I think it’s just your perception, I was responding with full respect to the other participants in the discussion.
1 Like
Good to read that.
1 Like
I have no issue using that tool, so why would I care if someone else knew? We can even ‘erroneously’ call it AI. Still wouldn’t give a shit.
But: Whether you, me, or anyone else here agrees with it, the use of the term AI is very broad. In fact, I think the toggle to enable it in the upcoming darktable version uses that vernacular. Since it is so broad, it seems reasonable to be transparent about it in broad terms too. Then we let other’s decide how they want to interact with our work. Maximal freedom for all involved!
I have no problem with being Jewish, so why should I worry about anyone else knowing? I can even wear a “Judenstern” on me. Still wouldn’t give a shit. – Don’t you feel the fundamental flaw in your idea?
Please, don’t decide for others whether they want to be accused of alleged “AI editing”, as a result of them clicking on an image and easily getting a mask (which they would have done in a minute or so otherwise).
“Maximal freedom for all involved!”. In short, I am completely against this idea. Let’s be adults. There is no need to participate in either the AI hype or the AI demonization hype.
2 Likes
Choosing child rape as your example came over very badly. There are other issues with what you said, but dealing when them would move further from productive discussion. Perhaps another example would be better?
I’m not interested in the marking part of the issue, but do want to explore the differences between the two de-noise types. How is the AI de-noise architected? AI systems tend to be opinionated and will nudge images towards their training data. Does this system to that?
Tell me which ones, let’s discuss… Let’s explore together what issues were in my answers. I wonder what the claims against me will be, considering that I didn’t say a single bad word to the person I was responding to, but I firmly opposed unacceptable ideas.
Maybe. Maybe it should have just been said that there is a presumption of innocence and to say that something “has a lot potential for unethical things” is completely unacceptable in any non-totalitarian society?
1 Like
This isn’t the correct forum to be discussing the finer points of rape and pedantry on such a sensitive topic would be insensitive anyway.
I read it as a note that this is an area where potential issues are more likely and that care and review is required. If that were enough to ban a topic, I’d agree with you, but I don’t think that was what was meant.
I just want to say I don’t want to discuss this anymore. I think I my perception of people being not nice to each other here is a bit off for this.
I will not replay here anymore. Please ignore me suggestion.
1 Like
Thank you for understanding that not being nice to very questionable idea and opposing it doesn’t mean not being nice to the author. Thanks for the discussion.
It wasn’t an idea without merit. Many other programs and groups are trying to do similar -ish things.
Now that dt has opened the doors to ai, those in charge need to start having these discussions. User input is not ignored, so that means we should engage in these discussions as well.
Getting the ball rolling is a good thing.
You do know that comparing the experience of photographers who use ai to denoise and mask (which are acceptable use cases even in Magnum’s eyes) to the bigotry, persecution, and genocide of the Jewish people for thousands of years is deeply offensive right?
Let’s lay off the cringe analogies shall we?
Please, I would kindly ask you not to promote the cringe idea from this thread. I am very sad that you do not realize all its cringiness and unacceptability, instead accusing me of demonstrating it with, yes, very unpleasant but direct analogy.
Again, if you want people to know about YOUR edits, you are free to add these tags to YOUR images. Please do not decide for others what they want, they will add the information they want themselves.
I am discussing the absolute unacceptability of the IDEA, I am not attacking the PERSON who promotes it (we all can make mistakes). Therefore, I kindly ask you not to accuse me of deep personal offenses, since this was not even close.
I suggest not continuing this discussion here. I find it unpleasant to read your accusatory response. Thank you in advance.
Respectfully,
Victor.
Does not make sense. You can change metadata. You can use external tools.
And most importantly: nobody cares.
I just used AI to generate keywords. It added “beach” to an untagged image with, well, a beach - how does that effect anyone in any way? How does it differ from me writing beach manually?
Just add a tag ai created and the problem is solved for you.
1 Like
Your are funny. I like you!
Its completely optional. I am not sure why you think anyone is deciding for you!
Ok. Lets set aside what I see as the casual anti-antisemitism implied in your analogy. Opinions on what counts as anti-antisemitism can differ right? Since it is such a ‘direct analogy’, it should be simple to put your argument into the standard argumentative form:
- The Analog has property 1,2,3,4…N
- The Target has property 1,2,3,4
Therefore, - The Target probably has property N.
Take your time. I’ll wait.
darktable is about choice. It gives you options to use it as you want. darktable explicitly doesn’t force you to use it in a certain way. Therefore forcing users to somehow disclose that AI (or any other module/process) was used doesn’t align with darktable’s values, IMO.
In the interests of full disclosure, darktable has had AI for years. There is facial recognition and image recognition and tagging integrated as Lua scripts.
On another note there has been a LOT of discussion (sometimes heated) about what models are “acceptable” for darktable. All the models that are used by darktable have their licenses available as well as other information so that users can make an informed choice.
2 Likes
But I think that this functionality can be relatively easy be build in a lua script?
I am talking about the auto tagging feature, using some for of neural network.
But I would say it would be nice if the Content Credentials would be integrated. Not for people to not able to hide that AI was used, of course that is your own choice, but the other way around: to prove AI was not used.
This is good to know. Some groups specifically use the tags Adobe provides (I think other software has that stuff too? Not sure.) as the first filter for ai use. Because darktable has not been including any ‘ai’ (acceptable or not) in the image processing portion of its functionality, they were not concerned about my using it. Since dt will never flag in ways that fit their process (even if it implements ‘genAI’ in the future), I guess I will have to double check that.
As a side note: An opt-out default setting for any given functionality is, in no way, forcing anything on anyone. Dt writes nothing in metadata which is not easily altered.
For tools which have generative capabilities, having output tagged as AI is useful for those training models so they can avoid them. I don’t think that will be an issue for dt for some time though.
Opt-in is probably better. That way it is a conscious use choice.