Gigapixel 7.1.2


This release of Gigapixel increases the stability and usability of the application across the board. We have added a camera profile for the Sony ILCE-9M3, and also fixed an issue where certain DNGs would cause a crash. Lastly, we updated the EULA for Gigapixel to bring it in-line with our other apps for commercial applications.

This update for Gigapixel also brings a number of fixes related to rendering statuses, zoom, crop, display, and memory issues. We fixed an issue with TGA files sometimes not being able to open due to a mis-match in format.

If you upscale textures or other non-photo images using Gigapixel, please let us know in the comments what other file types we should be looking into!

Gigapixel is getting better every single day behind the scenes.

As always, a full change log is below. Please stay tuned for some exciting updates.

Please give us any feedback or report issues with this release. We’ll be updating Gigapixel regularly to address your feedback and any issue reports. But also - please do feel free to share any great export results you are getting with the application!

Released April 26th, 2024
Mac: Download
Windows: Download

Full changelog for version 7.1.2:

  • Gigapixel’s EULA was updated for commercial usage
  • Add camera profile for the Sony ILCE-9M3
  • Updated components and icons to bring them in-line with Wonder UI
  • Disabled zoom shortcuts when the Recovery mode is active
  • Disabled the tooltip for comparison in recovery
  • Fixed an issue with Photoshop sometimes returning 1 for the PPI
  • Fixed an issue with some TGA files not being able to open due to mismatching format
  • Fixed an issue where clicking preview caused re-rendering
  • Fixed an issue with scale resetting after rotation in crop
  • Fixed an issue where certain DNGs would cause the app to crash
  • Fixed an issue where deselected faces were not showing
  • Fixed an issue with status sticking while blending faces in Face Recovery
  • Fixed an issue that caused the cache to exceed normal memory usage limits
  • Fixed an issue where flipping the aspect ratio in the crop tool shrunk the selection
  • Fixed an issue where no scale was selected on custom width and height
  • Fixed an issue with the plugin not closing if no images are loaded
  • Fixed an output size misalignment issue in the open file view
  • Fixed zoom levels being reversed in recovery for small images
  • Fixed multiple text and icon misalignment issues
  • Fixed some dialogs overlapping the title bar on windows


Dakota Wixom
Product Manager
Topaz Labs


Thanks Dakota, I just noticed the update when I fired up the app a moment ago!

Did the lazy in-app update which seems to work much better now, no issues.

Quick full-autopilot upscale of great-grandma before further testing :wink:

Upscaling a low-res logo image 4X:

And upscaling a previously reduced-in-size iPhone pano 4X:


It already worked well for me on an M1 Mac with v 7.1.1. This version does as well, or maybe better for everything I have tried.

I noticed an oddity that may have already been there, but it is this:

The amount of Face Recovery has an opposite effect on sharpness in Recovery mode, compared to all other GP or TPAI modes. More Face Recovery make the face sharper in Recovery, whereas in other modes, more Face Recovery seems to make the face smoother but less sharp – either may be desirable depending on the image.

This is an observation, not a complaint, and it is probably not a bug. The only possible change I could imagine would be to change the name from Face Recovery to Face Enhancement when one is in Revovery mode, just to create an expectation of differing behaviour and to encourage users to learn how it works.



SO much better than the previous version. This one actually works! :slight_smile: Will we ever see the option to mask out areas we don’t want enhanced? This is a particular issue with AI, where eyes or hairs are over sharpened. If the Photoshop plugin is used, it’s not a problem, since the over sharpened areas can be masked out to reveal the bottom, un-enhanced areas. But thanks for the update. I had been communicating with Justin (Jason? I deleted the emails) about the issues I was experiencing on my 2015 iMac, my MacBook Air, and my M3 Mac. No issues so far.


I bought Gigapixel back when it first came out (2018) to support a project involving upscaling a thousand *.dds game files. I had to save each one as a *.tif first, and then save the resulting upscaled image back as a *.dds. I also had to extract the Alpha layer and upscale that separately, and then recombine it. It would have saved me a ton of wasted time if Gigapixel could read the native *.dds format.

Warning, within the DDS family there are many subtypes. Even if you only did DXT1 (no alpha), and DXT5 (interpolated alpha) that would helpful. Similar to the TIFF with an alpha layer issue, Gigapixel would need to upscale both the Texture, and Alpha layers correctly, and leave them intact (not flattened) when you save the result.

