Comparing the AI Auto Processing of Lightroom Classic CC and Luminar 2018

Recently, there's been a lot of buzz building around artificial intelligence, or AI, in all sorts of areas. And it's just starting to filter down into the kinds of photography workflows that we can all use.

For a while now, Adobe has been using their AI engine, called Sensei, in several different interesting ways, but one of the newest additions is in incorporating it into a major overhaul of Lightroom's Auto Tone function. I've previously posted some real-world examples of the new Auto Tone feature. Not long before that, the Lightroom alternative Luminar 2018 was updated to include a new dynamic filter they call Accent AI. I'd also posted some examples of using that. So I thought it worth comparing the results of both of these new AI implementations directly.

Both of these are pitched as using AI. Both apply a whole range of edits based on an image's individual characteristics. Both offer the promise of one-click processing to significantly speed up workflows and make a more visually pleasing image.

Nevertheless--and to be clear--despite these significant similarities, this isn't a true apples-to-apples comparison. That's because each of these work in a different way. Lightroom's Auto Tone function is toggled either on or off, but once it's on you have full control to change any of the individual settings.

Luminar's Accent AI filter is applied a bit differently in that you use a slider with values from 0 (off) to 100 (fully on). You don't then have the option to edit the individual component edits, but you can selectively apply it to parts of the image using the brush or various masks. And you can also combine with other filters and effects, including by using layers.

In these comparisons, I chose to apply Accent AI at a setting of 50 rather than 100 because I've found that 100 is usually overcooked, and a setting of around 50 or 60 often seems to lead to results more similar to Lightroom's new Auto Tone.

Complicating things more is that these aren't adjusting exactly the same settings. Lightroom's Auto Tone changes are fully transparent in that you can see them applied to the sliders for settings like Exposure, Shadows, Clarity, and Saturation. The Accent AI changes are under the hood, and you can't edit individual component parts of it, but you can apply it selectively and combine with other filters and effects.

So both are very much just starting points. But because the most attractive promise of both of these is in providing a quick and easy way to improve the look of an image, I still think it's worth looking at their results side by side when they're used for one-click processing, in order to give a sense of which might be a better fit for someone's preferences and needs.

Adobe Lightroom's Sensei AI vs Luminar's Accent AI: Examples

Which is Better?

Much of the time, both of these AI auto image processing features improve the image with a single click. The results are neither revolutionary nor spectacular, but these tools can be useful additions to a photography workflow.

It's hard to say which of these is better because it very much comes down to personal preferences and tastes. Personally, I find Lightroom's new Auto Tone to be too aggressive with its shadow and highlight recovery sliders. That said, it's something that's very easily fixed with a quick tweak of the slider. Viewed only as one-click functions rather than starting points, most of the time the Accent AI produces results that I prefer. But that mainly comes down to my own preference for slightly higher contrast rather than the flatter, more even images that come from using the shadow and highlight recovery tools.

Perhaps the simplest examples of this from the ones above are the shots of the Lincoln Memorial and the canal in Copenhagen. The Auto Tone function just seems to be trying too hard to do it by the numbers.

But another reason that it's hard to pick a clear winner is that the results vary image to image. While I prefer the results that Accent AI gave on many images, there were some, like the shot of the aqueduct at Pont du Gard that Auto Tone have a better result.

All of this, of course, comes with a giant asterisk. The whole point of AI and machine learning is that it's a moving target. And these are very early first versions. So I would expect that we'll see the technology get better and better in future iterations.

Overall, these results are not going to make me stop using Lightroom. There's so much else about Lightroom I like. As I've written before, I'm somewhat underwhelmed by the latest update of the Auto Tone function, but that's just one tool among many, and, frankly, while I use it sometimes, it's not something I rely on. If anything, these results just confirm what I already knew--that better results are possible from automatic processing.

Lightroom vs Luminar

Both of these apps can be used as standalone apps, and that's really how they're both designed. But they can also be used side by side, with Luminar including the built-in functionality to add itself as a plugin for Lightroom.

So it's entirely possible to use the strengths of both apps at once, if you like.