The new A.I. Gigapixel from Topaz is being released today as an app for upsizing pixel-challenged images or monster-sizing RAW ones. Beta testers are commenting that it beats the current market leaders in resampling quality.
About A.I. Gigapixel
From the press release: Topaz A.I. Gigapixel is a standalone application made for batch resizing your images (it is not a plug-in or Topaz Studio adjustment). It uses the power of Topaz’s proprietary Artistic Intelligence engine to make upscaled images sharper and clearer than traditional upscaling tools which use “interpolation” (bi-cubic, Lanczos, fractal, etc.). These traditional methods produce images that are blurry, unrealistically painterly and lack the details that are present in real high-resolution images.
A.I. Gigapixel is the first desktop application to use the power of Deep Neural Networks to realistically enlarge your images up to 6x.
The unsurpassed quality of A.I. Gigapixel can enlarge full-size RAW images to over a gigapixel in size. It can also enlarge thumbnail images to create full-size versions that are practically identical to original high-resolution photos.
(Note: Topaz is not promoting a coupon code or discount for this release but we are! Use coupon “plugsnpixels” here to save 15% off the $99.99 retail price (this discount also applies to ALL Topaz products).
My experience (See my more recent tests for additional eye-opening results)
Experimenting with the pre-release version of A.I. Gigapixel, I went directly to my oldest digital images (slightly mushy, circa-1999, 1-megapixel, 1536 x 1014 @ 150ppi Kodak DC265 JPEGs) to see what it could do to wring some further quality out of them.
First, here is a screenshot of the summary of the settings that are available to work with. As you can see, A.I. Gigapixel is not complicated at all. First choose whether you want to scale or custom-size the image enlargement, and whether to add corrective enhancement or not. Next, choose an output format with related batch or individual settings, and finally set the output parameters:
The test
Hint: Right-click and open these images in a new browser tab to see them at actual size, where the results are more accurate.
My first attempt was with a photo of a fountain statue, shown here in a Photoshop screenshot as follows: Original at left @ 100%, a 200% enlargement and a 400% enlargement (both saved out of A.I. Gigapixel and shown at 100%, though your device might downsize my examples). As I said above, these late-’90s 1-megapixel images are a bit mushy at best, but A.I. Gigapixel did a fine job of enlarging the subject data that was present while maintaining the level of detail that was there.
Here is another image taken with the same Kodak around 1999, with the original at left at 100% and a portion of the 600% A.I. Gigapixel enlargement at right, also shown at 100% in this Photoshop screenshot:
For fun I enlarged the original in Photoshop to a 600% screen view (not an interpolation) so you could see the actual improvement A.I. Gigapixel provided (at its 600% scale setting) compared to the original pixel data:
I think A.I. Gigapixel did a fantastic job with the water, given how little data it had to work with, and the curves and lines of the architecture look much better as well.
The competition!
On the photo forums, people are asking how A.I. Gigapixel compares to Photoshop’s new “Preserve Details 2.0” Image Size feature. So I pulled out a 2007, 4-megapixel image and worked at a 600% enlargement. First, here is a reduced version of the original so you can get perspective:
And here are a few comparison enlargements from various areas of this image, with Photoshop CC 2018 latest build’s results on the left and A.I. Gigapixel’s on the right (100% screenshots as seen inside Photoshop):
Now off to CalTech (with another late 1990’s 1-megapixel shot, shown here in reduced size for reference):
This time we’ll give Photoshop a better chance, using both Preserve Details 2.0 and 10% noise reduction, with Topaz Gigapixel on the bottom (with Enhance Image checked), 400% enlargement:
Here’s a final example, a 2008, 6-megapixel image, shown here in its entirely (reduced):
And again, with the 600% enlargement treatment, Photoshop on the left, A.I. Gigapixel on the right:
WOW!
Be sure to grab the 30-day trial and give A.I. Gigapixel a spin on your own data-challenged images, or even bump your RAW images up to gigapixel size!
System requirements
I did the examples and screenshots above on a 2013 i7 MacBook Air (8 gigs RAM, and just meeting the minimum graphics spec). It handled even the beta version of A.I. Gigapixel just fine, though a bit slowly of course. Tip: Dedicated graphics cards are always better for digital imaging app performance (as opposed to integrated cards such as the Air’s).
Here are the official requirements:
A.I. Gigapixel is compatible with 64-bit Windows 7+ and Mac OS X 10.11+. It also requires a GPU with OpenGL 3.3 or later (Intel HD5000 or better, Nvidia GTX 760 or better, AMD R9 280 or better).
(Note: Topaz is not promoting a coupon code or discount for this release but we are! Use coupon “plugsnpixels” here to save 15% off the $99.99 retail price (this discount also applies to ALL Topaz products).
Your purchase of any software through the Plugs ‘n Pixels website or this blog gets you the best deal available and also helps generate a small commission to help keep these resources online. Please subscribe above to receive automatic notifications of any new posts. Thanks again!