Gigapixel 7.2.2


Gigapixel receives yet another update today with further improvements to color accuracy across an assortment of images types including for 16-bit JPEGs and grayscale RGB images.

The Photoshop plugin for Gigapixel also received some more love with this update, and should now open an image more consistently when Gigapixel is called while a layer is selected inside Photoshop.

Gigapixel is getting better every single day behind the scenes.

There’s a few other fixes in this release - so please check out the changelog below, and of course stay tuned for more exciting updates.

Released July 3, 2024
Mac: Download
Windows: Download

Full Changelog:

  • Fixed the JPEG preserving bit depth defaulting to 16
  • Fixed grayscale images with RGB profiles not loading
  • Fixed the PhotoShop plugin not opening image when layer is selected
  • Allow esc/enter to close face selection and the recovery GPU warning
  • Fixed various button focus issues
  • Fixed button being on top of separator in upscale settings

Let’s post some fireworks in the comments to celebrate.


Dakota Wixom
Product Manager
Topaz Labs


The latest Gigapixel takes on a set of 1-megapixel digital images from 1999! I love this app :wink:

Request: There is an option for Gamma correction. I would rather have Levels, thanks, as you see these images could use a little pop that they are not otherwise getting.

The originals for reference, reduced:


Reducing the Save/Export Window to a fixed smaller size is a bad step backwards. It is hard to see how far the processing has gone, and I don’t need to see any of the prior screen in any case.
So please put it back to full size Window or have it as a user an option under Preference.

I wish that new versions would not arbitrarily alter any of the interface for no logical reason.
Another example is having the Sharpness above the Noise Reduction.

For years the order was the opposite, then once again for no good reason it was changed.
PLEASE make no arbitrary changes. The position of all items in the interface is fixed after lots of usage, and any alteration forces users to relearn where things are. Things that we look for and reach for almost by instinct.


It is presumptuous and condescending to the developers to characterize any decision they make as arbitrary/devoid of logic/devoid of reasoning simply because you don’t like it or feel inconvenienced by a change.

You may even be right sometimes about a certain change not being good or logical, but you’re using incredibly flawed and entitled logic to conclude that it is impossible for any logical thought to have driven a decision solely because the reasoning is not immediately apparent or agreeable to you.

You’re just passive aggressively calling the devs stupid, which hardly seems necessary or beneficial to the product improvement process, for not perfectly catering to your personal and immediate satisfaction, which hardly seems like a viable way for them to go about software development.

I gather that Gigapixel recovery doesn’t work properly on Intel Macs. It generates a really bad image even on the Quality setting. A bit surprising since StabilityAI’s Stable Diffusion works beautifully on my Mac locally. I’m using AMD RX6800XT gpu.

1 Like

Oh, come on, lighten up! :grinning:

Irreversible assumption, of someone creation, without the knowledge of basic punctuation, create chaotic manipulation.

I was disappointed in the quality output of v7.2.2 on a somewhat pixelated family image, just because the facial features were overly exaggerated. Using the Recovery (beta) feature however was almost spot on with those restored facial features and other photo enhancements. Great work on this feature.


I am really very satisfied that the Gigapixel still works great for me and is well controlled. I tried to improve a photo from a common (but quite good) Sony DSC-HX90V small compact camera that allows 30x zooming in on the photographed subject. According to my liking, it was a bird, a finch (Fringilla coelebs), i.e. Ms. Finch, which is not as colorful as Mr. Finch, but it was not bad. From the original size (4896x2752 pixels), I made a 960x720 crop and used Very Compressed 4x. I then scaled it down to 640x480 (whichever is the larger of the dimensions) and ran Recovery 6x hoping it wouldn’t take hours. And indeed, it took less than 20 minutes (the small input data reduced the non-linear computational complexity favorably, because also non-linearly downwards). Well, that’s all.

p.s. Sorry for the boring description of the triviality, but I wanted to commend Gigapixel and its development (and also my photo :slight_smile: ).


Thank you for sharing your process on it! Great output result!

Nice plugs - thanks for these! I’ve seen requests for levels before, making a note :slight_smile:

Thanks for the feedback Dan!

Hey Johnny - can you share the input image by chance? Dropbox

Glad you like it Stephen. Are you saying Face Recovery was too strong/exaggerated?

I am also continually surprised by the examples which the Recovery upscaling model nails. It does seem to do well on facial features.

Wow, I am amazed that this is an output from the Recovery model. Awesome! Thanks for sharing.

Recovery Model worked VERY well, where the other options like Standard, or High Fidelity, etc really distorted the faces. The Recovery (beta) did a very good job at not botching up the faces. Here is just a quick comparison. You can see how pixelated the photo was.


I’ll do that, thanks.

I was also (pleasantly) surprised. For some reason Very compressed followed by shrinking and then using Recovery (beta) works well for me. As an example, I am attaching a print-screen of the original (*.jpg) from the camera – after enlargement, quite significant flaws are clearly visible (blurriness, noise not only in the background). So I’m glad that the photo of a nice bird could be improved a lot. Recovery is an excellent tool. I look forward to further improvements!


I have selected all of the photos, but once I move from one photo to another the selection checkmarks disappear, all photos are suddenly de-selected. This is not what I want and if I forget to re-select all of them and start processing I don’t get all of them processed. Past versions had this right, but this version changed it and it wastes a ton of time.

Once I have selected what I consider the optimal parameters I almost always wish to apply it to all of the photos. I should be able to change this behavior in the Preferences.

That way everyone would have the options they want.

This is ONCE AGAIN, a change from the past 3 years of how this software was fine tuned by tons of beta testers. Why do you keep throwing out all of this work?

If you or someone requests a change like this, please give us an option to keep it working the old way.

Finally, compared to GigaPixel 6.3.3 this current version is INCREDIBLY SLOW. I can see quality improvements, but when you work with many hundreds of photos, multiplying the time needed to process them by a factor of 5 or possibly 10 gets to be a real frustrating experience. At least let us keep the old version when we install this new one.


Do you have the legacy models enabled? You can use those instead if you want the speed.

1 Like