Topaz Photo AI v3.6.0

[dakota.wixom]

Hello everyone,

We’re excited to announce another major update to Photo AI

• Super Focus v2: A second-generation of Super Focus and now our most powerful sharpening and focus recovery model EVER. Powered by the same architecture as Recovery v2, this model is 500% faster than before.

• Super Focus is MEGA FAST in the Cloud: It’s true! This model is even FASTER in the cloud than some of our standard sharpening and upscaling models. This means the credit cost is super low in comparison to before. If you can’t run Super Focus v2 on your machine - no worries, because you’re gonna love this in the cloud.

• Color Correction, Highlights, and Shadows with Adjust Lighting v2: Our second generation AI color and lighting filter is incredible, and now you can pick whether you want color correction on or off when applying the model to your images. You can also adjust highlights and shadows to boot.

• More Improvements are also packed into this release, including a much-requested Zoom to Fit preference, new image formats and extensions (fancy JPG or JPEG? TIF or TIFF? no problem), and of course many other fixes and optimizations under the hood as well.

We hope you enjoy using this version of Photo AI. Please let us know in the comments below!

And always, if you’d like to dig deeper into this release, keep reading!

You can find the installer links and the full changelog below for this version of Photo AI. For any issues, make sure to send us an email to support@topazlabs.com

---

v3.6.0
Released March 27th, 2025
Windows: Download
Mac: Download

---

Super Focus v2 - Fastest and Most Powerful Focus Recovery Model Ever

500% Faster than Super Focus Generation 1

We’re introducing Super Focus v2, a major update with a new model architecture that delivers much faster processing compared to the previous version of the Super Focus model.

Save the cats with Super Focus Gen 2

This model is 500% faster than generation 1, and more powerful and consistent in the results than ever before. It’s so fast that you won’t even need cloud rendering in most scenarios, but more on that later!

Super Focus v2 also features an updated workflow that gives you more control over your final results.

After processing, you can adjust the sharpening intensity and fine-tune details without needing to re-run the entire process. This new flexibility allows you to explore the full range of results and find the best result for each image. Whether refining subtle textures or restoring heavily blurred details, this workflow makes it easier to get the best possible results with significantly reduced effort.

Paired with the increased speed, you can now experiment and view multiple outputs faster for a much smoother deblurring and focus recovery experience.

I’d like to bless you with one more amazing before/after example from this amazing model:

How to Use Super Focus v2

1. Click Super Focus from right panel
2. Select a model version, Super Focus V2
3. Pick a focus boost depending on your blur level. Extreme blur should use Major focus boost.
4. Use quick preview to process and review a section of the image
5. Use selection to display selected areas of the image
6. Render to process the whole image
7. In post processing change the selection or strength to tune your result
8. Change focus boost to return to the processing menu and reprocess the image

Tips for Best Results

• Use Focus Boost only on large images with a lot of blur
• For smaller images, avoid focus boost entirely, as the effect may be too strong
• If your image has a shallow depth of field use the selection to choose the subject
• Use selection to hide artifacts or unwanted detail in the background
• Run face recovery and text preserve after Super focus for best results

Please share some of your best focus recovery images with us in the comments!

Super Focus v2 is Mega Fast in the Cloud + Cheaper than ever

If you can’t run Super Focus v2 due to system requirements, that’s no problem. If you have a cloud rendering plan, then you can render images even faster than if you had a good graphics card, and cheaper, too.

I’m serious - Super Focus v2 runs so fast using Cloud Rendering that it is actually faster than some of our standard sharpening and upscaling models. We’ve dramatically lowered the credit cost for this model to reflect this.

Previously, processing a 16-megapixel image would cost 4 credits — but with our latest optimizations, that same image now costs 1 credit instead. This makes cloud processing more accessible, allowing you to upscale and sharpen images at a fraction of the previous cost.

Cloud processing allows anyone to access our most powerful models, no matter what hardware you’re using. If your graphics card doesn’t meet the requirements or you’re relying on CPU processing, running Super Focus locally can take over an hour even for smaller images. That’s why we’re bringing Super Focus v2 to the cloud, delivering high-quality results without long wait times or hardware limitations.

Here are the costs for processing your images, based on the input size in megapixels. Note that focus boost will decrease the size of the image thus decreasing the cost.

Please note that this is subject to change, as we often adjust our cloud rendering hardware and pipeline.

Highlights and Shadows in Adjust Lighting v2

Adjust Lighting v2 uses artificial intelligence to optimize the color and lighting of your image for incredible results. But of course - even our AI doesn’t always know exactly what you want!

We’re introducing a highly requested update to Adjust Lighting — manual Highlights and Shadows controls. You now have dedicated sliders to fine-tune these elements manually, giving you more control over lighting adjustments.

