For the last year I’ve been shooting landscape photos with the Sony A7CR which is a 60 megapixel full frame camera. The resulting images from that camera have absolutely incredible detail straight out of camera, but a lot of my older landscape photos were captured on 24/26 megapixel cameras. What’s more, Lightroom has always struggled to render Fujifilm’s X-Trans raw files.

With the release of the Enhance feature in Lightroom a few years ago, I was finally able to overcome this issue of wormy/soft pixel detail in the X-Pro3’s X-Trans raw files, and the “enhanced” versions of these look great when zoomed in. It really makes a big difference to the way Lightroom renders the details.

This is something I even run on all of my Sony 60mp raw files as there is a subtle but visible difference in the fine pixel details.

When stacked against the 60mp Sony A7CR images, the 24 megapixel resolution is a little behind. If I want to make huge prints of these images, they would inevitably be a little soft on pixel details at large sizes of A2 or beyond.

Don’t get me wrong, this is not a showstopper by any measure, as I’ve made some lovely prints of some of my earlier landscape photos, some of which are hanging framed on walls, but it’s a much better experience to walk up to a large print and find that you can see even more detail than you thought possible, and to inspect the details of a beautiful framed print.

So I started to think about Topaz Gigapixel and if I could upscale these earlier images.

Note, this article is not sponsored by Topaz Labs. I purchased Gigapixel 8 myself.

Topaz Gigapixel AI 8

To try upscaling my earlier 24 megapixel landscapes, I downloaded a trial of Topaz’s Gigapixel 8 to see how it worked and what kind of results you could achieve with it. I was initially impressed with some of the results, and thankfully, Topaz Labs still offers non-subscription perpetual licenses with regular updates, a trend that is slowly disappearing.

In the end, I decided to just buy a license and have a current copy of it in my photography tool belt for those times when I want to try to upscale an image.

Let’s try this 26 megapixel Fujifilm X-Pro3 image, shot on the Fujifilm 23mm F2 WR prime lens at Mount Cordeaux, Main Range National Park in South East Queensland.

Please consider supporting my work by purchasing a lifetime license of **Like A Frame for iOS, as demonstrated above**. My framing app offers a 7-day free trial so you can try it out.

The Gigapixel user interface is fairly straight forward and works out of the currently loaded image with multiple images being shown at the bottom of the window in a list. You can edit the upscaling settings for multiple images then export them all in a queue.

When starting with a high quality image file, most of the settings can generally be ignored, but I have found that for these types of landscape photography images, the best AI model to use if the High fidelity model. It will take the high quality input image and output a very sharp upscaled image.

2x or “Less is more”?

By default, 2x resolution is chosen and for many landscape images, being mostly filled with natural colours and random textures, this will yield a good image and be generally successful, but for a lot of work it can be overkill if your images are already higher resolution.

I’ve found that 1.5x upscaling will yield slightly better results if you don’t need the massive resolution, as you aren’t asking the AI model to generate too much in between pixels and the resulting file will rival the resolution of a medium format digital camera.

24-36 megapixels: 2x upscale
42 megapixels or more: 1.5x upscale

It all depends on what you’re trying to achieve in the end and how it manages the task.

Loading up the Fujifilm X-Pro3 JPEG into Topaz Gigapixel 8.

Below is a sample of 2x upscaling (click to enlarge), which on this particular image is actually quite successful and produces a very detailed result that is 103 megapixels from a 26mp source. To my eye, though, the pixel detail is almost too crisp, and might benefit ever-so-slightly from a blur, but this is not likely to be visible at any reasonable printing size, letalone digital viewing.

The results on 2x upscaling on a natural fine art landscape style image is quite impressive! I should print this big…

Gigapixel 8 Model Settings

While the UI doesn’t require much input to get a good result, each model allows you to adjust some settings including:

Sharpness (use this on softer images)
Denoising (use this on higher ISO images)
Fix Compression (this is most likely not necessary for high quality JPEGs).

I find that the Sharpness setting is mostly irrelevant if you don’t have a soft input image, though it may help with slightly out of focus shots you need to correct. The output of the upscaling is already very detailed and may look overly “pixelly”.

Denoise may help if you have a higher ISO shot, though I haven’t needed to test this much myself as I generally try to shoot landscape photographs at the lowest possible ISO.

Play with these to see what the differences are that they make.

What About Mountains?

What landscape photographer doesn’t love to stare at mountains for hours, waiting for the perfect light or trying to find the perfect composition? This is an example of a 60 megapixel I captured on my Routeburn Track hike in New Zealand over the new year. It’s an image that at first I dismissed but then it grew on me as I realised I really liked the negative space of the clouds hiding a snowy Mount Xenicus above.

I took this as a single frame on the Sony A7CR with the Sony Sonnar T* 35mm F2.8 ZA lens from the other side of the plateau. The image is super sharp already, but why not go for mega-resolution for such an epic location?

Here is a comparison of what only 1.5x upscaling can give you, in terms of print size at the same DPI.

Loading the image into Topaz Labs Gigapixel 8 and inspecting the sample crop.

The detail is impressive and at no point feels unnatural. Let’s try the always spectacular Aoraki/Mt Cook, New Zealand’s highest mountain (3,724m). For this image, I’m going to use 2x upscaling to produce a massive 240 megapixel result (19008x12672 pixels).

Even at a high detail of 300 DPI, this image could be printed at 63 inches or 1.6 metres across.

Framed in **Like A Frame for iOS**. Please consider supporting my work by purchasing a lifetime license. Aoraki/Mt Cook from the Hooker Valley Track in New Zealand.

Aoraki/Mt Cook original image (60mp). Click to enlarge.

2x upscaled image. It fairly successfully resolves the details of the suspension bridge.

In Conclusion

Frankly, I’m quite impressed with how far upscaling software has come for photography. I have tried these kinds of upscaling algorithms in the past a while ago and they weren’t very successful, but Topaz Labs Gigapixel 8 does a really impressive job with the natural, fractal and randomised textures of landscape photography.

If you already shoot with a high resolution camera, there isn’t as much of a need for such a tool, but for those who have previous fine art landscapes or other images shot in a lower resolution and want to be able to print them large while still retaining a high level of detail, I can recommend grabbing a license while the perpetual licensing model still exists.

It’s a sad fact that the world of software licensing has changed a lot and finding software that you can buy outright is becoming hard to come by. Thankfully, Gigapixel 8 as it stands is quite affordable, especially given that you only have to pay for it once.

That aside, I’m looking forward to trying out the 1.5 and 2x upscaling on a number of my earlier works such as my Sunrise at Cabarita Beach image and others. There’s no doubt that it will improve over time as the technology gets better, but already this version is really impressive.

If you enjoyed reading this article and find the results help you make a decision, please consider checking out my digital photo framing app Like A Frame for iOS. A lifetime license only costs a few dollars.

As always, please feel free to comment below or send me a message directly. I love to hear people’s thoughts.