Pages: [1]   Go Down

Author Topic: Scaning and printing B&W graphics artwork - how to linearise the workflow?  (Read 1411 times)

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254

Hi all,

I have fought with the task for the last several months, and finally decided to ask on this forum.

I need to reproduce graphics made by charcoal and black ink. They are in A3 format. I scan it wit my Epson V500 in 1200 dpi resolution (2xA4 and merge) to 280 MPix 16bit greyscale scan and then print it with my inkjet printers in 1:1 scale (to be enlarged in the future). There is a lot of playing with levels and gamma during scan and then processing in PS. The results sometimes are more then satisfactory, sometimes are really poor, especially when a lot of mid-grey shades appear in the scan. Finally, the result is printed on Epson Hot/Cold Press papers in ABW (QTR will be used in the future). All parts done in GreyGamma 2.2.

I would like to have a print which would be very similar to the original drawing. After try and error phase now I would like to do it in more objective way. I have V500 scanner used with Epson software, Kodak greyscale 19 patch step-wedge reference, Epson 3880 and P800 printers, x-rite spectro (CM) and quite a lot of Epson cotton Press papers. I tried to put some corrections after neutral scan (0-255 levels, no gamma correction in scanner software) in PS curves after measuring the original reference and the 1st print but result is not satisfactory.

The original artwork L range starts at L*=20 up to the white point of the paper so it should be reproduced with no problem on both printers. Linearity is a problem. I would appreciate any ideas that could help me in this linearisation scan-to-print problem. Many thanks in advance!
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com

The quality of the scanner profile - and of the scanner itself - are critical to the linearity of gray-scale tonal rendition, if by linearity you mean that the L* values of scanned output will be the same or nearly the same as those of the media being scanned. It is difficult to correct major non-linearity in post-scan editing. Custom-profiling the scanner would be a good place to start.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com

I would skip the ABW part. It is not linear by default. I would use QTR as my starting point.

Are the tones important? Getting the black tones to match could be problematic.
« Last Edit: June 30, 2018, 07:16:02 pm by Jeff-Grant »
Logged
Cheers,
 Jeff  www.jeff-grant.com

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254

Mark, Jeff, thank you for your replies.
One approach would be to everything linear independently, ie scanner and printer pass, but I doubt I can handle this (mostly with scanner since I am advanced with printing). The other would be intermediate corrections to linearise the overall flow while I know the input and know the output and this is the part I am struggling with. In such a case ABW wouldn't be harmful.

As for the tones, the original drawing fits within L* range reproducible by the printer on my matt paper, so in theory it should be possible to reproduce given amount of black. That's the theory, my practice is still behind it...

I don't know how to make scanner profile. The new i1studio enables simple scanner profiling (tah's what I can afford) but to be hones't documentation is not clear how to use the profile...
Logged

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com

I would think that the paper and ink are an issue. You may get the linearity right but the paper will add colour, and the inks add their own colour. I'm wondering whether you would be better treating it as colour. It's not something that I have ever had to do so it's a bit of a shot in the dark.
Logged
Cheers,
 Jeff  www.jeff-grant.com

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005

I would think that the paper and ink are an issue. You may get the linearity right but the paper will add colour, and the inks add their own colour. I'm wondering whether you would be better treating it as colour. It's not something that I have ever had to do so it's a bit of a shot in the dark.

Yes, stay in color up to the print proofs. Easier to retouch faults in the original like smudges, silver degrading in B&W prints etc. . On the other hand if the original is quite even in color both in paper, greys and blacks then a switch to B&W can reduce artefacts like CA in details at the edges. It is sometimes easier to add color again for the paper etc after that B&W switch than trying to get color correct from the original color takes. Color in B&W originals is subtle and by that close enough to the delta E drifts of the total process. The eye is very sensitive for color differences between the B&W original and the B&W reproduction. Be careful with the viewing light and paper OBA content too (both in original and print).

That said pixel shift takes already produce better color evenness of drawings and gravures (observation of reviews, no experience with). Flatfield compensation (Lightroom plug-in or Corner Fix, there are more) reduces unevenness in lighting, both color and luminosity. Good macro lenses already deal with CA issues. Scanner's lower repeatability however makes it difficult to use flatfield compensation. There are issues with illumination in scanners too, either increasing paper texture or too flat. RAW development is more easily done from camera output.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
March 2017 update, 750+ inkjet media white spectral plots

Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com

Mark, Jeff, thank you for your replies.
One approach would be to everything linear independently, ie scanner and printer pass, but I doubt I can handle this (mostly with scanner since I am advanced with printing). The other would be intermediate corrections to linearise the overall flow while I know the input and know the output and this is the part I am struggling with. In such a case ABW wouldn't be harmful.