1 Like

I don’t know if I’m missing something, but I don’t see how to change the models in the compare view. It would be very useful to be able to replace any poor models with others while staying in the view. (So basically similar to v6.) Is this possible somehow, or if not could you please consider it, now we have so many models to play with?

(Remembering which ones to keep, going out of the Window and then selecting those and two more you might not screw up once, but wanting to try a couple more afterwards is getting to be pen and paper territory to keep track… Sometimes an unusual model choice will be the best for what you want to do with an image, especially if text and photographic images are present.)

P.S. Literally the only other change that would interest me is if Standard V1 isn’t the Standard model that was in 6.2.0 can we please have that as, say, Standard V0, as it avoided a number of noise-like pixilated effects.


Wow love that pano

Hey samb - I have noticed this as well. We are thinking about disabling it by default for Recovery – I honestly get mixed results. Sometimes it does well, sometimes it does not. Let us know what your ratio of results is. I agree on the workflow/behaviour note either way, great points.

Hey Daniel - for masking and selection, most of those capabilities will be present in Photo AI, whereas Gigapixel will focus on whole-image enhancement. There’s still much to do in this domain, and we’re excited about the future for Gigapixel.

Glad to hear this is working well for you overall!

Thanks for the note Andy! Yes, I’d like to take a look at alpha channels in general for Gigapixel as well.

Thanks! These pass for clouds these days…


Hey jml you cannot currently change models in comparison view, and I agree it would be a nice thing to be able to do! I’ll leave it at that :wink:

Standard V0 - do you have any examples of better results vs v1 or v2?


If I have Topaz photo AI, is Gigapixel worth getting also? - is the algorithm different or better?

1 Like

Hey Robert, yes - I have all the Topaz apps and I regularly use them for different purposes. It depends on your workflow, but Gigapixel currently has THREE new amazing upscale models, including a diffusion-based generative AI model called “Recovery”, Low Res v2, and a model that excels at Text & Shapes. And more to come. It also has a comparison view which I personally love.

TPAI will be receiving model updates as well of course, but Gigapixel is the powerhouse for upscaling. Do stay tuned for more updates to both apps. If you need iterative denoise, sharpen, and masking or work heavily with RAW files, you will want to use TPAI.

Sometimes I send my TPAI files to GP and back for upscaling. Many users have workflows like this.

It’s tricky as, alas, you can’t have two versions installed at the same time, and I can’t easily go back and run images that don’t work well currently (very very few of which I would have kept anyway) with 6.2.0 (as 6.3.3 had the issue IIRC) - I’ll see what I can do…

1 Like

Bring back Quick Save please! Quick Save function to Export > Save > Close image

1 Like

What’s happening?
Selecting Low res v2 makes the images blurry.

I have to note something. During the update, I had an Internet problem and it didn’t download all the models again as it always does. But the installation worked well.

Is it possible that some are corrupted now? If so, how do I force Gigapixel AI to redownload them all without reinstalling the whole application?

Edit - I used the installer to repair the installation. It works again


I have both (Gigapixel as well as TPAI). I recently tried to update Gigapixel (from v. 6 to v. 7) and I don’t regret it at all. Stand-alone Gigapixel can already for me do a lot, and at the moment it seems to me much better than its implementation in TPAI. Just a quick attempt at a “portrait” of my cat Racho – I couldn’t improve it properly in TPAI (but it’s quite possible that more experienced professionals could do it, and better than me). In Gigapixel 7.1.2 it was quick and easy even for me (an enthusiast, not a professional): Crop → Scale 2x → select Very compressed. Of course, it could probably be improved further, but I just wanted to find out how Gigapixel 7 works. And it works: Excellent automatic Denoise and Sharpen and then selected Scale 2x. Only the AI-model didn’t offer a good solution, but I also tried “Very compressed” and it worked nicely (but I have no idea why this, it’s a mystery to me). I will try to attach a picture (printscreen).

So I wouldn’t be afraid to use a stand-alone Gigapixel – I’m very comfortable working with it thanks to the good UI (still avoiding the TPAI update after several tries).


the new low quality v2 delivers awful results where the older v1 made at least acceptable ones! might depend on the image though.

1 Like

So far I’m finding Low Resolution V2 introduces artifacts that V1 doesn’t. Then when tweaking the sliders to reduce artifacts, some are improved, but others are made worse, like an unwinnable game of whack-a-mole. In fact High Fidelity V1 generally produces better results than LR V2, because it doesn’t amplify artifacts as much.

Ironically, it seems that LR V2 doesn’t handle low resolution images very well.