Generative Fill - Cleaning up a reservoir filled with trashy sticks and muddy water
I tried this exercise using Generative Fill in the Beta Version of Photoshop (ver 24.6).
We had a lot of rain this past winter in California and it filled up the reservoirs. In the first image, you can see how trashy the shoreline is with sticks and muddy water.
For image two, using Photoshop's Generative Fill, I drew a selection around the front and right side shorelines being sure to go out into the water past the trash and up on shore out of the water, and with that, hit the Generative Button with no text in the box, and it figured out that I wanted to get rid of the trash. I can see that in places the shoreline did change a bit.
For image 3, decided I didn't like muddy water and drew a selection around all the water in the image. And in the box filled it with "clean water" for which it produced this image complete with reflections in the water. The shoreline did change some more as it dreamed up a new look.
Generative is not based on faithful reproduction of detail. It invents as it goes and tries to produce something that makes sense.
Original image with lots of trashy sticks and muddy water
(
Download)
Cleaned up the shoreline
(
Download)
Cleaned up the muddy water
(
Download)
A nice illustration of the programming power.
That is a particularly good example of how AI is useful!
Wow, very nicely done, Jim.
Humm, did you send a bill for the cleanup... with photos to prove your work !!! We live in dangerous times with AI.
I'm amazed, but I'll stick with retality.
terryMc
Loc: Arizona's White Mountains
JimH123 wrote:
Generative is not based on faithful reproduction of detail. It invents as it goes and tries to produce something that makes sense.
Because it cannot possibly reproduce detail faithfully. It simply analyzes the surrounding pixels and then draws on its "knowledge" to create what may have been there. Unlike Content Aware Fill, it can create something from nothing, whereas Content Aware simply borrows from the surrounding pixels, much as we would have done in olden tymes using the clone stamp.
As noted in other responses, the purists will never accept this, even when there is no way for anyone to tell the difference. They will forever from now on be suspicious that every good photo is a fake, and will pore over every one of them looking for "evidence."
Yawn...
dpullum wrote:
Humm, did you send a bill for the cleanup... with photos to prove your work !!! We live in dangerous times with AI.
What a great idea! I'll prepare the bill right away.
terryMc wrote:
Because it cannot possibly reproduce detail faithfully. It simply analyzes the surrounding pixels and then draws on its "knowledge" to create what may have been there. Unlike Content Aware Fill, it can create something from nothing, whereas Content Aware simply borrows from the surrounding pixels, much as we would have done in olden tymes using the clone stamp.
As noted in other responses, the purists will never accept this, even when there is no way for anyone to tell the difference. They will forever from now on be suspicious that every good photo is a fake, and will pore over every one of them looking for "evidence."
Yawn...
Because it cannot possibly reproduce detail faithf... (
show quote)
Yep!!! They will do this.
Thanks for sharing your work!
Wow, you did good! Very nice.
I don’t think I’m going to believe what I see in a news photograph is reality anymore.
Stan
terryMc
Loc: Arizona's White Mountains
StanMac wrote:
I don’t think I’m going to believe what I see in a news photograph is reality anymore.
Stan
Only now, when for decades any photo could be imperceptibly altered to change the reality? Journalism still has ethics and standards, and I see no reason to believe that AI is going to force photojournalists to cheat any more than they have in the past...
If you want to reply, then
register here. Registration is free and your account is created instantly, so you can post right away.