Contact us: +44 116 291 9092
Title Image

Upsizing and sharpening for making a print

  |   Articles and reviews, Article, Image Editing, Photography Ideas, Printing, Topaz, tutorials   |   4 Comments

Upsizing and sharpening an old digital photo for a big print

How image processing software has improved



Site update: Keep up to date with all Keith's work
...Get our Newsletter for new articles/reviews and please subscribe to Keith's YouTube Channel
...Keith's book about how to use tilt/shift lenses is now available.
Our site contains affiliate links - these help support the site. See our Advertising policies for more

Digital cameras have improved immensely since Keith started using the full frame Canon 1Ds in 2004. Its 11MP sensor was considered huge at the time, but what to do if you want to make a print from an old image from a digital camera?

Keith has written up the some of the processes and software used to go from a crop of a 2004 11MP image to an A2 sized print (~16″ x 23″).

The principles involved are applicable to a variety of image editing software packages.

There is a 30min video version of this article
Making big prints from low megapixel images

The seagull

The source photo

In October 2004 I was visiting the Oregon coast, and on a bright, but quite stormy day was walking along the beach at Pacific City, near Cape Kiwanda.

As well as the surf and misty views, I took a few photos of some seagulls flying around me. The lens was the (still) rather good EF70-200mm F2.8L IS. I was still relatively new to using the camera and lens, and having no experience of nature photography was exploring how the camera’s AF functionality worked. It’s not long before this that I’d given up film.

pacific city beach Oregon

I keep every shot I’ve ever taken in an archive, so I can quickly get a good feel of what it was like on the 15th of October 2004.

The seagull – upsizing and sharpening

Here’s the seagull – the photo is open in Adobe Camera RAW (ACR) and shows that I’d still not quite got the hang of exposure.

[Note: many images here can be enlarged if you click on them]

dark-original-file ready for upsizing and sharpening

The histogram shows I could easily have upped the exposure a bit and still not clip highlights.

I’m using Photoshop CS6 here for this example – but am sticking to fairly basic processing with just two bits of external software. I’ve several other articles about such image processing, which I’ve collected together at the foot of this article. This is most definitely not a step by step guide, since it’s about how I processed this particular image. The idea is to show why I made some of my choices and hopefully help guide other people’s choices/exploration when working on their images.

The image gets about a stop in exposure increase in ACR, but other than that, nothing. I’ll work in the large ProPhoto colour space for the image, partly to avoid any clipping and give me plenty of working room (I’m also working in 16 bit).

Since I’m going to be enlarging and sharpening the image, I know from experience that turning off noise reduction and sharpening is likely a good starting point for the image I’m going to work with. This is set to zero and no lens correction is applied.

seagull-original-11MP

Normally I’d tweak more in the RAW processing, but I’m looking to do quite a bit of processing to the file and I want the data ‘relatively’ untouched.

Note, I’ll come back to some of my image editing choices and options later, so if you have disagreements with my choices, feel free to comment (real names get longer answers!) or email me ;-) I do change my mind, but it’s usually based on evidence, not dogma ;-)

The composition I want for my print is a crop of this image, which further challenges my edit workflow. This is what I’ve got to work with – a roughly 3000×2000 file.

original-image-size

So, I can get a 10×7 print from it… How to get to a larger file.

I’ve two bits of software that immediately come to mind in making this a simpler process than I’ve used in the past.

For resizing, I can use Topaz Gigapixel AI and for sharpening I can try Topaz Sharpen AI. If you want to try these packages, there are 30 day free trial versions available.

There are other ways of doing this, but these two are my go-to choices when normal resizing in Photoshop is being pushed or sharpening less than optimal quality images is needed. Both also work in standalone mode, so you could use other software (LightRoom/DxO PhotoLab) to do the RAW processing, export TIFF files, process them and then re-import to do printing (if you must – none of these packages would ever be my choice for printing large prints – YMMV)

The original image, shown here at 100% is quite soft, particularly since I’ve turned off sharpening.

original-sample

Upsizing and sharpening or is it the other way round?

So, do I sharpen and then enlarge, or enlarge and then sharpen – or a bit of both?

Experience with making very large prints in the past suggests that very careful attention to sharpening (only where needed) and then enlarging. However I need to consider aspects of when I used Gigapixel AI in my look at making large prints from low megapixel images last year. Add to this the fact that Sharpen AI has just been updated, so I thought it worth testing various methods…

I decided to produce three versions of the image for enlargement, one would get a light sharpening in Sharpen AI, one a stronger setting and one would be the version straight from the RAW converter.

Here’s a screenshot from when I was applying the stronger ‘shake’ sharpening. The split screen view shows just how much is being done. There is also some noise reduction going on – I’m just using default settings here. You could try the auto settings as well, since these analyse the image to work out settings to try.

original-image-with-sharpening

With these three versions I can get an idea just what method is working best -for this image- and then decide if I want to fine tune settings.

