>>1550364
maybe painting all the speech balloons with a similar hue and then adding noise?

File: result_0.png (2.15 MB, 832x1248)

2.15 MB PNG

>>1550364
Photoshop has a content-aware fill tool, but haven't tested it for comics yet.

Also using this https://unwatermark.ai/ai-photo-editor/ with the nanobanana model I just prompted:

"Remove all text, replace with clean empty space, professional"

and 2 seconds later it gave me pic related, is not perfect but you can experiment, for example you can cut problematic parts, or you can try different more complex prompts like >>>/wsg/6039111
You can prompt stuff like maintain colors or add a paper texture etc.

Lastly to maintain as much of the original as possible, what I do is on photoshop just place the new ai image on top and use only the textless parts, there's like 5 different ways of doing that depending of the image, most times I just duplicate the ai layer + use some selection tool + select inverse + feather edges a little + delete + merge and remove small imperfections manually, sounds like a lot but having keyboard shortcuts for all of those makes it almost instantly or only takes a few seconds but I mostly remove watermarks on photos.

>>1550386
This is actually really useful.Thank you.

File: vdfnqzwdydrma0ctvdp8s2gj5(...).png (2.25 MB, 864x1184)

2.25 MB PNG

>>1550481
>Remove all text, replace with clean empty space, professional
Funny thing is AI removed speech baloons entirely not just text