Thanks! Big problem solver for me! I print a lot, and haven't been all that pleased with the generative AI because of the low resolution.

Thanks for a super-interesting point. I come more from the sound technician world, so I see great similiarities all the time… and yes 90% of the people will not notice when you are doing some cheating, and yes compression do not matter much if you are doing it a little bit one time you listen to song.
But if you are doing one compression after another compression… and repeat the process 30 times over then will your music start sounding like crap and deep bass and high frequency sounds will get removed and everything sounds ugly.

And that is the reason why I wanna avoid compression at all costs. It always easy to turn a high quality music file or image and turn it into compressed junk, but you cannot do things the other way around, and turn junk into high quality.
And yes it is complicated and time consuming to avoid lazy solutions that leads to compression. But if you get compression on your entire image every time you add a generative fill (like in Leonardo AI) then will your image looklike crap in the end…. and everything looks blurry, and the richness in colors are gone.
So it is time consuming to add only the edited area into GIMP on top of an uncompressed BMP file of the original to avoid as much compression as possible.
But the end result is always hundredfold better and worth it, when one works with time consuming projects where you might need to do a hundred changes to the original image before you are done.

Photoshop have both pros-and cons compared to Leonardo. Leonardo offer more options and allows you to steer its generative fill in Canvas to a higher degree. But on the other hand do you have to first paint into the image what you want, and then in the next step let Generative fill fix things.
In photoshop you got no ability to steer prompts other than with text, and you cannot use negative prompts, and you cannot decide how many % you wanna change the area you wanna modify – which leads to crappy results. But photoshop could still sometimes be superior if you accept a 100% to the marked area in an image.
But as pointed out here… generative fill is a lazy solution with flaws. Perhaps things change in the future. But for the time being do some other methods have their superiority.

When generative fill with AI first came out I thought it was pretty cool and solved some tough case situations. After using it for a while, I grew to dislike it and have almost entirely abandoned it. I found that I spend three times as long trying to make things look right with it rather than using various older tools and refining the result.

Thanks. Really helpful.

Thanks, Matt. I completely forgot about contrent aware fill!

Great Tutorial Matt. It's good to have this reminder of how good Content Aware Fill is before all this AI came along.

Many thanks for the reminder!

Thanks Matt great video as always

Great video as usual. Thanks!

AI remove/fill from Photoshop is a game changer for photographers.

Great stuff, thanks!