Last month I wrote about how I go about retouching digitized images of my paintings. In that article, I admitted that I do not own a camera that is high enough quality to photograph my work and so I am forced to use a flatbed scanner to to pull my work into a digital work space.
Initially, I thought that a discussion of that scanning process would be a waste of everyone’s time, it became clear in the weeks since that there actually was value in that discussion. And so I wrote this.
Now, there are some reading this that might rant and rave against the use of a scanner to digitize artwork, and admittedly it can be less than ideal. Still, in all the years I’ve been doing things this way, I’ve never heard a complaint from any of my clients about the quality of images that I’ve supplied them
With that out of the way, here’s how I go about doing things.
When scanning, there are two surfaces that one need pay attention to. The first is the scanner’s glass plate, the second is the surface of the artwork itself.
The scanner’s glass plate, I find, benefits from regular and thorough cleaning. It’s important not only to get as much dust and lint out of the way, but also any streaks such cleaning might leave behind. Sure, you can always retouch any hairs and random skin flakes sprinkled about the scanner’s surface, but why make more work for yourself? Cleaning is pretty quick and is pretty easily dealt with.
Before we move on, though, I want to point out that while the particulate that may find its way onto the scanner’s glass is easy to deal with in retouching, the streaks are a bit more difficult to take on. So even after I’ve finished cleaning, I always rub the surface down thoroughly with a dry, lint-free cloth or a super cheap brand of paper towel that doesn’t leave a lot of lint behind (seriously, your Bounty or Brawn paper towels are great and absorbent, but they leave SO MUCH paper dust behind—the cheapest paper towels on the market are FAR better for this kind of thing as they’re less likely to undo any previous lint removal you’ve already done).
If you want to be certain you’ve gotten it all, you can always put a black board or piece of paper down onto the scanner, scan it at a reasonably high resolution, and examine the result Photoshop to see all the spots you may have missed.
The art surface is a different story altogether. In this case, before I even begin scanning an image, I make sure that the surface of the painting is even. By this I mean, if the surface is glossy, I do what I can to make sure that it’s consistently glossy. If the painting’s surface is matte, it needs to be consistently matte. The reason this is an issue is because glossy and matte surfaces will appear very different in the scans and I’m trying to capture the image with as little surface awareness as possible.
To achieve this even surface with oils, the day before I scan a piece, I’ll coat the piece with retouching varnish. This tends to liven up the dull areas quite nicely.
If there are dull spots on an acrylic piece I’m working on, I tend to go over the entire piece with matte or gloss medium, depending on which makes more sense with the piece (though generally, I find that the matte medium ends up creating better results).
In the cases of watercolor and gouache, my experience has been that the surface of finished work is already even enough to proceed to the scanner without further action.
Point is, the more even the surface is, the easier it’s going to be to retouch the image later on.
My scanner is an old Epson Expression 10000XL. It’s a large scanner that is capable of scanning things as large as 12”x17”. This means that any work I do that’s 12”x16” or smaller can be scanned in a single pass, which is really nice. Unfortunately, about half the work I do is larger than that and requires multiple scans that then need to be assembled in Photoshop.
Multiple scans would barely be an inconvenience were it not for an unfortunate aspect of this scanner: the scanner’s glass is not flush with the rest of scanner’s surface on three sides.
What this means is that depending on the placement of the piece when doing multiple scans, there’s a lip along the scanner glass that prevents the work from laying flush and so the scans can sometimes be slightly distorted within an inch or two of that lip. How much this needs to be accounted for and corrected down the road depends on the image and the number of scans involved. Frankly, I don’t always do much to account for this distortion because it tends to be very subtle and is something I can largely eliminate as I assemble the various scans.
Side note: In order to eliminate the issue of the inset glass, I’ve contemplated having an additional piece of glass cut to the dimensions of the scanner glass and placing it on top. This would give me enough height to eliminate the problem altogether. However, I’ve never followed through with it, and it’s possible that this solution could instead cause additional problems as there’d now be three glass surfaces to clean, the possibility of additional glare, and the possibility of that extra bit of separation resulting in scans with a soft focus. All that’s conjecture, however. Regardless, if anyone’s ever tried something like this, I’d be very curious to hear about your experience.
Anyway, I tend to scan my work at 100% and minimally at 300dpi—often going as high as 600dpi. This may sound excessive, but I find it’s better to have to downsize an image after the fact rather than wish I had a larger version of it later on. Plus, I find that the larger image makes for more precise retouching, which yields better, sharper results.