As for the tones, the original drawing fits within L* range reproducible by the printer on my matt paper, so in theory it should be possible to reproduce given amount of black. That's the theory, my practice is still behind it...

I don't know how to make scanner profile. The new i1studio enables simple scanner profiling (tah's what I can afford) but to be hones't documentation is not clear how to use the profile...

If you want linear, ABW is not a correct approach because as Jeff said it is non-linear intentionally and you'd have a difficult time trying to correct that.

I don't have a clue what you mean by "intermediate corrections to linearise the overall flow" - this phrase makes no obvious sense.

The colour values of the media you are scanning need to fit within the colour space of BOTH the scanner space and the printer space.

If you don't know how to make a scanner profile, please consult resources on the internet or any books on colour management, where it is explained. If you can follow what instructions i1Studio provides for making the scanner profile, do so, then using it is a matter of loading that profile into the colour management settings for your scanning software. If the software you are using to scan with can't accommodate that, you need a better application that does.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254

Just some clarifications. The drawings are made on paper which white point is very similar to non-OBA matt cotton or alpha cellulose, so no problem here. Since it is drawn with charcoal and black ink it is just greyscale with virtually no chroma component. That's why I scan it as monochromatic 16bit B&W picture. All (!) I need, is a curve which would linearise the overall flow.

"intermediate corrections to linearise the overall flow" - if you threat the overall flow from signal theory point of view we have one transformation function (non-linear) describing scanner job and the other, also non-linear, describing printer ABW work. Both transformations put on each other go also to non-linear final result.
What I would like to do, is to add the third transformation (in PS) between the scanner and the printer which would make appropriate corrections to make the OVERALL process linear, end-to-end. In such a case, I would not need to worry about linearising the scanner and ABW printer.

It looks like, after my today's trials, that I need to design appropriate look-up table. more to come...
Logged

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com

I'm still confused but that's not unusual. if the artwork is truly L only with no A or B, which I doubt, then the B&W conversion, would be fine. However, at the end of the line you need an ink to match the L100. AFAIK, all K inks have some A and B. I would think that, were this my challenge, I would be working on finding an ink to match the artwork first. If you can't do that, the rest is just wasted time.
Logged
Cheers,
 Jeff  www.jeff-grant.com

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com

I'm still confused but that's not unusual. if the artwork is truly L only with no A or B, which I doubt, then the B&W conversion, would be fine. However, at the end of the line you need an ink to match the L100. AFAIK, all K inks have some A and B. I would think that, were this my challenge, I would be working on finding an ink to match the artwork first. If you can't do that, the rest is just wasted time.

As far as I understand it, the O/P's problem is about linearity of tonal transitions, which is L* - the smoothness and accuracy of tonal transitions in the scan relative to the media. Hue is another issue. A good scanner profile should render both the tonality and the hue of the original. If the profile is no good, it puts an extra burden on post-scan editing to adjust the hue to the right values compared with the media. This can be tedious, especially if there is hue inconstancy through the tonal range. So the O/P's primary interest, as I said from the start, should be focused on the quality of the scanner and the profile. A V500 with a canned profile using EpsonScan may not cut it for this kind of exacting work, which can be pretty complicated to colour-manage properly. But at least the profiling should be under his control if not the scanner for the time being.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Jeff-Grant

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 568
    • http://www.jeff-grant.com

No argument there, Mark, but I think that there are more problems lurking beyond the scan. I'll happily bow out. I just wish that we got feedback as to the result and process followed. That would be very interesting to see. I have to reproduce artwork for an artist pretty regularly, and it's always a challenge so seeing how someone else attacks it is extremely interesting. I haven't tried the scanner approach preferring to capture with a D850.
Logged
Cheers,
 Jeff  www.jeff-grant.com

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com

No argument there, Mark, but I think that there are more problems lurking beyond the scan. I'll happily bow out. I just wish that we got feedback as to the result and process followed. That would be very interesting to see. I have to reproduce artwork for an artist pretty regularly, and it's always a challenge so seeing how someone else attacks it is extremely interesting. I haven't tried the scanner approach preferring to capture with a D850.

Properly profiled, your D850 will provide more faithful outcomes. Scanners are a fickle species.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."
Pages: [1]   Go Up