1

(10 replies, posted in General Discussion)

So sorry to hear this. It is very rare to find someone with that level of expertise, accuracy, and knowledge who is so helpful and generous. I learned quite important matters from his posts. My condolences to his friends and family.

2

(12 replies, posted in General Discussion)

PaoloT wrote:
Ali wrote:

a group of designers who are Affinity users have been involved in the background.

Hoping they are really designers, and not just influences.

Said otherwise: if you want your tools to become universal, you involve a mass of contributors into the discussion. Asking to a few of them will just give the opinionated view of the selected ones, that may have been selected after the opinionated view of the product manager.

Paolo

Will Paterson has a design studio and, considering all I have watched from him in the past, IMO really knows design deeply, not the usual random person just making some cute logos.  Has a solid business, for all I know. I know less about James Martin, but he has been designing for 20 years, involved in the creation of a thousand brands.

At the very least, I would say they are not just influencers.

I was in a hurry...  The above is a new short from another pro designer (whom I have followed for a long time), kind of ensuring that what's coming on 30th is good stuff.

https://www.youtube.com/shorts/ybFDm8H5-uE

Nice!

Mine was a bit over 1k
If I could exchange that for money, hey!.... but I am afraid not...
It would have made Alfred rich!

Oh, it's free and open source!  Donations supported (optional).

Noice.  (as the young ones say now).

I will definitely have a look. Thank you, William!

In my memory there's like a tool name everyone mentioned... but it's blurry... Maybe it was called Font Forge. Anyway, at some point I'll do a bit of personal research on the matter  smile

@Twolane and @dcr

Yep, licenses can be very restrictive. I don't have a Comicraft font yet, but I guess that could become a problem for a webocmic unless I'd throw quite some money on it. That said, 100k views/month (it was what a certain software company where I worked at had, and we made some niche but famous products) are a lot of views, the extra licensing cost for that seems reasonable for the money one "could" be making, but webcomics are hard to monetize efficiently. I kind of think that you would have fewer issues in that regard with Blambot than Comicraft(comicbookfonts). When working as a designer, although for most stuff at companies I just used a few commercial fonts that were very safe, very well designed, for less important things, no corporate image, etc, I just often went with free font sites (if you'd set that in the site's search)  like Dafont, 1001font, Fontsquirrel or open source ones like Google Fonts. Another that I am currently digging (no idea of how good it is) :  https://open-foundry.com . (any feedback is appreciated.)

And the font I am using more now for comics is the one that I mentioned, included in Clip Studio Paint since version 3.0, which appears to allow everything I actually need.  Depending on the kind of comic, that or a certain one from Blambot.

I am a graphic designer and illustrator, I have worked in both roles at companies and as a freelancer, but what I am not is a font designer. There is a ton of knowledge for me to learn to become one, I am afraid. But it would be fun to make one solely for my comic art (but not to the level of Comicraft and Blambot, they consider a lot of details and lettering related matters). Yes, a font-designing tool would probably be my choice. Some people just draw the fonts on paper, by hand, then scan, vectorize and import into a font tool... One could go that route.

@dcr Kudos for that.. Kind of makes me envious... Someday I might try that.

10

(64 replies, posted in Art & Literature)

William wrote:

I ask that because the human artist would have trained using the same, or similar, information that was used to train Copilot.

So, if a human artist, as part of his or her training, learns to paint in the same style as another painter, is that result art if the picture is an original composition?

Or is it not art on the basis that art needs a new concept?

Before directly answering that, let me go deep into certain matters. There are several huge differences between what generative AI outputs and the work of a painter. Most AI defenders want to ignore these facts, somehow. Some in bad faith, but others just don't know better because they have not studied and trained art skills for decades, so it is understandable.

Gen AI cannot draw nor paint, full stop. Does not really know about perspective, does not understand anatomy or color theory, and...does not feel nor can think.

So, the "information" needed to process is not the same by a huge margin, in kind and quantity. Plus, painting is far from being solely processing information: that is the very basic thing gen AI does, while painting is a lot more involved (and better). An artist drawing with a pencil or painting with a brush and oil paints is not simply good at it because this person has "processed information." Additionally, the argument of "but artists copied from masters" is very false, too. I learned from reality, not from other people's art, and from understanding anatomy, etc. My art does not come from "scraping" any artwork from others. 