Changes to the sliders should render quickly in the preview to show you results when recovering lost details in bright areas or bringing out shadows.

Color Correction in Adjust Lighting v2

And there’s more! We made another change to Adjust Lighting v2 based on your feedback.

It turns out that sometimes you don’t always want the color adjustment that comes with the effect, just the lighting. This is now possible!

We’re adding a new option to Adjust Lighting to prevent color shifts when optimizing your images. With this update, you’ll have the ability to toggle color correction on or off, giving you more control over the final look of your edits.

Here are the results with color correction off (top) and color correction on (bottom). The orange Shell sign and yellow Sherwin Williams sign show the difference.

Example with Color Correction:

Without Color Correction:

Color correction will be off by default, so Adjust Lighting will improve lighting while maintaining the existing colors of your image but if you prefer a more enhanced look, turning the switch on will further optimize colors alongside the lighting adjustments. This flexibility ensures that the feature is useful for a wider range of images and editing styles.

You can also easily change the default setting for color optimization in the Preferences > Adjust Lighting section if you so prefer.

Zoom to Fit Preferences

We’re also introducing a highly requested feature that allows you to change the default zoom behavior when opening images. Previously, images are always loaded at 100%, which can be annoying when dealing with bigger images. Large images would load at an extremely zoomed-in level, forcing you to zoom out every time. No more!

This update adds a default zoom preference with zoom to fit, allowing images to open at a more practical view by default, making navigation smoother and faster.

This feature is especially useful for people with high-performance machines, as it eliminates unnecessary zooming and speeds up the workflow when working with large images or using plugins and allowing a more streamlined editing workflow.

You can find and modify the default zoom option in the Preferences > General menu.

HEIC, WEBP, TIF, and JPG export

We’re expanding Photo AI’s export options to give you more flexibility in file formats with your workflow. Previously, the app lacked the ability to export directly to .tif or .jpg, which could create issues when running workflows. With this beta update, we’re adding support for both, along with new file format options: HEIC and WEBP.

JPG fan? TIF lover? Not there’s an image format extension for everyone. Even if TIFF is still the best.

Optional: Help Us Improve our Future AI Models

We’re introducing before-and-after collection in Photo AI, starting with Super Focus v2. This feature allows us to gather original images alongside their processed versions to help us develop even higher-quality models in the future.

This is only applied to Super Focus v2 after rendering, and participation is entirely optional. This helps us understand how our models are performing, and your contributions will directly help us refine and develop better AI models that produce cleaner, sharper, and more artifact-free results in future updates.

The pop up will appear a few seconds after processing on some images. If you’d like to help, you can provide feedback through the text box or even share your processed image to show how well the model is working—or where it can improve. Once your feedback and file upload starts, you may click anywhere on the app window to hide it and continue with your editing.

But if you don’t want to participate, not a problem! You can opt-out, and there’s a button that will prevent it from showing in the future. This also will not show up if you are working offline or if you have disabled data tracking in preferences.

---

And that’s all! There’s so much more to come for Photo AI, so stay tuned.

Cheers,

Dakota Wixom
Product Manager
Topaz Labs

---

Known Issue

• Banding artifact in super focus v2 on skies and blurry backgrounds (will be fixing this soon)

Changelog

• Add Super Focus V2 model
• Add Super Focus V2 cloud processing
• Highlight and shadow slider for Adjust Lighting v2
• Color correction in Adjust Lighting v2
• Add zoom-to-fit default in preferences
• Added .tif and .jpg extension option in export
• Fix processing crashing on macOS 15 (Sequoia)
• Fix heic export
• Fix cloud credit estimation not correct for focus boost change
• Fix application not opening for some users with foreign characters in file path
• Fix errors and crashes with some Intel Arc graphics cards

[christian_p_schaefer (X)]

Hi @dakota.wixom
Is SuperFocus V2 now finally working on Intel iMac with AMD Radeon Pro 5700 XT 16 GB GPU?

[Ange.topazlabs (ange.topazlabs@gmail.com)]

Intel Mac AMD seetups are not meeting the requirements for Super Focus and this has not changed. This is because these setups are not optimized for AI processing, and stable diffusion technology used in Super Focus goes over what they can handle. This is mentioned in the System requirements. While we do not foresee them being able to handle this, we can circle back with you if that becomes the case. I have added your contact information to be contacted if it becomes possible. If you get a Mchip computer, which are optimized for AI Processing, you will not have this limitation!

[christian_p_schaefer (X)]

But @dakota.wixom and @Lingyu said, that they work on Support for Intel iMacs with AMD GPU (16 GB minimal)… Is this still the Case?

