Andy Bell Photography

Image Resize Without Losing Quality using Topaz Gigapixel AI

How to do an Image Resize Without Losing Quality

Easy! Just put it into Photoshop and do a re-sample, right?

Absolutely not!

If you are enlarging an image, you face this reality – pixels need to be added and your image’s quality will suffer. But only if you use a standard method, such as bicubic or lanczos to do the enlargement.

There is a better way.

I first reviewed Topaz Gigapixel AI a few years ago, and it amazed me. I had never seen such high quality enlargements. Since then, Topaz have steadily improved Gigapixel and this post will show you how it performs against standard bicubic and lanczos enlargements that editors such as Photoshop and Affinity offer.

The Challenge

I selected a few images from my collection and resized them all to 200% using Gigapixel AI, Bicubic and Lanczos (using Affinity photo). In then compared the results. Testing resize software is easy – there aren’t many options to choose from and judging the results is straightforward. So, let’s see the tests.

Test 1 – Landscape Photo

Gigapixel AI vs Bicubic vs Lanczos

As you can see, this is a nice, detailed image. Viewed at 100% you can see it is sharp. It’s an ideal candidate for upscaling.

Here are some comparisons of the results. The output images are named by the tool used. Click the images to see them full size.

All three resize methods have done well, but Gigipixel’s is sharper and more detailed. This means a better print.

Test 2 – Landscape Image

This image, too, is detailed and sharp. Let’s see how the tools handle resizing it to 200%:

In each case, the Topaz Gigapixel AI result is noticeably sharper than the others. A superior result leads to superior prints and happier customers.

Test 3 – A portrait

Portraits have their own challenges. You want the picture to look sharp but not in any way unnatural. Upsizing pictures of people will expose flaws with both the photo and the resize method, and people will notice this! Especially if they are the subject of the picture!

Gigapixel AI has, once again, produced the sharpest and most detailed result.

Gigapixel AI is the best way to resize a photo without losing quality

The above examples all tell the same story – Gigapixel AI produces excellent results. With high quality, high-resolution images Bicubic does OK, but Gigapixel betters it significantly. My previous tests have shown that Gigapixel AI is vastly better than ‘standard’ resize methods with massive enlargements and/or lower resolution images.

But is there a downside?

There are things to consider when using Gigapixel AI:

  • Cost – it’s another tool to buy. However, it is a single purchase with free updates for a year. But if you rarely enlarge images, then it may not be worth it.
  • Time – Gigapixel AI takes longer to resize an image than Bicubic or Lanczos.
  • Time – Gigapixel AI now offers a quite a few options, and I found I had to preview each of them before deciding which settings to use. This isn’t a tool that does batch processing well unless you are upsizing a batch of very similar images.
  • Hardware – Gigapixel AI works best when you have a supported graphics card and the latest graphics driver for it. It will work using the CPU instead of a dedicated graphics GPU, but then it is much, much slower.

Image Resize Without Losing Quality – Conclusion

If you enlarge your photos for any reason, then Gigapixel AI is the best way to do it. There are a couple of scenarios where Gigapixel AI is unbeatable:

  • Enlarging to make a big print
  • Enlarging a cropped, or heavily cropped, photo. This is significant news for wildlife photographers, where you need to get close to the subject, but that is not always possible

Adding Gigapixel AI, Sharpen AI and DeNoise AI to my toolkit and workflow was one of the best decisions I ever made.

Try Topaz Tools for yourself, using these links:

Gigapixel AI

DeNoise AI

Sharpen AI

The Photo Browser/Image comparison tool used in the screenshots is Digital Photo Guardian by AB Photo Tools.

Exit mobile version