23 December 2024

Adobe Blog: “Removing window reflections in Adobe Camera Raw”

In this blog we describe Reflection Removal, a new technology that can eliminate reflections from photographs taken through windows with a single click. Our technology is powered by AI, but it’s not generative AI. This first iteration of the tech is designed to address only one kind of reflection — from plate glass windows that cover most or all of your field of view. It’s not designed to remove reflections from windows that are small or far away, or where the window frame is within the field of view, or reflections from objects like wine glasses, car bodies, or clouds reflected in a lake. We might address some of these applications in later updates. Our goal is to help you turn a photo you might otherwise delete into one that is good enough to share. Reflection Removal is available now in Camera Raw as a Technology Preview, to get feedback from the community, and will be coming soon to Lightroom.


That said, removing window reflections is a hard problem, and this is our first stab at it, so there are inevitably some rough edges. For example, we don’t currently do well on cityscapes at night. In fact, removing reflections is what mathematicians call an ill-posed problem, meaning that for a given photograph it’s not possible to decide with certainty which objects are in the original scene versus the reflected scene. One polluted photo could have many plausible separations. The key is training our model to understand scenes that are likely to exist in the real world. We’re constantly improving the model, so stay tuned!

Marc Levoy, Eric Kee & Adam Pikielny

Here’s a genuinely positive use of AI – machine learning to be more exact. Though the results are impressive, the use case seems fairly limited for me. There are multiple methods to minimize reflections at capture time, as this blog post admits, and in many cases reflections can be put to great creative use. The only situation where I would always apply this tool are images taken through airplane windows. Nevertheless, it’s good to see Adobe investing in these sort of niche applications of AI – maybe someday they will get around to flare removal as well.

Training example for Adobe Reflection Removal showing a winter scene with trees covered in show with and without window reflections
One of our training examples. Ordinary photographs of an outdoor scene and an indoor scene are added together to form a simulated image (right) polluted by reflections. Our Reflection Removal model learns to separate the third image into the first and the second, with the two original images serving as ground truth.

On a less positive note, days after this article Adobe also announced ‘updates’ to Photography and Lightroom plans – as you might expect, essentially a price increase. I can’t say that I’m too pleased at the idea, as here in Romania I was already paying significantly more than people in the US (the Photography plan (20GB) has the same price as in the US, but in euros, and VAT is added on top for a monthly charge of 11.89 €). Moreover, I don’t seem to have any option to select another plan or to switch to annual billing (which allegedly keeps the current annual price) in my Adobe account. I’m seriously considering canceling the subscription when my annual billing cycle ends, I think in May of next year.

Finding a suitable replacement may pose an issue though, as competitors either don’t seem to have the same feature set as Lightroom, or have equivalent prices, considering their yearly updates as well. Obviously, it is not a decision I need to take urgently but annoying all the same to start hunting for alternatives and find the time to try out and learn new systems.

Post a Comment