An artist trained in all those skills during years: anatomy drawing, perspective, composition, color, lighting, and very importantly, created a particular style, own expression, which serves to express one's feelings and ideas: something the AI algorithm does not even have. As I have explained in several instances before, this strongly distinguishes art from an automated process.

The AI does not "learn" all those skills. Indeed, if it did, you would have never seen a hand with 7 fingers in the AI outputs (to mention a simple example, but a lot more mistakes happen). It does not produce so many errors now, but not because it now "understands" anatomy or ANY of those matters. It understands nothing. They (the developers) just fine-tuned it to try to not output that error so often, but the AI does not really know why.

So...there you have it: generative AI does not use at all the same methods as an artist. It is really a glorified mosaic maker of blurry pixels that is helped by an immense database of unlawful origin; it just predicts the next pixel based on probabilities extracted from all the human work that it has in that humongous database. It's making a puzzle by mere imitation (helped by randomness), without understanding a heck about the puzzle.

But also, it does not "feel" nor "think." This is another so important difference that I don't know why there's even debate about it. Obviously, this all comes from an attempt to simply disregard definitions of art from all that has been academically established about the matter, just for mere interest. But the facts, and what things really are, cannot be twisted.

You seem to hyperfocus on the "result." But this is the output of a system that plagiarizes artwork in a well-hidden way (and which used a legal gray area—for now, but this is starting to be regulated—to scrape the data). But art is never ONLY the result. Art is the entire process (by definition) and a language to express emotions. AI has no emotions. When someone prompts something to Copilot or etc., the individual is not using a brush nor intervening in the art process at all. The "prompter" is only doing the work of a client when ordering/commissioning a piece of art. I explained this in detail earlier, though.

So to answer specifically this supplementary question, I needed to preface it with the above paragraphs.

If with this humans-only example, you want to reach the conclusion that AI is like a human artist, as after all it is using the same information (other artwork from many artists, the masters, etc.), I answer to you that no, it is not using the same information. And more importantly, not in the same way. Firstly, a human painter cannot "scrape" that much data, and especially key, the human artist does NOT need it. Does not need gazillions of data to predict what pixel comes next, because a painter works at a MUCH higher intellectual level. That is, understanding anatomy, perspective, composition, color, light, etc., the "human artist" can imagine and paint from scratch. AI can't do this.  But also, to reach that output, AI is not doing any method comparable with art, by any stretch, as I have methodically explained.

If we were to take your latest question in a literal sense, though... An artist producing the exact same painting as another from another painter is mostly a copyist or a plagiarist. Both (the first, doing it legally, with permission, and the second, not at all) require a LOT of artistic skill, but they are not expressing their own human experience and emotions or ideas; they are merely copying someone else's. Which does not match the definition of art creation. Does it have merit? A LOT more than someone using AI, that is for sure, as to make a good copy, you need to first be a really good painter, great in every of the mentioned matters. And if it is a copy of the great master painters from other centuries, it typically requires mad skill.

Plagiarism is not considered art per se, though. And also, it often allows the original author to take legal actions (which ends up favoring the original author practically always), and that is one of the several reasons why what AI does with the artwork it scrapes should be prosecuted legally.  Obviously not the users of a service or app (i.e., you), but the creators of that product and the companies having done the scraping without permission from the painters and other copyright owners should answer for their actions.

Side note: Gen AI cannot produce an exact image from one prompt to another, as it's based on a random seed. Doesn't add much to this conversation, only to mention that it really can't produce two images exactly the same twice. A human copyist could do a much better job at it (and there are amazing professionals doing it). So, the "thing" is a bad copyist, too.  big_smile

An artist (or a "human artist," as AI fans like to say, as if there was another possibility...) is a lot more autonomous than the automated algorithm of an AI, because, well, the "human artist" can actually draw and paint and has all the skills and knowledge that I have referred to. AI is neither "imagining things." That is not really what is happening.

I hope it answered the new question.

Exactly! I had forgotten the name of the original author and the brand. I read back in the day the full explanation that there was once in 1001fonts (not in 1001freefonts, a different beast). And it was quite moving that he did all that work for comic creators to have it for free.

