I used the generative fill for removing object but the output was worst. It made photo unusable. Few months back it was working pretty good. But now it's output is bad. It's not even looking real.
Thank you for your feedback. I encourage you also to thumbs down results like this and provide feedback:
I like to add a note to describe specifically what is wrong.
I have noticed myself on generative fill, more generations seem to be required to find the right result with model 3. Our collective feedback will helpfully provide information to Adobe to adjust the model.
One additional tip, try selecting a larger area around what you want to replace. I am not sure what you are selecting, but if you are selection only the object you want to remove by closely selecting to the edge of that object, the results will be poorer than if you also select around the object as well. For example if I start with this image and want top change this spot:
Not so good selection:
Better selection:
Result:
I hope this helps. Let us know if we can help further.
Yes you are right. But I have been using generative fill for past few months. I have edited photos. I am even them on my Instagram as well. What I meant to say is that the results few months back were good. But now suddenly the results are not the same.
Take a look at this photo there was a person sitting on the road but I removed it using generative fill. Look at the results but now it's not the same.
@Watch It37257370ncv3 , in general, I agree the new model has made generative fill harder. It needs some updates. I think Adobe expected that as it is listed as being in "preview".
If you need to create better results faster, you might consider using gen fill in Photoshop which I believe still uses model 1 (for now). Photoshop beta is using model 3, so it there is a visible path for Photoshop to be updated.