Most consumer grade flatbed scanners are designed for Letter and A4 paper. Some are a bit longer for Legal sized paper (8.5″x14″). While there are some larger flatbed scanners available, they tend to be pretty expensive or come with usability compromises and often aren’t really a practical option unless you do a lot of scanning of large images. You can find them occasionally available for rent, but that depends on your local area. And the self-service scanning services at places like FedEx-Kinko’s can add up quickly when large-format scans are priced per square foot. Another option is to photograph the original, but that entails a setup where you line up squarely above the original as well as control the lighting to keep it even.
But a lot of times you don’t really need to look for a completely different option. With a modification to the usual flatbed scanning workflow you can get excellent results.
You can use your regular flatbed scanner to scan images that are larger than your scanner’s window. The trick is to scan it smaller sections and then use software to stitch those sections back together. It’s the same principle as creating a stitched panorama–you’re just creating a stitched mosaic instead. It takes multiple parts of an image and assembles them into a single whole.
Here’s how to do it with a regular flatbed scanner and PTGUI, among the most popular and powerful apps for creating panoramas. I’m using an Epson V600 scanner in this example, but it will work just as well with any flatbed scanner.
Scan the Images
The first step is to create the scans that you’re going to work with.
In this example I’m going to scan the cover of this old Look magazine. The cover is about 10×13 inches, but my Epson V600 scanner’s scan window is only about 9×12 inches. I’m going to scan it in two parts, but you can use the same process to scan larger images in more parts–say, 4 or 6 parts.
If you’re scanning a large image that’s fragile or valuable, where laying it on top of a flatbed scanner might damage it, here’s an idea for hacking a Canon LiDE scanner that might be of interest.
When you make the scans, you want there to be some overlap between the images. Something around 30 percent overlap is a good target, but you can often get by with less. PTGUI will take the overlapping areas to map the corresponding sections between the images and see how they should align.
Every manufacturer’s scanning software is a bit different. You can also use Vuescan, which is compatible with a huge number of scanners.
In general, you want to turn off any automatic image enhancements like color restoration or auto-contrast. The reason is that you want each of the scanned images to use the same scan settings so that they blend more accurately when you stitch them together. If one scan is brighter or more saturated than another, it’s going to look strange when you stitch them together.
You can try using any dust and scratches corrections like Ice, but I find that for this type of scanning they often do more harm than good and tend to leave them off and fix any major issues in Photoshop afterwards.
Once you’ve scanned each part, fire up PTGUI and load the images as you normally would.
PTGUI is designed to work with photos from a camera, and it’s algorithms specifically aim to compensate for lens distortion and perspective issues. The main trick in using PTGUI for a flat mosaic is to fool the software into not applying its sophisticated algorithms. A way to do that is to tell PTGUI to treat the images as though they were taken with a very long telephoto lens.
PTGUI will be confused when it tries to read the lens EXIF data from the scans and won’t be able to automatically find the fields for focal length, etc. So you’ll need to enter something here. You won’t be able to proceed until you enter some values.
In Lens type, uncheck Auto and select Rectilinear.
In Focal length, choose something very long. I use 1000. Basically, you want PTGUI to treat it as though it’s a flat image rather than a curved one.
In Crop Factor, choose 1. The imaging sensor size will then fill automatically–you can ignore them.
Hit the OK button to close the popup.
Choose Align Images. It will then begin processing, and you’ll get the Edit Panorama popup.
In most cases, I find that it does a pretty good job all by itself. This is the result I got. As you can see, there’s really not much I need to do in this case. I can straighten a little if I want, but the alignment and the control points have all worked perfectly in this case.
For this one, I would simply do minor straightening and then hit the Create Panorama tab to go to the settings to generate the final image.
There are many things you can tweak here if you’re inclined, but the ones to start with are setting the output size and filetype. If you hit the Set Optimum Size button, it will automatically update to the ideal maximum size for that image based on the images you’re stitching. You can also make it smaller. I generally prefer to output from PTUI at maximum and resize later. You can also set the file format and where the file will be saved.
Under advanced, try the default settings first. I find that they do a pretty good job.
Once you’re happy with the settings, hit the Create Panorama button.
If you’re not getting the results you want, here are some steps to work through.
If you get something that looks like this, it’s probably because you set the focal length far too small. This example uses a focal length of 10mm (rather than the 1000mm used above). As you can see, PTGUI is doing all sorts of algorithmic gymnastics to try to make sense of the distortion from what it thinks is a wide-angle lens.
Another thing to try is to go to the Optimizer tab and change the Minimize Lens Distortion option to Heavy. Then hit the Run Optimizer button at the bottom. You’ll then have to go back to the Project Assistant tab and his Align Images again.
If neither of those options give you the results you’re after, it could be because the images are not aligning correctly. It’s time to start messing with the control points. Basically, these tell PTGUI which part of one image corresponds on the other image. It’s where the images overlap.
When you click on one image, you’ll get a cursor on the other so you can choose the corresponding section on the other. It’s best to choose small details, like sharp edges, spots, or other details that are easy to match closely.
If you’re starting with good scans with a reasonable amount of overlap between the images, those steps should give you good results. There are some more advanced options for fixing, using masks, etc, but I don’t normally need to resort to those when stitching mosaic images like this.