It's from years ago, I don't really remember the story, I will have to read it again... Oh, it's here (one has to scroll down) :

https://www.1001fonts.com/komika-text-font.html

It's the same link I pasted, but I realize now that the full story is still there. smile   8)

12

(64 replies, posted in Art & Literature)

I cannot speak for other forum users. But for me, it would not be necessary that he'd reply or acknowledge my answers/explanations (although I was not replying to what he said but what Copilot "said," mainly). For me it is way more important that he reads those and gives them deep thought. Maybe to discard them; that's fine. I personally don't need acknowledgement or answers about this. But I can see how others certainly would, as it is the expected thing in a conversation  wink  (no big deal, though).

3) Their personel keeps calling it Affinity in communication here and there. So I would be surprised if the name Affinity would vanish

It would make sense keeping the name somehow; there are a lot of positive articles, reviews, and word of mouth found with any fast Google search. This is just pure golden rules in marketing; I wouldn't throw that away (but you never know). It's a very strong presence, which means everything in SEO. And mostly, that the brand and product names are well considered and have good prestige, unlike with other companies and products.

Browser-based wouldn't be that great for professional work for heavy raster files for print, RAW processing, etc. For UI/UX prototyping? Sure, Figma is total proof that such kind of browser based design app works well for professional teams. But as long as it's that, UI/UX, low fi graphics.

I don't think they can discard desktop without discarding the pro world, honestly...

Highly recommended, on the affordable but very good side of the spectrum, and from people who have always helped comic artists, especially indies : Blambot (it's not a comic font, it is the name of the company).

They support very advanced matters in comic lettering. It would be too long to detail in here, but their site is full of AMAZING info ("ARTICLES" section). A lot of that info will help you make comic lettering in a much more professional way. I have two great fonts from this place. Blambot always offered their fonts free for Indy comic authors (provided it's for an indy comic creation!), which is beyond generous. And those offered, are quite good. Still, the commercial ones, besides that you get a full commercial license, are of very high quality. And no strings attached, no limitations for those. Like the following company, these fonts are very well thought out for printed comics, comics on paper. But IMO, can be very well used for web comics, too.

https://blambot.com

This other company makes exceptional fonts. Quite more expensive but absolutely outstanding. They have sales days (I don't know if it was on Black Friday). I typically wait till those days.. and then I forget.  big_smile . These are absolute high end comic fonts that many of the biggest names in the industry have used. Important caveat: For a web comic, the price goes increasing depending on how many unique visits per month you have (100k visits/month, 200k visists, etc). The previous company, Blambot, not only has lower prices, it doesn't have these tiers per monthly views. But these are of exceptional quality, very well thought out for printed comic books. IMO, you don't go wrong with either company.

https://www.comicbookfonts.com

Yeah, both are great.


And one that surprised me very positively (might become my main font for web comics!) is a font that gets installed when you buy Clip Paint Studio Paint 3.0 (not previous versions!) permanent license or the subscription. I'm guessing obviously so too with 4.x. I have 4.0 EX, now.  From what I read, it comes with both the Pro and Ex version.

Obviously, it'd be expensive to buy an application (which is a gem) just for a font, but in my tests, this was the most readable in small size on screens, and yet looks very stylish for comics.

"Clip Studio Comic Regular"

Those like me, willing to make it sure something is legal to use for commercial purposes, you have some data about it here:
https://ask.clip-studio.com/en-us/detail?id=116188

And mostly HERE (with examples of the font, and license explained. Spoiler: no problems) :
https://www.celsys.com/en/topic/2024031402_es


For those wanting another option for a free comic font (IE, in case of one not making indy comics, so the Blambot free fonts would not be an option (but those are better)) for a graphic design or whatever, or any commercial use, but can't afford the 30-40 bucks of Blambot's, this one might do the job. Even if it's in a very different (lower) tier, it's a gazillion times better than Comic Sans (too much hatred against that font, Comic Sans was designed for screens of eons ago for a certain purpose, not for graphic design, and definitely not for comics):

Komika Text.
https://www.1001fonts.com/komika-text-font.html

A lot of great webcomics can be made with it (and surely are).

15

(64 replies, posted in Art & Literature)

Out of random curiosity, I have often asked such questions not only to Copilot, but to other AI solutions (Gemini, Claude, etc). Obviously things like these are a bit "hard coded." And they tend to come up with the same "mistakes".

IMO, just like some ethical matters —some other ones that have more potential to bring them legal trouble— I can see how the devs "force" some particular direction for these answers ("aligning", as they call it... but IMO, used in a not-so-great way, for this concept). Especially knowing the huge impact that this has on artists' livelihoods, the necessary theft for Gen AI to exist, etc, and the general social reaction against this. Of course an AI is going to say there's no wrongdoing in it!