Remember, this is about making a print, and not a scientific analysis of software functions. The print is the final product, not some massively zoomed in view on a monitor.

Next I need to upscale the resulting images. I’m going for a 4x upscaling, even though it will give me a bigger file than I need.

I no longer worry about producing my image files to some ‘magic’ resolution for a particular printer and I don’t use anything more than the printer driver for the printer. Why? Well I’ve been making huge prints for a while and increasingly realised that software is getting better all the time and that when I can’t see differences in prints, then I’m pretty sure no-one else will, and if they can, it’s their fault for getting out a loupe to look at my prints. There is a time to say ‘This will print well, but is the photo any good’

I’m using Gigapixel AI, which lets me batch convert files. Here are the three files showing how the resampling will look.

First – no sharpening applied to the source image.

resize-original-image

Then normal light sharpening.

resize-sharpen-setting

Then the stronger correction that addresses movement blur as well.

resize-sharpen-at-shake-setting

The photo was taken at 1/1000s but at 200mm there could well be a bit of movement blur. Anyway, I know from my testing that some types of len aberration are also amenable to fixing at this setting.

Details of the three processed images [click to enlarge]

three-4x-resized-versions

There was one last option – applying sharpening after resizing the basic image.

sharpening-a-resized-image

You may need to enlarge to see it, but it’s not something I’d want to use.

You might also be thinking that those earlier sharpened images are too sharp? However, remember that we are making a print and have over enlarged to some extent.

This suggests to me that when printed, this level of detail is going to look fairly good. if you’re unsure, then make a test print on a smaller bit of paper and look at it from a reasonable distance. How the print actually looks on paper always trumps what you think looks over processed on the screen.

Too much detail

Before moving to my photoshop layer tweaks and printing, there’s the matter of looking what’s going on in those big flat areas of sky and cloud.

The 1Ds collected dust – it pre-dated built-in sensor cleaning. That means I have to go through the image looking for dust spots to fix. This is a time to be thorough – I don’t want to discover them from the print ;-)

dust-spots-and-banding

Just above the dust spots you may be able to see some fine horizontal banding. This is from the 1Ds sensor and exacerbated by the resizing (I found this before when looking at making big prints).

My personal preference is to go back to the original file and produce a second version with the noise reduction turned up.

reduced-noise-for-background

I’ll resize this in Photoshop using Bicubic Smoother up to the size needed for my other processed image.

image-size

I’m compositing the images and doing a few adjustments using layers (my number one reason for using PS over many other editors).

Here are my layers in the final print image.

photoshop-layers

The base layer is the soft resampled image I’ve just made and upsampled in Photoshop – it has no banding. It does have dust spots so I need to fix these with the healing brush.

Then we have the sharpened/resized layer – this has an image mask painted in to show just the seagull – which is sharp and detailed.

Then there is a curves layer – I’ve used this to brighten some of the underside of the seagull. The brightening is painted in using a low flow soft brush, painting onto the layer mask. Remember, the curve layer brightens the whole image, but the mask restricts this to just where you want it.

curve-to-lighten-bird

This is a view of the mask itself, after I’ve painted in some white to locally apply the curve – a bit like when I used to ‘dodge’ prints back in my darkroom days (only a lot more accurate and reproducible).

curves-mask

Here’s the lightened seagull – no you are not meant to see where I’ve done it…

bird-lightened

I’ve also added a vibrance layer, to deepen and intensify the blue areas.

This is for a print, so what looks intense on the screen may look a little ‘flat’ as a print. I’m printing on Epson TPP (lustre) paper using the P5000, so I’m expecting something broadly similar to my screen. If I was printing onto a matt paper I’d have to be a bit more careful in adjusting overall contrast. The print would look very different on a matt art paper – I think a lustre paper works for this image…

vibrance-boost

The final levels layer lets me expand the tonal range a bit to fit the range available on the paper. Prints have a much smaller tonal range than screens. I’m working in 16 bit here, so adjustments are unlikely to cause any banding or problems.

levels-adjustment

Making the print

Although I’m pretty confident that the printer/paper combination will cope just fine with this image, I do a quick soft proof, using the icc printer profile I made for this paper when I did my original P5000 printer review.

soft-proofing

It looks fine.

Soft proofing is sometimes a good confidence building tool, but I always caution against its over-use, as a crutch to avoid having to make the effort to really get a feel for how your printer works with a particular paper. Knowing your printer, along with sound colour management is the key to consistently getting great looking prints. Forget the idea of ‘prints matching your screen’ – learn to appreciate the print as -the- end result. BTW I’ve more on this in my article about Improving your photography by printing your photos.

Here’s my 2020 version of the image, compared to part of a 2004 version that gave me a quite nice looking A4 print.