When placing the artwork, I do not abut the edges of the scanner plate. I have found that scanners will often cut off a small sliver of the image if fully abutted, and so I always leave a little breathing room. That way I can be sure I’ve gotten the full image, all the way up to the edge.
With multiple scans, an important goal is to scan as much overlap as I can. In essence, I’m looking to have coverage of as much of the middle area of the image in as many of the scans as possible. This helps both in aligning the various scans, as well as disguising the transition of one scan to another down the road.
A Note About Color Software
Many scanners come with color software built into the scanning software. And some operating systems have their own color software. I’m personally not a fan of any of them. Depending on the color software and settings, scanning results can end up being WAY off color-wise, and in some cases subtle details in light or dark areas can be lost. So, personally, I scan with all the color correction software turned off, preferring to do the color correction later using multiple color correction layers in Photoshop.
However, I confess to being old fashioned, and sticking to what’s always worked for me. It’s likely that some experimentation with the scanning and color softwares available to you might yield superior results. As with all techniques and processes, your mileage may vary and it’s important to experiment to find what works best for you.
Some (Automatic) Assembly Required
Once I have the necessary scans, I open them all up in Photoshop and drop them into a new file at the appropriate dimensions (with a little wiggle room that can always be cropped out later). There are two ways one can assemble multiple scans. The first is by using Photoshop’s Photomerge feature, which can be found under File>Automate>Photomerge.
I have personally found Photomerge to be extremely hit and miss in its quality. While it does tend to do a decent job overall, I find that I often end up needing to manipulate some of the layer masks it creates because invariably, there will be an awkward hard edge or two that remain visible. I’ve been told that Photomerge works better if you have a photograph of the full artwork included with the other layers as sort of a base for Photoshop to compare to. I’ve not personally seen a huge difference in doing this, but your results may differ. Also, if I had a good enough quality image of the whole painting, I probably wouldn’t bother scanning the darned thing.
For the most part, if I need four or more scans to capture an image, I utilize Photomerge and then fix any flaws as best I can using small sections of the various scans to cover or disguise any ugly bits.
If fixing the Photomerge results requires too much work, then I will roll up my sleeves and do it all manually.
Some (Manual) Assembly Required
When I assemble the scans myself, I decide which scanned section I want to base the others off of and then turn off all the other layers. My choice is sometimes arbitrary, but more often than not, I work from top left to bottom right. I square this base layer (which we’ll call “Layer 1”), move it into position, make sure it’s level, and then move Layer 1 to the bottom of the layer menu (if it’s not already there).
Then I decide on the next scan (which we’ll call “Layer 2”), turn that layer back on, position the layer above the base layer in the layer menu, position it to what I think is approximately the right position, and reduce it’s opacity to 50-60%. The semi-transparent layer now becomes a lot easier to align with the layer below.
Looking at key features where the scans overlap, I use the move tool and the arrow keys to adjust the placement. Since there are often a lot of specular highlights and potential lint visible at this stage, those “features” can help fine tune the second layer’s placement.
If rotation of layers ends up being necessary, I usually choose to type in the degrees of rotation (which tend to be minute—somewhere in the .02 degree range) at a time. It gives me a lot more control than for fine tuning.
Once I get the first two layers lined up, I then lock them both in place. If there’s are more than two scans, I repeat the process above until all scans are placed and aligned.
Once that’s complete, I create a layer mask for every scan layer that isn’t the base scan.
Using edges and features within the image, I then create new edges for each layer in order to better conceal where one scan ends and the next begins. Generally speaking, I utilize a fairly hard brush while doing this in order to keep the minute features (brush strokes and bristle marks) sharp and clear. The reason I use layer masks rather than an eraser to remove the parts of the image I don’t need on each layer is simply so I can easily push and pull the edges as necessary.
Once I have all the parts lined up and the edges masked, I select all the scan layers and create a group with those layers.
Here’s a look at a couple of other images:
After it’s all placed, aligned, blended, and grouped, I start my retouching process.
Hopefully that helps clarify my process and my reasoning. Probably not the most engaging reading, and certainly not particularly helpful to the digital crowd, but hopefully it helps someone out there.
The next couple of entries from me will likely be reruns. I’m starting a move from one coast to the other and my next couple of months and I’m not sure I’ll be able to keep up with the posts amid the transition. Hopefully I will be able to do it all, but if not I look forward returning with a vengeance and with new art to share. Until then, stay safe everybody, and keep at it.