Well, it won't actually work on a mono image. But it will if the image is converted to RGB.
Here is one where I was standing too close to the bridge and decided to make it look like I was not so close. Expanding the front worked well. I tried expanding to the right, and it had no idea that there was a ravine with a creek at the bottom. So that attempt failed.
I did like what it accomplished. And I like the added shade from an unseen tree beyond the right. Looks very convincing that the lead up to the bridge is legit.
Next. I tried a side view of the bridge to see how it would be handled. Neither end of the bridge is in the image, so I thought I would try to see how it ends the bridge. On the right, after the 1st expansion, it was getting confused and I decided not to go any futher. On the left end, the 1st expansion kept the bridge going. Same with the 2nd, and I wondered if it would ever end. But it did on the 3rd expansion. It didn't know how the bridge should end, so it designed the simplest one it could. Not the other two Generative Fill choices kept the bridge going.
Before Generative Fill
(
Download)
After two sections of Generative Fill to move the camera farther away
(
Download)
Before Generative Fill
(
Download)
Generative fill - 1 time on the right, 3 times on the left
(
Download)
Generative fill works much better when you take small bites
If you want to reply, then
register here. Registration is free and your account is created instantly, so you can post right away.