Repeatedly, I have found Gigapixel to be unable to improve on noisy images, even unable to recognize faces, when the original file is actually quite large, but of poor quality (low res, noise, out of focus, etc.). Reasons unknown.
It helps greatly to FIRST reduce the file size (in another application) to become commensurate with the original image quality. Then, all of a sudden, Gigapixel does what it is supposed to. And you can scale the image back.
When the file size is too large, Gigapixel thinks this is a high quality image. Downsizing tricks it into doing what we actually want.
Hi Gerhard. Recover Faces was only intended to work with small low resolution and low quality images for example, group photos with several people.
It was never designed to work with large images as you have discovered by reducing the image size then running Recover Faces in fact, using Recover Faces on larger images you may experience the opposite affect of reducing the quality by softening the photo.
Here’s the link for Topaz information on the Recover Faces option
there seems to be a misunderstanding here, on your part: I am not talking about the recover function, but generally about the ability of Gigapixel to improve on certain image files, regardless of which type of method is being used.
And the failure I was referring to in my post is not only about face improvement - some images aren’t improved at all unless you do what I described. And then, all of a sudden ……
Sorry if I used a wrong catch phrase there. I still think my post of valid.
Hi Gerhard. Recover Faces was only intended to work with small low resolution and low quality images for example, group photos with several people.
It was never designed to work with large images as you have discovered by reducing the image size then running Recover Faces in fact, using Recover Faces on larger images you may experience the opposite affect of reducing the quality by softening the photo.
Here’s the link for Topaz information on the Recover Faces option
Utilize Face Recovery to improve images where facial attributes could be enhanced. Low resolution, blurry, and unclear details of human faces are best used for this feature.
Without seeing your original image it’s, impossible to properly identify the issue perhaps it would be possible for you to post the image
Gigapixel 8
Gigapixel is primary used for upscaling and only has minor denoising for fine adjustment on individual AI Models. It doesn’t have the same AI denoising or sharpening capabilities as Photo AI.
Faces are selected automatically with the button to the right of the Face Recovery label turned on however, the user is still required to select Apply.
You don’t have to be in another application to downsize your image before sending it to Gigapixel.
The ability to downsize an image before upscaling has always been a function of Gigapixel
Within the panel on the right under the navigator window, you’ll see the Upscale section.
Under the word Upscale you’ll see a scale group 1x, 2x, 4x, 6x and Custom.
Under the Scale group you’ll see Scale Factor with a number.
If an image source came from scanners, screen captures, or somewhere online; it may have the issue of false resolution.
False resolution happens when an image is upsampled, boosting pixel count without adding real detail. The image source may have large dimensions, but the extra pixels are just interpolated and don’t reflect true information from the original image.
The solution is to downscale the image first (downsample, reduce pixel dimensions). Downscaling removes excess pixels and condenses real detail, restoring clarity by eliminating false resolution.
| 
