Topaz Labs; the creators of an AI range of photography utilities (e.g. Topaz Denoise AI, Topaz Gigapixel AI, Topaz Mask AI) have released an update to the Gigapixel AI package (Version 5.5.0).

This update brings a number of changes, including interface improvements, a new AI engine which allows processing images faster, and supercharged AI models allowing you to upscale and enhance low resolution images up to 600% and offers improvements with newer hardware including slightly sharper output than the legacy models.


Topaz Labs have provided the below from the changelog:


Major Features

  • AI engine has been brought into Gigapixel. This should greatly improve performance on modern hardware.
  • All models updated for those on newer hardware. These models should be slightly sharper than the legacy models
  • New model – “Very Compressed”
  • New trial flow. No more 30-day trials, but instead a trial will place a watermark on your image
  • “Send Feedback” option in the Help menu – have a feature request you want to send to the dev team? Use this option to send us a 500-character report.


  • Interactive tooltips have been added for each model, to get a quick sense of what each can do
  • Controls tutorial has been slightly updated
  • Better caching when saving


    • Preview window should not show as blank before previewing for the first time
    • Filenames should elide in the middle instead of the left
    • DNG support should be fully-functional for those coming from pre-5.4.x versions of Gigapixel
    • “Loading images” text no longer overlaps the FileListView caret
    • Saving text has been changed to show you what the program is doing when processing (downloading, processing, finishing, etc.)


Users with a current upgrade licence should be able to download this upgrade at no cost through any of the links in this page, otherwise if you are interested in purchasing or trialing Topaz Gigapixel AI you can also follow the links.

Prev DxO PureRaw
Next ON1 NoNoise-AI

Comments are closed.

%d bloggers like this: