Printing giant-sized posters and billboards has long posed the problem of how to get very large images to reproduce without degradation or pixellation.
Even the latest, highest resolution digital camera can’t capture enough pixels to give truly photo-quality results on something the size of a truck or even a ship. So, how do you get artwork, especially something that needs to look like a photograph, up to huge size?
Over the years a lot of techniques have emerged. However, the first thing to consider is whether you actually need to do anything. The need for resolution is closely tied in with viewing distance and the halftone screen in use. If your big print is going on a billboard designed to be viewed from 50 metres away, and printed at 80 dpi, you may not need dramatic enhancement of the pixel count.
We talked to Austin Coates, studio manager of UK large format specialist PressOn, who said “most of what we print is for large viewing distances. We always try to print at 100 dpi, and ask our clients for 75 dpi at the lowest.” That is the final printed size: if you’re starting with a normal photograph that has to be enlarged, it will still need a lot of pixels to start with.
The techniques for increasing (or decreasing) pixel count are generally called interpolation or resampling. Any worthwhile photo editing program will have built-in interpolation, which you’ll normally find in the image size menu.
Naturally Photoshop had this from the start. Its options have changed over the years and the current CC 2015 version gives you a choice of several automatic and either enlarging or reducing methods.
For enlargements, which is what we’re mainly interested in here, there’s a choice of Preserve Details, or Bicubic Smoother. Preserve Details tries to keep edges sharp around fine detail, but this can also enlarge image noise, which might look ugly. So Photoshop gives you a slider that lets you progressively reduce noise, though this can also blur real detail. Bicubic uses a different method to avoid graininess, but this can flatten the detail a bit.
If you enlarge by 300% to 500% there’s a visible difference between the three options. Which you choose depends on the image, but I’ve found the Preserve Details option with 10 or 10% noise reduction best for the sort of technical images I often photograph. Faces on cosmetics images might work better with Bicubic Smoother.
There are third party pixel resamplers, of which probably the most well-known is On1 Software’s Perfect Resize. Until 2012 this was called Genuine Fractals, which was originally launched in the 1990s by Altamira Group. The current price is US $39.98 (about €36.50) and it runs as a plug-in to Photoshop though it launches into its own standalone window. Perfect Resize provides a lot more controls over the settings than Photoshop, and it can import and save presets.
Alien Skin Software uses a different technique with its Blow Up software, priced at $99, about €90). This analyses the image and maps it into vector shapes that are then enlarged. This works particularly well for scanned logos and the like with distinct shapes.
Both Perfect Resize and Blow Up have optional controls to sharpen as they are resizing and also to restore a little grain, which can look more natural. Both can also crop and scale to pre-set aspect ratios and sizes, which may save time and mistakes if you’re doing a lot of these.
With our test there were some small differences in appearance between Photoshop, Perfect Resize and Blow Up. When examined closely on-screen Blow Up produced marginally the best results, but it’s unlikely that you’d notice much difference in print.
It’s hard to show enlargements in web pages, but this series of pics show the effects of 300% enlargements of a small section of the original image in Photoshop CC2015, Alien Skin BlowUp 3 and OnOne Perfect Resize 9. BlowUp is marginally the best, but Photoshop is still very good and “free” one you’ve signed up for Creative Cloud (with some restrictions).
Here the original photograph has been converted to a full tone vector image using Adobe Illustrator’s Live Trace. When reproduced at a similar scale to the pixel enlargements above, it’s hard to tell it is no longer a contone photograph.
Austin Coates at PressOn concurs: “We tried Genuine Fractals and the like some years ago and to be honest I couldn’t see any difference to what Photoshop does for free. So we just use Photoshop and various filters for everything now.”
With any of these techniques you may still get jagged edges on curves and diagonals. A bit of judicious blurring can help here, either using the Blur filter or slider offered in the enlargement controls, or by going into Photoshop’s Blue menu. You can use soft-edged masks or Photoshop’s Blur paintbrush to confine the blurring to where it’s needed.
