Three ways to fix your pixellated images: the 300dpi problem

Picture this. You’re preparing to travel to present a conference poster. The deadline to submit your poster for printing is tonight and you only started designing your poster this afternoon. You planned to do it weeks ago but have developed a nasty habit of binge-watching crime dramas late into the night instead of catching up on work. So, you’re now on a coffee-fuelled last-minute bender to design your conference poster.

It’s going well until you paste in that photo you took years ago. You scale it up to size to take centre stage in your poster and… it looks awful. It’s pixellated and blurry. Your perfect picture simply isn’t big enough for your full-sized poster! You go scrambling through your folders of images and figures and there’s nothing else you can use. With no other options, you crawl back towards the couch having decided it’s probably time you quit your job anyway.

Then you remember something you saw during your last binge session: a crime show where they had a blurry photo of the bad guy, taken from security footage. They couldn’t make out his face until some technician started clicking buttons and saying the word ‘enhance’ over and over again. Suddenly, the image was beautifully crisp, showing every detail in the villain’s face (who gasp turned out to be the good guy’s brother’s wife’s dog-walkers’ yoga instructor!).

If there were a program that could fix your blurry image like that, you could make your perfect poster in time and wouldn’t have to quit your job! There are programs that will do this. Well, kind of. Those scenes in the movies are still a fantasy, but there are ways to take a pixellated image and tweak it enough so that it's suitable for printing. Here we’ll go through three options that we’ve tried and tested to hopefully help you solve your poster-making problems.

To learn more about why you’re facing this problem, read on! Or skip ahead to learn three ways of fixing this problem using the programs Affinity, GIMP, or Upscayl.

What’s the problem and how do you fix it?

Most of the photos and images we see these days are on digital screens where the standard resolution is only 72dpi. When it comes to printing, however, the minimum acceptable resolution is 300dpi. So, a picture at maximum size might look great on screen, but terrible when printed out on a conference poster. You’ve also probably come across this problem again when journals and publishers ask you for print-quality images.

The first and most obvious way to avoid this is to make sure you always have suitably high-resolution images. Photos taken up close with digital SLR cameras, or images from high-resolution scanners are ideal. But even these images can’t scale infinitely, and you won’t always have these options available.

These programs below can ‘upscale’ images by making algorithmic extrapolations of the data being fed to them. Keep in mind you don’t get extra ‘real’ details in the high-res images they make. This isn’t the fanciful world of the movies where a detective can zoom in so close on a blurry photo that they can identify the artisanal blend of the coffee stains on the bad guy’s teeth. There are limits to what can be done, and they might only work in certain circumstances. Try out some of the options below and hopefully get your poster design or paper figure over the line and ready for printing.

AFFINITY

Affinity is our go-to program for design and training. It’s powerful and completely free. Once you’ve downloaded the program and signed in using a Canva account, here’s how to fix your image resolution problems.

• Open the image file in Affinity.

• In the top banner, click on Document.

• In this dropdown menu, click on Setup.

• Then click on Resize.

• Then click on Document.

This opens a pop-up window where you specify the final dimensions that you need, and the resolution you require (remember 300dpi for printing). There are a few options for what method the program will use to resample your image (e.g. bilinear, bicubic). Try a few different ones and see what works best.

There is another way to upscale images in Affinity, but it isn’t available in the free version. It’s one of the AI features tacked onto the software hidden behind a paywall. If you’re wondering when, or if you should, use AI for image upscaling, I test out a dedicated AI upscaling app below. But first, one more very impressive free option.

GIMP

While GIMP doesn’t have the suite of features that Affinity does (e.g. vector graphics), it’s perfectly suited to things like photo editing and illustration. There’s no need to register an account for this program, just download the software, open it up, and get started.

• Open the image file in GIMP.

• In the top banner, click Image.

• Then, in the dropdown menu, click Scale Image.

This again opens a pop-up window where you specify the dimensions and resolution you need. Like above, you can choose the resampling method the program uses to upscale the image. What do these methods mean and how do they work? I honestly couldn’t tell you, but try them out and find one you like.

UPSCAYL

The first thing you’ll notice about this program is the sleazy tech name – Upscayl. The era of lazily throwing around misspelled words in the hopes of stumbling across trendy app names is apparently not behind us. The second thing you’ll notice is that this free app doesn't seem to be free. If you head to the Upscayl website the first thing you’ll find is their subscription cloud service, which is bizarrely expensive. You’ll want to look for the desktop version of the app, but don't click the link that takes you to an App Store, it will prompt you to ‘donate’ $20 for this free app. Find the dropdown menu that lets you click on the free version.

Once you’ve opened the program you just drag and drop your image in. Then on the left hand side pick the upscaling method, image size, and output folder. They have a handy model specifically for upscaling illustrations which works great for bold design work.

Like everything else being shoved down our throats at the moment Upscayl is an AI tool. The benefit of using an AI tool for image upscaling is that it creates an entirely new image. This means that it can turn an almost indecipherable cluster of pixels into something crisp and highly detailed. But this is also the biggest problem with using AI image upscaling. Here’s an example using a very low-resolution image.

The result is a much more high detail version of the original. And while it might feel like the fantastical technology we see in crime shows, it’s not. Because it’s been AI generated, you could argue that it’s not a photo anymore. The program has produced a new image based on its training data. Upscayl is making a guess as to what a this crab could look like, if smushed into the layout of the pixels you prompted it with.

This is where we, as researchers, should consider whether AI use is appropriate or ethical. If the image you are showing is purely illustrative or there to look cool, then maybe an AI generated image is fine. But if your image is there to demonstrate a study result, showcase your study subject, or depict research participants, it gets ethically muddy very quickly. Especially if you are implying that you are showing a real photo, then something like the above crab is deceptive. This is why many peer-reviewed journals are explicitly banning any AI-generated or modified images, and the use of AI-generated images could start people questioning the integrity of your work.

Final thoughts

One last thing to note is that, so far, we’ve been working with an image like a photograph: something that uses a pixel-based format, also known as “raster”. But as researchers, we deal with all sorts of images like diagrams, graphs, charts, maps, illustrations, you name it! For other kinds of images, it’s possible that you might not need to worry about dpi and pixels at all. There are other file types that don’t rely on pixels to store image data and can be printed and scaled to any size imaginable. Stay tuned for an upcoming blog post on how this works, or, to learn how to make these images yourself, check out our online course on using Affinity for graphic design.

For now, remember that there is no way for these programs to extract extra information somehow hidden inside individual pixels. They make guesses as to what the new pixel values are based on nearby pixels or training data. While these tools are very handy, they can’t work miracles. The next time you have a conference coming up, or you are preparing a figure for a paper, make sure you have high-resolution images ready for printing. But if you’re stuck, these methods above can work wonders and help get your designs looking sharp in a pinch!

Yours creatively,

Dr James O’Hanlon