Richard.Taylor wrote:
Iceberg1962 wrote: … I tried all the tips mentioned above and they don't work.
Hold it; didn't I just post pictures showing editing a face without pixelation? Why would it work for me and not for you? What are you doing differently?
Generative ai algorithms like this can produce unpredictable results and are something of a black box. We don’t always entirely understand what the “logic” is behind the results the algorithm produces (particularly since it’s not actually using logic in a critical thinking sense, as it’s not sentient and does not have intent, obviously).
I am trying to remove some straight horizontal black lines from a watercolor swatch. I had actually already zoomed/cropped the swatch in quite a bit because I needed it filling up the entire square canvas.
I have no idea why it keeps activating the safety filter on this image, especially since it seems completely inconsistent about it: it applied it immediately when I tried removing the right line, but started to let me erase the left line (fading it out a bit). But when I tried to remove the remaining faded left line, it activated the safety filter for that one too!
I finally gave up and started over using the Repair tool in Photomator instead, which worked perfectly (and was so much faster and easier—it removed each line in one go, no need to go back over them). I use this app for the bulk of my photo editing and I highly recommend it! But it does require a subscription though, it’s not free like just using iOS photos. (To be clear, I don’t work for them or anything, I’m just an artist and am picky about having good editing tools, so for me it’s worth the subscription).
I’m assuming if enough people submit reports about this issue, Apple will fix it eventually, as this clearly isn’t intended behavior. I’m wondering if the safety filter algorithm just hasn’t been trained sufficiently yet or is over-tuned and recognizing way too many things as “sensitive” content.
But yeah ideally they should just let us choose when to use it? I really don’t understand at all why they’d automate something like this, especially since it’s not fully automated, it only activates when someone chooses to try to remove something, and if they’re already making a choice and doing something manually, letting them have full control over that choice makes much more sense.
