Copy link to clipboard
Copied
Hello everyone, I'm trying out the new generative fill feature that uses artificial intelligence, and it's truly impressive! However, I'm currently experiencing an issue, and I'm not sure if it requires specific settings or if it's a problem that still needs to be resolved since it's in beta. The problem I'm facing is that when I generate an AI-assisted edit, the resulting image is blurry and unusable. I'm attaching an example here. In this case, I enlarged the original image and had the AI reconstruct parts of the background that didn't exist. If you zoom in, you'll notice the clear difference between the original photo and the reconstructed part. Am I doing something wrong with the settings? Please let me know. Thank you.
I am using a Macbook Pro M1 Max
Mac Os 13.3.1 (a) (22E772610a)
AI is limited to resolution of 1024x1024, Meaning it will not generate clear results if your image is bigger than that (There is a workaround thought)
Please see this video: https://www.youtube.com/watch?v=pGJ0KjT4Xds
This will fix your issue most likely 🙂
Copy link to clipboard
Copied
Admittedly some presentations I have seen do not do the limitations of the Generative Fill feature full justice.
Copy link to clipboard
Copied
of course its cumbersome and not a valid workflow especially when you need to edit hundreds of pictures.
What kind of editing do those hundreds of images need that cannot be achieved with other, so-to-speak »classical« methods?
Who created the images?
Copy link to clipboard
Copied
i work on photo sourced textures and i need to fix bad photos, remove people and unwanted objects, make the image tile, etc.
so far the results are a mixed bag and i might move onto a more "classical" program without ai and without 12 euros a month
Copy link to clipboard
Copied
I can't help thinking about how this will be tended to if and when the element gets a general delivery; the transmission of the information for pictures bigger than 1024px x 1024px to and from the Adobe servers appears to be a conceivably enduring bottleneck …
Copy link to clipboard
Copied
I can't help thinking about how this will be tended to if and when the element gets a general delivery; the transmission of the information for pictures bigger than 1024px x 1024px to and from the Adobe servers appears to be a conceivably enduring bottleneck …
Is this a rewrite of a previous post of mine on this thread?
Copy link to clipboard
Copied
Ha, that is funny. I mean it's not a high crime of plagerism of fiction writing, its more an agreement with your viewpoint and brought up again in a different conversation. It does feel a bit like an attempt to be perceived clever based on the writings of another. Speaking of AI, we all now wrestle with the thoughts that maybe your original thought was helped or entirely created by AI. Not saying it is, I would wager it's not but everything has to be questioned now.
Copy link to clipboard
Copied
Some of the recent posts seem like possible AI-generations.
But I am not sure who would »slop« Fora such as this one and to what end exactly.
Copy link to clipboard
Copied
My image is 1080 x 1080 and i am only trying to use Generative fill for a portion of my image and it is still producing a lower res fill than the rest of my image. What am i doing wrong here?
Copy link to clipboard
Copied
Could you please post screenshots taken at View > 100% with the pertinent Panels (Toolbar, Layers, Options Bar, …) visible?
Ultimately the feature may just not live up to your expectations …
Copy link to clipboard
Copied
You are not doing anything wrong its just that the AI results are made in low res and not fit for purpose.
Copy link to clipboard
Copied
The 1024px x 1024px-maximum Generative Fill produces should (pretty much) suffice for a section of 1080px x 1080px.
Copy link to clipboard
Copied
I have also experienced the blur with the extended AI area. I tried it on a landscape (5616 px x 3744px) and was initially impressed with the results until I zoomed in to 100%. I found the areas that were unsharp in the original were matched perfectly but where the foreground was in focus the generated extension was noticeably unsharp.
I didn't see a size restriction (I looked at on the Help section within PS). I looked at Adobe's site (https://www.adobe.com/uk/products/photoshop/generative-expand.html) but didn't see a restriction there.
I did see the limitation discussed in Julieanne Kost's very informative video (https://www.youtube.com/watch?v=VhYQsA7leFM) where she shows the option to enhance the generated detail (about 7 min 14 sec in). I triedthis but the results went from blury to something else quite unusable.
I'm not sure 1024x1024 pixels is very high unless you're just using an image for the Web. Otherwise it's just not going to be useful.
If Adobe did manage to scale up the technology then I think it's a game changer, but, as of August 2024, it's of limited use.
Copy link to clipboard
Copied
Someone recently mentioned that Generative Fill should render 2000px x 2000px now (if so I suspect it’s possibly 2048px x 2048px).
A test with a graphic checkerboard seems to indicate that might still stil 1024px x 1024px for Generative Fill.
See the difference in below screenshots.
706x1014
1129x1945
Copy link to clipboard
Copied
This video is helpful: https://www.youtube.com/watch?v=QtIq5HPNHu0