Real-World Distraction Removal
Using Generative Remove to address problems in the field
This is article #15 in my series on Practical Generative AI for Photographers. You can find more related content in my Generative AI section on my Behance page.
On a recent visit to a nearby city park, I lucked out. Within minutes of arriving, I spied this young 8-point buck foraging near a marsh. 

Unfortunately, no matter how I positioned myself, there were always obstructions to my view. I was shooting with a very long lens, so I knew shallow depth of field would help reduce distractions, but not eliminate them. 

Not to mention that focussing through branches is neither fun nor easy...

The background was VERY busy, but I liked the overall environment of the composition.

And when he raised his head to look straight at me, I knew - distractions aside - I knew this was THE shot. At no other time did he look directly at me.  
As much as I like the framing created by the two out of focus branches, the fact one limb ran through the deer's ear was too distracting for me. In all honesty, without  Firefly's Generative Remove function, I could have lived with it, but I would not have been as happy with the shot.
Eventually, the buck moved on to a less busy background and foreground, allowing me to capture the shot below, but that front-facing image stuck in my mind as a hopeful keeper.
The monochromatic look of this scene is one of the things that makes this image so beautiful. Unlike the branch in the other scene, the foreground grasses add depth and some needed contrast.
After viewing the shots in Lightroom, I could see a thin, blurry branch running through the deer’s face. The contrast was just strong enough that the branch couldn’t be ignored, and it did NOTHING for the composition. ​​​​​​​
Tech note: for both of these images, I was shooting with my OM1 Mark II, ISO 2000, using my OM 150-600mm zoom, 1/1000 sec at f/6.3, handheld.
Having tested and used Generative Remove in Lightroom often (I have numerous articles on the topic), I hoped it would come to the rescue. My very first attempt, brushing over the entire area was not a great result, but when I zoomed in further and applied multiple shorter strokes, the result was mind-blowingly successful. ​​​​​​​
As a final step, I ran the image through Topaz Photo AI for both denoising and sharpening (below). Lightroom's AI Denoise would have worked just fine (I tested it) but sadly, there is no AI Sharpen yet, and I felt the image needed both.
The dreaded finger vignette
It happens to all of us, right? In my case, I was trying to shield the lens from direct sun (and a resulting lens flare), and my hand got a little too close. I did shoot another frame where my hand stayed out of the photo, but I thought this was a good example to share, in case it happens to you.

For this image, I was able to brush over the entire area in one go and apply Generative Remove. Not only did it get rid of my hand, but all three initial variations recognized the foreground branches on the right and gave me thin, out of focus branches at the top! 
​​​​​​​
I also used masks to color-tone the scene, enhancing the blues in the shaded background and bringing out the whites of the birches while knocking back the dark tones.
Tech note: this image was captured with my OM-5, using my 12-45 f/4.0 Pro lens at ISO 200, 1/125 sec. at f/8
Tech Note: Resolution is still a limitation when using Generative AI; if those existing branches in the foreground had not been blurry, I couldn't have gotten away with the replacement branches at the top if I planned to make a print. I would have had to go with a different variation that didn't use so many branches, like the example below.

You may also like

Back to Top