DxO PhotoLab 9, the latest version of its RAW photo editing software, is one of the more significant updates in recent memory, simply because it finally tackles masking in a meaningful way. But there’s also more to it than a single feature upgrade.
Using machine learning to precisely highlight selections, DxO AI Masks sort of emulates Photoshop’s Object Selection Tool when you hover over an object, person, or image area to isolate it. It’s also possible to get there by just drawing a selection box around the object or using one of the automatic subject types, like background, sky, area, people, faces, and hair.
Fast and Efficient
DxO says it prioritized the balance between speed and accuracy to make the AI masking feel like it works efficiently. It appears that this balance exists, based on my initial testing, albeit with a beta version of the software so far. PhotoLab 9 is quick at recognizing a subject, and effective enough to separate it from others within the frame.
In cases where there may be tricky hair or fur, it is adept at capturing those errant ones to maintain the integrity of the subject at hand. It’s not necessarily perfect every time, but if you choose to try a separate mask to isolate something else, you can also opt to combine the masks or create a sub-mask to keep it all together if the need arises.
This may be necessary in more challenging images, where subjects are complex or lighting poses difficulties, making it difficult for the AI to determine where a subject’s pixels actually begin and end. Either way, I can definitely say it’s faster than Lightroom and Photoshop.
To refine those further, DxO’s previous U Point technology, brushes, Control Points, Control Lines, and Graduated Filters give you the chance to combine automated selections with these manual adjustments when necessary. DxO posits that combining AI Masks with U Point saves time because it doesn’t require you to paint or lasso anything to isolate something anymore. It even attempts to dig down to a more granular level by applying a mask only to pixels with similar colors, brightness, and texture.
It will be interesting to see how this performs once the beta is lifted and DxO updates the software over time. As is, masking looks to capture a subject entirely, though you can separate how you edit, say, a subject’s face versus their body. Except that there are limitations in what the AI identifies, at least compared to how AI masking works in Lightroom. For starters, PhotoLab 9 doesn’t distinguish between clothing and body skin on its own, nor teeth and lips, all of which can be done automatically in Adobe’s program.
Here, you would need to isolate those things manually, unless the AI is smart enough to do it once you hover over them. There are just no distinct AI masks for those details, and I found it hit or miss so far in my brief testing.
Either way, DxO’s U Point tech is robust enough to recognize at least the difference between a hand and a head, for instance. You’d still have to select both manually, but it’s very possible to create masks for them together or separately. It can even isolate parts of each based on pixel similarity, helping you edit only a portion of either one without manually brushing to include or exclude other areas. If you need to, you can always refine your selection further with the U Point tools.
Support for Mobile Photographers
PhotoLab 9 enhances file format compatibility by introducing support for Apple HEIC/HEIF and ProRAW images, thereby improving integration for iPhone photographers. But that really doesn’t stop anyone using an Android phone from importing RAW files from their phones, either.
The new feature is essentially a way to bridge across Apple’s own image formats, which is not a problem on Android since RAWs and JPEGs are pretty standard across the board. At least now, anyone with a batch of iPhone images can process them with the same level of support on the go. There is no PhotoLab mobile app, nor any announcement regarding one, so you’ll need a computer to get the photos over to the desktop.
Noise Reduction Boost
This is probably DxO’s bread and butter, and bringing in AI Masking may finally keep some users staring at PhotoLab 9 instead of denoising there and then migrating over to Lightroom or Capture One for the rest of the editing workflow. Plus, the DeepPRIME XD3 engine now works with all Fujifilm X-Trans sensors, offering improved noise reduction and detail recovery for these RAW files.
DxO’s DeepPRIME earns its reputation as a stalwart in cutting down noise and grain, and it’s always there when isolating a background or subject in PhotoLab 9. In this case, it’s much easier to apply denoising separately to one mask from another. You probably would want to do it to an entire image if it shows up uniformly to begin with, but the option to do it otherwise is readily available here.
It also works considerably faster than Adobe’s own Denoise AI, applying the denoising effect almost instantly. Not to mention the various denoising options and sliders to further refine it.
Takeaways
I would like to see DxO work with Logitech to support the Creative Console because something like PhotoLab 9 can only benefit from more precise adjustments. The sliders are a bit small, and making adjustments feels more overt than what I think Adobe users are used to. Logitech’s Console makes that feel even finer thanks to the dial.
The cool part is DxO’s AI Masks and denoising all work on-device, so there’s no need to upload anything to the cloud to get a result. Nor is an internet connection necessary to run through images on the fly.
Other workflow improvements in PhotoLab 9 include organizing related images with image stacking, marking folders and projects as favorites for quick access, and using the Customize tab to open projects directly. A notable feature is the new batch renaming tool, which can generate new file names using metadata, EXIF data, or custom text, with support for saving presets to streamline large editing sessions.
DxO admits PhotoLab 9 will run better on Apple’s own M chips and Windows 11 with better GPUs, but it’s otherwise pretty fluid. I encountered no issues running the beta on a 2021 MacBook Pro with an M1 Pro CPU.
Then there’s the fact that DxO doesn’t charge subscription fees. You pay $240 to buy it outright or $120 to upgrade from PhotoLab 7 or PhotoLab 8, and that’s it.