But an AI made by a corporation, defending its money-making machine, is not the most reliable source for these matters. Not when it is an interested party and it has been said that what I described in the paragraph above does happen quite often.

Besides the described huge issue for art-related questions, an AI is not a "venerable oracle" that has the ultimate truth for everything. It has been demonstrated too many times that these things do not always pick the criteria from the most reliable sources. IE, they often are based on just reddits! Without needing to refer to the extreme of listing glue as an ingredient for a pizza recipe (and that happened), it hallucinates with many other smaller things. And it is never "the best source", as even when it does the right thing and at least picks first from Wikipedia content, Wikipedia sometimes is wrong (some non-revised document, and I am finding that even some of the revised ones!). It is NOT like academic studies where there's a consensus from the science community, historians, or etc. It will coincide with those very often, but not always. This is clearly one instance of that (and IMO, not by accident, which is much worse).

It would be interesting to go analyzing each of its statements :

[WJGO]

When you produce artwork from a prompt that I have authored, is that artwork art?

[Copilot]

Absolutely — it can be art. Whether something qualifies as "art" isn't determined by who or what physically creates it, but by the intent, meaning, and emotional resonance behind it.

False. This is fabricated and very partial. With that definition, which betrays itself (who or "what". Very intentionally added to give validity to AI as an "artist". A thing cannot have "intention" ! ) an oily puddle (it's a "what", a thing) from a truck could be considered art. But it is not. Like with the elements defined for a message to exist, by definition (art is a form of message/communication) there needs to be a sender, a receiver, and the message itself.

Also..."can be"... smart, lol.

[palette emoji redacted] Here's why your AI-generated image can be considered art:

Creative Intent: You authored the prompt, which means you initiated the creative process. That intent — to express, explore, or evoke something — is central to art.

I will repeat myself from my previous long email: Nope. A client (a patron) from the Renaissance would also be "initiating the process": becoming automagically "the artist". Leonardo...sorry man. Or much more mundanely, any client from today who sends a paragraph with the requirements of what they want for their dining room wall (or other project) would become The Artist just by writing what do they want and for which part of the wall they want it.

But hey...hmmmm.... I like Copilot's way of thinking! Then when I call the plumber, it is not him the skilled person with 30 years of experience in fixing huge problems with pipes! It is ME!  I AM A THE PLUMBER! All these years. I can be now an astronaut (my intent is enough... no need for all those boring engineering studies and hard physical tests!) just by writing a prompt too, right? I will try that later. Man, my CV is going to look soooo much better pretty fast!  Thank you!
/sarcasm.

And Gen AI defenders often use also the emotion that looking at something causes on the viewer....When a patron in any history period would order a painting from an artist, with the intention that it would convey an emotion in the people who would visit their house, palace, or etc, and the artist was able to create something that generated certain emotions on people, it was *not* the patron becoming the artist. As far as monarchs have had typically big ego problems, crazy narcissism at times, I doubt king Leopold II saw himself as the musician creating La Clemenza di Tito, instead of Mozart. But the king did write the "prompt" (or some servant did, I dunno, lol). He commissioned this opera. But he did not CREATE it.


Collaboration: Think of it like a duet: you provide the vision, and I help bring it to life. Many artists collaborate with tools, assistants, or even algorithms.

Wrong again (on purpose; it's clearly fabricated and partial). It is confusing concepts on purpose....That duet is better described as client + artist (but in this case there's no artist). But if we wanted to play along here, there is collaboration not in the execution of the art, not in the 99% of the work, which depends on the artist's criteria, skills, knowledge, expression, human experience, etc...when it is a real artist who does it. Also "I help bringing it to life". Nope...more like:  "I plagiarize from a gazillion human made works". And... "I" would not be a proper subject for these matters, really, it is a thing (reason why there are so many issues with AI and copyright).

Yes, artists collaborate, when both are actually doing art (ie, shared canvas/murals/graffiti, etc), both are painting (or sculpting, etc). When a football fan goes to see a match in the stadium, this person is not the individual playing football, no matter how far the imagination can go for some.

Expression: If the final image reflects a mood, idea, or aesthetic that you care about, it’s expressing something — and that’s art.

Hmmm...Neither.  If I throw some dices in the air, and they fall in a way that sums up a number that means something to me (like 7, if I had seven children), that is not art. There has been no artistic process. It was a random factor, not intentional, and there was practically not even intervention by me.