2004-against-2020

Some things have come on quite a lot…

All that remained was to print.

I’m using the Photoshop print dialog – the scaling is just because I forgot to set the image PPI at some point in the process.

The paper is Epson Traditional Photo Paper – it’s being printed at 1440 dpi.

print-from-photoshop

Here’s the image – downsized for web use and saved in the sRGB colour space.

seagull_2020

And lastly, the actual result.

seagull-print

More thoughts and ideas

Having known this image since I first started using digital, I’m rather impressed by how much modern software lets me create from the original source image. Of course, it also reminds me why I always keep my original RAW camera files.

Upsizing and sharpening? no, definitely the other way round.

Could I have produced a similar print without the Topaz software? Well, with some real care in sharpening and enlarging, probably something at A3+ or even A2, but without the fine detail.

Topaz Sharpen AI now has built-in masking, so I could have just applied the sharpening to the seagull and hoped that Gigapixel AI might not have picked up the sensor banding. I guess you could do much of this in something like Lightroom, but to myself it would be a poor imitation of what I can do with a years old version of Photoshop. I’d be able to do most of the work with software such as Affinity Photo. The idea was to chart a series of things to do with the image on my journey to a print, not a precise recipe.

I hope this article has been of help – it’s part of my work finding photographic things to try whilst like many others I’m stuck at home in 2020.

Related articles

I took several road trips in the US during the 2000’s and have some travel blogs if you like the scenery (I want to go back!)

Printing is one of the best ways I know of improving all aspects of my photography – there is an index page specifically for Printing related articles. However you might want to start at:

For making big prints

Software – I’ve quite a few Topaz articles/reviews which go into more detail about using it.

The site has over a thousand articles/reviews – see all the categories listed in the dropdown category menu at the top right of any page.

Check the free trial versions of Topaz Software
Use our discount code ‘Northlight’ for an additional discount on some promotions.

Never miss a new article or review - Sign up for our occasional (ad-free) Newsletter
and please do subscribe to Keith's YouTube Channel

Was this helpful? Tips help run the site and are appreciated
Buy Me a Coffee at ko-fi.com
Other areas of our site that may be of interest...

All the latest articles/reviews and photo news items appear on Keith's Photo blog

tilt-shift book

Keith explains tilt and shift lenses

Keith has written a book that looks at the many ways that tilt/shift lenses can benefit your photography from a technical and creative point of view. If not in the UK check for import issues and maybe try an on-line bookshop. Keith has no connection with sales of the book.

ISBN 9781785007712

Book now available

There is also a specific index page on the site with links to all Keith's articles, reviews and videos about using tilt and shift.

We've a whole section of the site devoted to Digital Black and White photography and printing. It covers all of Keith's specialist articles and reviews. Other sections include Colour management and Keith's camera hacks - there are over 1200 articles/reviews here...

Assorted Google ads - sorry but we have no control over external content. One day you might see one that is remotely of interest


 

We're an Amazon.com affiliate, so receive payment if you buy via Amazon US

4 Comments
  • Keith | Nov 2, 2020 at 9:58 am

    True, it’s not a panacea. I don’t have that other software to compare beyond basic Photoshop.
    However, I’ve found that with any resizing it’s paying attention to the source image that often is the key to success. That’s why I mention some other earlier looks at making large prints.

    http://www.northlight-images.co.uk/making-a-giant-print-from-an-11mp-image/
    http://www.northlight-images.co.uk/printing-low-megapixel-images/

    So for example, I might use different processes on different parts of the source image – once you look for a single ‘click to enlarge’ solution (from any software) you may well get sub-optimal results. It’s this flexibility of approach that I’m really pointing to. There’s no reason not to assume that something ‘better’ than what I’ve used here may come along, but even then, decisions about processing the source image will be key.

  • Tom | Nov 1, 2020 at 11:00 pm

    While Gigapixel in some images does miracles, in others it fails miserable. In such cases I would recommend to use state-of-art upsizing algoritms, e.g. Lanczos in RAWTherapee or QImage.

  • Keith | Mar 23, 2020 at 10:27 pm

    There are some more detailed comparisons of Gigapixel AI in the various review articles. Indeed, if you look at the article about making a 3m x 2m print from an 11MP image, it dates from before the two examples of Topaz software. Also, look at my review of Focus Magic – it handles camera shake very well, as well as sharpening.
    At the moment I don’t know of anything that matches GP AI for large amounts of enlargement. Of course, many times a combination of other techniques can work well, but sometimes you just need a heftier computer (mine will no doubt be inadequate before too long).

  • Charles Bragg | Mar 23, 2020 at 9:45 pm

    My computer does not have the oomph needed for Topaz Sharpen AI and Gigapixel AI. I use CS6 bicubic in 10% steps for enlarging these old images. Can you compare the CS6 method to Gigapixel AI?

Post A Comment