What to do about PDFs?
Print service providers may not receive original graphics files that easily can be edited in Photoshop or Illustrator. A lot of them will be supplied as PDF files, which still contain raster and vector elements, but they’re a bit harder to get at.
Fortunately you can pull PDFs apart and edit the various elements. Adobe Illustrator can do this up to a point, but a dedicated PDF editor gives more features. Enfocus PitStop is widely used and reasonably priced, and includes a wide range of editing tools. Callas’ family of pdfToolbox editing tools have similar functions. At the high end OneVision’s pre-press solutions workflows include Asura for manual PDF editing and Amendo for automated enhancements.
You normally only need to worry about image enlargement if you have photographs or scanned images in the job. These are called bitmap or raster graphics and are made up of pixels that form image cells.
Vector graphics, which are made of mathematically described outline shapes and fills, can be enlarged or reduced to any size with no loss of quality. Logos are frequently created as vector graphics, though they may reach you as pixels if somebody has scanned them off a product.
Adobe Illustrator and Corel Draw are the best known vector graphics design programs. However, layout programs such as Adobe InDesign or QuarkXPress also use vectors in the shapes they draw.
All modern computer-generated type is also vector based, so works at any enlargement factor (very small and very large characters are generally automatically modified for the best optical appearance using “hints” in their code).
Indeed, even the type in Photoshop starts out as standard vectors. It’s worth remembering this if you are incorporating type in a Photoshop composite image that’s destined for enlargement. It’s best to retain the type in its original vector layer until after the pixel image elements have been enlarged, and only then to “flatten” the whole file, which converts type and shapes pixels at the same resolution as the other layers.
It’s also possible to convert a raster image to vectors, so the image becomes thousands of shapes based on similar tones and image contours. You might imagine that this means you can then enlarge without limits. In some cases this will be true, but in others the contour edges between tones will become more obvious. However, if you want to take a photograph and enlarge it to something the size of a 20 storey building, it may be worth experimenting with.
A few years ago Adobe introduced Live Trace, an advanced converter that is built into Illustrator. This will convert full tonal range photographs to vectors, where most third party standalone trace programs are really intended to convert scanned logos to vectorised linework. Illustrator Live Trace has a wide range of controls and presets to get the best out of full-range colours and tones, as well as things like creating logos or converting to black and white. Corel’s Windows-only CorelDRAW X7 and the CorelDRAW Graphics Suite X7 include CorelTRACE X7, a similar raster-vector converter.
There are some more specialised trace programs that are intended largely for logos, including some cloud services. For instance Vectormagic offers an online commercial tracing system where you simply upload your images. This is automatic, though a hand tracing service is also offered for items that defeat the software.
There is actually a way to get truly enormous high resolution bitmap photographs, though it only works for certain types of images. The technique is called gigapixel photography, from the idea that the images can contain a thousand million pixels, in the same way that a Gigabyte hard drive holds a thousand million bytes.
True gigapixel images may be overkill even for billboards. They tend to be mostly landscapes and cityscapes, taken by enthusiasts and specialists and often exhibited on Gigapan. This has a viewer that lets you zoom in and into these pictures and see the tiniest of details: a bird sitting on the top of a distant tower, for instance. It even lets you order large format prints from the website.
The creation technique is to zoom in the camera for a small angle of view with maximum detail, and then to take lots of overlapping images, moving the camera by a mall amount in between each frame. These can then be electronically stitched together and blended using either specialist stitching software, or in some cases the automatic image blending menu that’s been built into Photoshop for years.
My landscape photograph here, of the beautiful Alderley Edge in Cheshire, is a relatively modest 24,000 pixels wide, but even that would still be sufficient for a 6 metre print at 100 lpi. It was stitched from nine overlapping 6,000 x 4,000 pixel photographs to produce a final 24,323 x 5,629 pixels taking up 396 MB.
In conclusion, there’s a wide choice of routes appropriate image quality for very large formats, at surprisingly modest costs. A lot of it comes down to experience rather than software.