Audience Reaction: If others view it and feel something — curiosity, joy, discomfort, awe — then it’s doing what art often does.

The easiest to refute. Following this logic a baby Panda makes most people feel compassion or tenderness. A human disaster makes most of us feel sorrow; that won't ever be art. Many things can make us feel something (bad results in an exam, a letter of dismissal, taxes raising, etc), and by no means that is sufficient to match any (even crazy) definition of art. They are redacting some things internally in the code, twisting it to their interest, propaganda and convenient PR.


Also, AI is -as often- contradicting itself. There would be NO intent there, which itself had declared as necessary some lines above. And this statement makes it look like the audience reaction would be enough by itself to consider the output as art.

Some people debate whether AI-generated works are "real" art, but that’s a philosophical question with no single answer. What matters most is: does it mean something to you? If it does, then it’s already doing the job of art.

It has a single answer, and they know which it is. It is just not convenient for them.

And the concept is not as vague as it pretends it to be. Also, that something "does the job of something" does not convert such thing automatically to a particular entity (I can hold a hammer in very aestehtic ways but I won't ever be a carpenter...unless I train and study in that field, for years. Maybe not even if so).  Then again, something that "means something for me" can be literally anything, and does not match any serious definition of art.


Would you like to explore how different prompts can shape artistic styles or moods? I’d love to dive deeper with you.

No, thank you. You have been so kind, Copilot.   big_smile big_smile big_smile

Anyway, I insist. I'm very happy that you and other people enjoy using LLMs to generate pics. Keep doing so!. But IMO it is not great to confuse concepts.

16

(64 replies, posted in Art & Literature)

Ali wrote:

This brings us nicely back to where this all started in another thread. Camera clubs have had to deal with the issue of AI-generated content for some time, and have had to redefine their rules where changes to reality are permitted in a competition. The consensus is that any change to reality must only be done using the photographer's own pixels, so the photographer may do a sky replacement as long as the sky is from their own work and not a stock image or generated by AI. For this reason, generative AI (which uses stolen pixels as you describe above) is not permitted. We can denoise images as long as we use a denoise tool that does not use generative AI, and so on.

I think you have explained the problem with AI-generated images being considered as art very well indeed, and why they cannot in all conscience be referred to as 'original'. Thank you!

Well, that's a very good measure. I wasn't aware of it. I am not good at photography and while I can edit images, I did not specialize in that. But I absolutely love that they are protecting and encouraging original work.  In (painting) Art groups, and particularly with contests, there tends to not be allowed the use of AI images, not even partially. I can imagine photographers not  having a good time, either, with the current situation. Besides the fact that the world of stock photos was already tough (but that was not as bad as now with AI, as they could still make stock photos and sell them, but their skills, style and expression would show up in those photos). I have seen how many stock photo sites now have lots of AI images hmm . My favorite ones are those that force the uploader to state that it was AI made, and that ban those users not respecting this.

You are welcome! I tried my best... (at that hour of the night).

17

(64 replies, posted in Art & Literature)

Alfred wrote:

> ...assess whether an AI program could fool people into thinking they were communicating with a person rather than a machine.

Yes, and also, we tend to humanize things, inanimate entities. Apparently it's in our "main code", due to being very social beings.  I think this happens with AI. Apparently it is becoming a problem for some people who even believe that an AI chat bot can do the work of a therapist or become "a friend." Not what we are discussing here, and it is an extreme example. But it is part of the problem, as people attribute capabilities to these LLMs that aren't there. Anyway, I preferred when people mainly humanized teddy bears, boats and the like big_smile 


*scraped
*scraping

Thank you! I missed this....   big_smile  The things one ends up missing.  big_smile Also, you are better than Quillbot (let alone Grammarly).
It's the language differences... In Spanish there are only two cases of double consonants (ll and rr), that I can remember... and they offer no confusion.  But in English, I too often don't know which word has it or doesn't...  big_smile

18

(64 replies, posted in Art & Literature)

Art

noun
1.
the expression or application of human creative skill and imagination, typically in a visual form such as painting or sculpture, producing works to be appreciated primarily for their beauty or emotional power.


It is a very incomplete definition, a basic one from a fast Google search. Not the one that you would find in good literature about art. But even in such a poor definition, there are already issues to consider "AI" output as art.