[Ange.topazlabs (ange.topazlabs@gmail.com)]

While they are considering testing these types of Intel Mac machines that have a lot of VRAM to see if possible to support these (8GB or 12Gb of VRAM or more), I haven’t heard about only 16GB of RAM ones, as Super Focus should not work on CPU alone, and this would be news to me. I will let them reply here! Potentially v2 only, since it was optimized a lot in this update. Will let them share their insight, I am curious to hear as well!

[dakota.wixom]

The issues with AMD GPUs have been fixed but the issues with Intel mac processors remain. I believe there’s only so much we can do @christian_p_schaefer almost all the other platform issues have been fixed. We reached out to Apple and have not received a resolution yet unfortunately. It’s possible they may be ending support for the platform.

[christian_p_schaefer (X)]

But the Rendering is working, but there are still these Color Artifacts visible! And that looks more like a Bug in the GPU and not a problem with the Intel Processor?

I always see a note in PhotoAI “SuperFocus is not currently supported on your GPU”!

This is clearly an Indication of an Bug with the GPU (AMD Radeon Pro 5700 XT 16 GB) and not an Bug with the Intel Processor!

[gosa]

Thanks a lot for the “zoom to fit” setting, that one will save (at least) me a lot of time.

[jayesh.topiwala]

Super focus is faster now but it process the whole image even the part of subject is selected while previewing

[Richard.J (Richard)]

We need a more aggressive (or better) Fix compression, especially when used with adjust lighting

By the way, I’d still like the option to unlock the Upscale module within the stack.

We did some rough calculations to compare the additional wasted hours required for the same type of project, versus the previous version where the Upscale module wasn’t applied at the end. The result? We’re losing approximately one hour per day.

Since we prioritize quality, as I’ve mentioned many times before,the current system often produces worse results in the all-in-one flow compared to the previous version.

To achieve similar (and good) results, we now have to go through two separate passes: first, running the upscale process, then reimporting everything to apply sharpening and/or face recovery.

This blocage of not unlocking this module must be ideological or something, because it is not technical for sure.

And i’ve produced many, many examples and proofs of how applying face recovery and sharpening on tiny image before upscale is such a waste and time and quality.

It costs nothing to let the user to decide to have the module this “all-in-one” software to be applied at the top of the stack… like it did before, for many, many years.

Again, when the change occurred, it was shocked of the quality of my output! I thought at the time that some model were messed up or something like that.

I don’t get it, I really don’t. Why nobody cares about the quality of the output of this program?

[dakota.wixom]

Hey @jayesh.topiwala - thanks for pointing this out. Just noting it isn’t a bug, this is currently the intended behaviour for all Super Focus renders, it was a design decision we had to make.

[dakota.wixom]

Great notes Richard! You’re right. We’re working on unlocking Upscale order but it’s a major architecture constraint for us at the moment. Stay tuned.

[Imo]

Not only in the 3.6.0 - seen often before - are these strange input boxes. I tried to type in something but nothing happened. The sliders itself work fine.

[Lingyu]

We’re investigating this issue. It seems to happen on devices with non-English locales.

[mikmod1]

Thank you for the extensive presentation and examples - definitely going to test these features in stable TPAI!

[Lingyu]

We managed to fix AMD gpu processing on Windows with a workaround that AMD shared with us. However, Mac processing uses different models which we do not have a solution for yet. Intel Mac is just the name we use for non-Apple Silicon Macs, the main point is that this older generation of Mac has issues with these new models.

[mikmod1]

I’m one of those users - Polish Win10 locale, but I’m using -comma- as a decimal point in the OS settings. And, indeed, I can’t enter anything, it zeroes instantly I press TAB.

There’s another bug - the value I set with the slider BEFORE I attempted manual input, [0,73], reappears as soon as I try using the slider:

And from this point on, the values set with the sliders don’t appear (but they’re probably being applied, I wonder about batch processing though):

[Ange.topazlabs (ange.topazlabs@gmail.com)]

@Imo and @mikmod1 - I have added both your contact info on the development team ticket we have for this issue happening with foreign keyboards, and will make sure to tag you here once we have resolved this! You can use the slider instead of a input value and it will apply on the image in the meantime!

[mikmod1]

Thank you for the heads up!
You’ve managed to tame this issue in Gigapixel already, partially (I can’t use [dot] there as decimal separator though)!

[gosa]

This thing with non English locales seems to be a very long ongoing topic. I remember commenting on it quite a number of versions ago, and when searching it I wasn’t the only one.

Personally I cannot wrap my head around this still being a problem considering that there’s a huge number of countries using the comma as a separator.

I challenge anyone who feels like it to Google the subject of comma vs period as separator in the world - you’d be amazed how many of us there are that are left out…