A friend of mine asked me to copy three 12×18 inch watercolors she had painted. To make it easy, she gave me a bunch of the same 12×18 inch cold press watercolor paper she’d used for them. I put the 45mm f/2.8 GF lens on a GFX 100, clamped the camera into a C1 on a Fobo camera stand, and pointed it straight down. I got a piece of board and a bucket to support it about 15 inches off the floor. I put one of the watercolors on the board, and lit it with two Aputure LED lights in 24-inch soft boxes. I used the GFX 100’s focus bracketing to capture a sequence of images long enough so that any alignment errors and focus curvature of the 45 GF wouldn’t give me blurry results. I developed the images in Lightroom with a profile that I’d made for the Aputure lights with document preferences. I stacked the images using Helicon Focus 7. I brought them into Photoshop, did some distortion correction and cropping, and set out to make a print.
Using an Epson P800, I printed a test pattern on the watercolor paper and determined the gamut inadequate for the image. How can that be? The watercolors are not real different from inks, right; and the paper is the same. Here’s the problem. The watercolors are pigments/dyes that can reflect many different spectra. Where they aren’t mixed with other colors on the paper, they are analogous to what the printing industry calls spot colors. Spot colors are inks that are loaded separately into the press, and usually are unmixed with other inks on the paper. The inks that are in the printer are called process colors. They are designed to mix with each other to produce colors not obtainable with any one ink alone. But you lose some things with process colors, and one of them is gamut. You can’t cover the range of color that you could get by using spot colors.
Then I discovered I couldn’t print borderless on the watercolor paper with the feed mechanism I had to use for it. Strike two. I crossed the watercolor paper off my list, and opened a box of Epson Cold Press Bright. I made a test print, and I’ve got plenty of gamut with that paper.
Now to the editing. I found that I couldn’t bring the luminance of the paper base up to the white point of the file without clipping delicate highlight detail, even when I applied curves through gradient masks to deal with lighting nonuniformity. So I stopped just short of there, which meant that the pure paper white parts of the original image wouldn’t be pure paper white on the copy. That’s pretty obvious when you can see the border of the copy image on the paper, but it’s not obvious at all if you can’t, so I think these will be fine if when they’re framed the mat covers the areas outside the image.
Okay. The highlights look good enough. The overall luminance of the midtones and dark tones is OK. But some of the colors are wrong, and which colors are wrong depends on which watercolor (there were three) and where you look. I was reduced to spot editing of the colors. I didn’t get them perfect, by any means, but I think I got them close enough. If I had accepted this assignment for pay, I would have been tearing what little remains of my hair out.
Lessons learned:
- I probably should have used strobes instead of the Aputures, which have a bump in the blue part of the spectrum.
- But then I would have had to make a profile with reproduction intent for the strobes. I probably should have bitten that bullet. Even better would have been to profile the camera with a target based on watercolor paints, but that’s a bridge too far for me.
- Even with good lighting, cameras don’t see colors like humans, and through a little experimentation, I determined that they’re pretty far off with some watercolor paints even with sunlight as the illuminant.
- I’ve copied oil paintings before, and they didn’t have bare canvas showing, so I never had to deal with the highlight clipping problem that I ran into with these watercolors, which had swaths of paint-free paper. I was surprised it was so difficult. I wonder if photon noise is partially to blame.
- I have renewed respect for the people who do this for a living.
Ilya Zakharevich says
> “I wonder if photon noise is partially to blame”
Is not it “ink drop” noise instead? With (say) 10 drops/mm² (in a very bright zone), the printer would have no way to put any gradation…
(Although this says nothing about “why watercolors are special”…)
Weston Harby says
In my experience (I do art repro for a living), watercolors are the hardest to reproduce. Getting the artwork flat is difficult, I usually have to hold it down with my hand. The other is color as you said, but its usually not a problem to fix that with other work. A color range selection with the appropriate correction layer makes it easy. Water colors are tricky since there are lots of tones between colors, or even in solid colors. This makes it hard to get a good selection. Water colors are also low contrast, which makes it easy to overdo any corrections.
The other issue is gamut. Most inkjet cotton rag papers have more than enough gamut to reproduce any color in a watercolor. This means than a colorimetric reproduction is possible. This pushes me to try and get it as close as I can.
JimK says
It helps to use watercolor samples as the training set for your profile making software.