'm still learning, adding 15 or so of these tweaks into a printing workflow can have a practical affect.
« Last Edit: December 02, 2017, 08:43:29 pm by Brad P »
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Compress RGB 1-255 when Printing?
« Reply #1 on: December 03, 2017, 12:02:39 am »

Continually trying to improve my printing, I'm reading two different books on the subject "Fine Art Printing for Photographers" by Uwe Steinmueller and Juergen Gulbins and "The Digital Print" by Jeff Schewe.  They appear to diverge on one step that I find curious.  Steinmueller and Gulbins advocate determining the native black point and white point of particular paper/printer combinations, then using Levels to compress the white and black point of image files that exceed that range prior to soft proofing and printing.  Schewe doesn't appear to advocate that step (or I may have missed it) and instead appears to rely on black point compensation, paper color profiles and rendering intents to make acceptable adjustments.  This raises a couple questions.

1. Does the automated black point adjustment work the same way as compressing the blacks using levels?
BPC, which is always enabled in Lightroom and an option in Photoshop scales images rgb=(0,0,0) to the black point but doesn't compress. It's a linear scale to the black point. The term "compression" normally implies some sort of non-linear compaction.

Quote
 
2. What about the white point?  Do both Relative and Perceptual rendering intents compress that assuming I have a good quality paper profile?
Relative, Perceptual, and Saturation all scale to the media's white point. Always. So the advice to drop the white point RGB from (255,255,255) is simply wrong and will reduce the brightness of an image unnecessarily. Schewe is correct.

Quote
3. I suppose there might be some advantage to even further adjustments.  For example, looking at my color profiles with ColorSync, I can see that the blacks actually terminate several points darker in the blue space.  If I were to adjust the RGB of my files to 6-6-6, for example, that doesn't appear like it would get me the darkest tone I can reach appears to be something more like R4 G4 B8.  That presumably could also have follow on affects achieving more accurate colors using whatever math Perceptual uses to render its print files.  Or maybe I'm over thinking that for tone - maybe it could just result in a slight unwanted color shifts and I should stay with the pure black tone.  Dunno and am just starting to test this out but interested if there were any thoughts out there on this.

Probably differences without much practical affect if any, but as I'm still learning, adding 15 or so of these tweaks into a printing workflow can have a practical affect.
Logged

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Re: Compress RGB 1-255 when Printing?
« Reply #2 on: December 03, 2017, 01:22:05 am »

Thanks Doug.  It’s a little of a black box without help like this. 
Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Compress RGB 1-255 when Printing?
« Reply #3 on: December 03, 2017, 04:32:21 am »

BPC, which is always enabled in Lightroom and an option in Photoshop scales images rgb=(0,0,0) to the black point but doesn't compress. It's a linear scale to the black point. The term "compression" normally implies some sort of non-linear compaction.

hmm, would you consider this option that BPC is non-linear, it compresses dark tones, but not clips them...
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Compress RGB 1-255 when Printing?
« Reply #4 on: December 03, 2017, 09:40:59 am »

hmm, would you consider this option that BPC is non-linear, it compresses dark tones, but not clips them...

My analyses of what BPC does conforms with Doug's explanation.

As well, the O/P should be aware that RelCol and Perpetual are alternative approaches for handling out of gamut colours and the relationships between them and in-gamut colours, as explained in Adobe's context menus in the Print command dialog. Which to use is a matter of personal taste depending on the photo. By editing under softproof and trying the options, you get to see which rendition you like best.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Compress RGB 1-255 when Printing?
« Reply #5 on: December 03, 2017, 11:47:11 am »

Also important is the Steinmuller book is quite dated (published in 2007) and Jeff Schewe's book is probably the best one out there in terms of how to get the best out of the new printers that are available.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Compress RGB 1-255 when Printing?
« Reply #6 on: December 03, 2017, 12:30:40 pm »

Also important is the Steinmuller book is quite dated (published in 2007) and Jeff Schewe's book is probably the best one out there in terms of how to get the best out of the new printers that are available.
Date or otherwise, I don't agree with their recommendations! Soft proof, pick a desired rendering intent, maybe produce some output specific edits if necessary, use BPC.
As for a white point, I see nothing wrong with a print that clips to paper white IF the print maker/image maker desires this. It's something I was trained to do in the conventional darkroom and see no reason why a digital print is any different. Often a paper white specular highlight (say a chrome bumper) where we see actual paper white, make the print look snapper and not as dull appearing.
First rule: there are no rules. :o
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Re: Compress RGB 1-255 when Printing?
« Reply #7 on: December 03, 2017, 12:59:05 pm »

The Steimuller book is now in its 3rd edition, dated 2013, but still the behind the scenes engineering changes so rapidly ....  Yes, Schewe’s book is more up to date and here is another instance where in case of conflict, Schewe’s seems more reliable. 

Andrew, good point on specular highlights.  Reading what Doug (with Adobe I think) wrote, it looks like the rendering intents may adjust for the white point.  I’m going to put that to the test all this today with some printer black and white point ramps.