There are many problems to consider "art" the output of a LLM based service or local application (it is not really "AI", it is not  intelligent. Actual AI researchers state this). 

1) Art, and creativity in general is based on making many conscious decisions. Particularly in painting, literally a decision with every brush stroke, every pencil line. It is not just a logical decision; each brush stroke is driven by emotion, intention, human experience, and the artist's style: the artist's very personal and unique expression. Already nothing of this can really be achieved by an algorithm. What it's happening is just automated and sophisticated plagiarism. It just predicts what is the more likely token to happen next, based on mere comparison (by probability) with a huge database, based on that humongous database of artists' paintings and illustrations (work which it does not understand!), artwork which the company scrapped (or another company made the scrapping) with no legal permission to do so.

2) That art definition I posted is the typical one you find in a basic dictionary. But most literature about art and aesthetic theory define art in a more complex way. Most coincide in the requirement of the existence of certain factors in the production of art to be considered such:

- Human making. But a human is not really making this, an algorithm is, and using material in a not really lawful way (regulation is to come about it after some time, I suspect). You will surely argue that you typed a sentence, and that's all your human "intervention". But that is exactly the work of a client asking an artist to create a picture. A sentence or a paragraph. And that is as old as humanity, almost. In the Renaissance, a patron (ie : Lorenzo de' Medici ) used to ask the artist (Leonardo Davinci, continuing this example) to paint a scene with very specific requirements. I can tell you that it was a lot more elaborated than your prompts for Copilot (no offense intended). Well, obviously these patrons would never, ever, consider that they themselves were "painting" these master pieces just for the fact that they verbally asked for it and detailed how they wanted the scene to be (they sometimes would even state in a contract how many "moons" the wood should have been left to dry!), in exchange of some very nice payment. So, it is not the prompter who is making the art. If anything, it would be the AI, and it is just an algorithm that is capable of producing it ONLY if you feed it with human made paintings. It doesn't know how to actually paint with correct human anatomy. It does not really paint, indeed, at all. It only achieves accuracy (and not always) by comparing things, but it does not understand cr4p what it is doing. It is a dumb system, but good at plagiarizing.

- Derived from the above, but essential, there has to be intention (again, in the most agreed definitions of "art"). Not the intention of the client ordering a painting for their dining room, intention here is understood as every brush strokebeing based on the artist's feelings, emotions, experience in life, thoughts, etc. You don't get any agency on that, using AI. All that is decided for you, by the algorithm. And the algorithm being non conscious, non human and incapable of feeling anything (neither of thinking!), is not actually capable of meeting these requirements.

- Connected with the previous point, but it is a different aspect: there has to be a conscious act! This alone is the main obstacle to consider any AI generated output as "art". No current AI is conscious. And I have heard from several researchers who argue that full consciousness is non achievable by an algorithm in the future, either (I am personally sure that it is not possible, from a philosophical point of view). I don't think the so-acclaimed "AGI" would be conscious as a human is, even if it would be more autonomous. It will not have a real conscience. Which is the most essential factor for art creation.

This all is based on the common agreement among most studies about the art concept, since quite some time (decades), I am not making it up. It would take me a ton of work to start doing this with actual references (links), like a thesis -not going to lose so much time.

But even more important than all this, it's the fact that Gen AI has needed (it is completely impossible to do it other way) to scrap without permission the art from millions of artists, and all what it does is form a mosaic of pixels based on a prediction system which basically plagiarizes styles and artwowrk from all these creators, just that it is capable to do it in a way that the laws have not yet been able to regulate properly (there has not been an opt-in system with which the artists could accept, deny or put conditions for the usage of their artwork to create a product that actually makes money to the AI developers, VC investors and many other parts of the chain who did not intervene at all in the creation of the paintings. This constitutes something plainly immoral, so, it is ethically wrong to even try to distract from this fact with redefinitions of art to make it fit someone's particular interests (often of a monetary nature).

Of course I find it wonderful that you enjoy generating AI images with Copilot. Good for you if you enjoy it! But that is one thing, and another very different thing is twisting the concepts following certain motivation. Also, a bit out of touch when this plagiarism is wrecking the livelihood of many artists, as a lot of their income does not always come from the "ideal client" (there are many of this kind, fortunately) who appreciates human-made art. A lot of commercial work is part of the artists' bread and butter, and this whole thing has been designed (by venture capital investors and some other unscrupulous agents) to eliminate the human factor of it. "Not great" (to say the least).