Thanks for all the comments.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Compress RGB 1-255 when Printing?
« Reply #8 on: December 03, 2017, 01:02:48 pm »

Andrew, good point on specular highlights.  Reading what Doug (with Adobe I think) wrote, it looks like the rendering intents may adjust for the white point.  I’m going to put that to the test all this today with some printer black and white point ramps.
If the original is indeed 'clipped' to white (255/255/255), I'd hope outside of cross rendering with an Absolute Colorimetric intent, the output would be clipped as well.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Re: Compress RGB 1-255 when Printing?
« Reply #9 on: December 03, 2017, 01:35:42 pm »

Yeah. Only having a spectrophotometer built into my Z3200, I’m only going to be able to evaluate it by eye and with a broad spectrum flashlight, so room will be left for further analysis.  But what I hope to see is that my image white point (255,255,255) will appear to scale to paper white and will not overlook specular highlights which I consciously process my images to retain (or disregard).   The ramp I’m using will go in (1,1,1) increments. 
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Compress RGB 1-255 when Printing?
« Reply #10 on: December 03, 2017, 03:15:27 pm »

Yeah. Only having a spectrophotometer built into my Z3200, I’m only going to be able to evaluate it by eye and with a broad spectrum flashlight, so room will be left for further analysis.  But what I hope to see is that my image white point (255,255,255) will appear to scale to paper white and will not overlook specular highlights which I consciously process my images to retain (or disregard).   The ramp I’m using will go in (1,1,1) increments.

You probably won't see differences between printing RGB(254,254,254) and RGB(255,255,255) (typically less than .4 dE76 or .2 dE00) but if you get a strong magnifying glass you will see scattered light gray dots.  Have fun.


Oh, and check out this utility from Graeme. It lets you see what a profile generates as device RGB values from input Lab and vice versa for all 4 rendering intents. It's interactive.
http://www.argyllcms.com/doc/icclu.html
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Compress RGB 1-255 when Printing?
« Reply #11 on: December 03, 2017, 03:49:52 pm »

If the original is indeed 'clipped' to white (255/255/255), I'd hope outside of cross rendering with an Absolute Colorimetric intent, the output would be clipped as well.
I recall reading one of the many internet articles on printer profiles which had a graphic showing Rel. Col. clipping source RGB whites below 255 because of the lower paper white point. It showed all values above RGB 240 or so being "printed" as paper white. Only Abs. Col. does that, not Rel. Col. or Perc which always scales L=100 to paper white.  That error is consistent with the one in Steinmueller and Juergen Gulbin's book. It was nicely written too, just wrong. I wonder how widespread this misinformation is.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Compress RGB 1-255 when Printing?
« Reply #12 on: December 03, 2017, 04:20:02 pm »

I recall reading one of the many internet articles on printer profiles which had a graphic showing Rel. Col. clipping source RGB whites below 255 because of the lower paper white point. It showed all values above RGB 240 or so being "printed" as paper white. Only Abs. Col. does that, not Rel. Col. or Perc which always scales L=100 to paper white.  That error is consistent with the one in Steinmueller and Juergen Gulbin's book. It was nicely written too, just wrong. I wonder how widespread this misinformation is.
I can easily dig up ICC profiles from MonacoPROFILER that had a bug and produced a 'scum dot' in one or two color channels when none should be there.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Compress RGB 1-255 when Printing?
« Reply #13 on: December 03, 2017, 04:30:22 pm »

I can easily dig up ICC profiles from MonacoPROFILER that had a bug and produced a 'scum dot' in one or two color channels when none should be there.

Yeah. There are some bad profiles out there including some I've made with PM5. Ones that don't comply with the last revision of V2 or later. The worst ones in terms of ICC compliance are those that implement BPC in Relative Colorimetric. (Epson canned USA 9800). They totally hose in gamut Abs. Col. which some of us use for repro work. IIRC, they are also ones that produce really crappy results with one of your test images because of the way they map OOG colors.
Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Compress RGB 1-255 when Printing?
« Reply #14 on: December 05, 2017, 01:56:46 pm »

My analyses of what BPC does conforms with Doug's explanation.

I have just realised that we were both right depending if we considel linear or logarytmic scale... :-)
Logged

Brad P

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 258
Re: Compress RGB 1-255 when Printing?
« Reply #15 on: December 07, 2017, 01:45:00 am »

You probably won't see differences between printing RGB(254,254,254) and RGB(255,255,255) (typically less than .4 dE76 or .2 dE00) but if you get a strong magnifying glass you will see scattered light gray dots.  Have fun.

Yep, you’re absolutely right.  I just spent half of the last two days recalibrating, double checking and tightening up my entire color management workflow and printing test prints.  In detail as you know.

Everyone printing from LR or PS at least would do better to not follow the clipping advice I read and posted (with a question mark), but maybe consider some very subtle white and black point curves to mellow out the hard endpoints and bring in some detail that you can actually see.  Seemingly the curves should be a bit deeper at the lighter end.  But that last bit is an unscientific observation. 

BTW Doug, it wasn’t my flashlight or a 20x Zeiss loupe that allowed me to see the near +/- 1,1,1 end points.  Rather dark nightfall and moderately dim daylight balanced LED lighting by naked eye.

Now I need more paper. 
« Last Edit: December 07, 2017, 02:04:53 am by Brad P »
Logged
Pages: [1]   Go Up