AI, art, guru disappointment, AI done right

Like most of the world, in 2025 I'm still coming to grips with AI and how it can be useful in our daily lives.  I suspect there is a place for it, and I hope AI will be doing the stupid boring jobs I hate.  Leaving all the fun jobs to me.  Although, it seems a bit the opposite so far.  hmmm.  Time will tell.

In the Art World, AI is making big ripples.  As I'm about a year into my journey of learning to paint seriously (and two years since I started to dabble).  It's been very interesting to see how AI has changed the art world in this short a time.  

And I don't yet know how I feel about this.  That's what I'm exploring here.  Some observations, some situations I've encountered - both in person and online - and how it's making me feel about AI and the way humans interact with it.


There are three events that happened to me recently.  I want to share them with you and how these expierence made me feel.  The first two expose the limitations of... um... not so much the limits of AI.  These are known.  They exposed the problem with using AI in art.  The last event gives an example of someone using AI in art without these problems.

THE CONTEST

Here's the first time I noticed AI being used in art.  



Basically, there was a contest.  And this contest had rules.  Rules included not using certain digital assists.  They seemed pretty clear.

Now the video goes into a huge amount of detail with the problems of the AI used by two of the artists.  One used the AI art as their digital version of their submission then when they won, they replaced it with a painted version of lesser quality.  The other painter (who also placed well) used an obvious AI image as a reference photo, but looks to have painted it himself (we know this because pre-ai, he was an amazing painter).  

The conclusion of the video is basically, the judges ignored the rules and got called out for it.  And if you look at the pinned comment by one of the judges on the video - he didn't care about the rules.  

...

That's all well and good drama and stuff.  But the impact of this on me and my journey into the art world is a bit different.  

Right now, I'm at the phase in my journey where I've learned to paint good enough that now I am searching for what I want to paint.  What is "good art"?

More importantly, what is art that is good enough for me to paint?  I'm looking to history to learn this.  I'm also looking to contemporary painters.  What do they paint and are any of them painting stuff I like?

One of the categories I'm drawn to is Imaginative Realism.  And considering the ARC people who run the contest claim to be all about promoting traditional arts in realistic style, they seemed a good starting place to learn about what good, realistic art looks like

The issues raised in this video hit hard.



This painting was removed from the contest after some people pointed out the ai weirdness and the difference between the submission on the left, and the actually painted work, right.

Both versions are not a style I enjoy.  Before the drama, I was mildly confused about why they placed so well in the contest for any kind of realism, imaginative or otherwise.  I shook my head and moved on.

One of the arguments in favour of this "painting" is that it falls under the style of Waterhouse who painted beautiful works of what we would now call fantasy art.  But I don't think it looks a bit like his work as the person seems flat and the background is missing something important to the story.  With Waterhouse, one can see immediately some of the story even if we don't already know what it is.  

When I see art like this and I discover that it's made by AI, it gives me the impression that AI just isn't good at making art yet.   If it makes such bad decisions, why is AI art so popular?



This painter is amazing!  His work is stunning online and from what I can learn of his history, he's possibly one of the top ten living painters of Imaginative Realism art.  

This specific painting, however, it doesn't match the level of skill he normally shows.  There's something weird about the way the light bounces around.  The musical instrument may not make sense, but it's imaginative.  Hands on the other, um, hand, these need to make sense in realism.  

From Jakedontdraw's video above and what I can hear from people who know this painter in person, he's moved towards using AI references.   AND IT SHOWS!

I don't feel so bad about him using AI references if it makes a good painting.  But what really bugs me is that he has the skill to fix the AI mistakes and he doesn't do it.  He accepts what the AI gives him, this imperfect image, and declares it good enough to be worth painting.

Or maybe he doesn't have the skills to fix the AI mistakes?  Could someone as good at painting as he is actually lack the ability to see bad hand mutations?  I don't want to believe this.

It also speaks to the importance of having a good reference we can trust.  Painting from life may not always be an option.  But surely our source material matters.  

---

Are current day artists only as good as their source material?

---

THE DISILLUSIONMENT AND A GURU

I'm about to talk about an online teacher I greatly admire.  I feel that anyone serious about learning to paint would benifit from taking time learning his method - even if you never want to paint that way again.  His ability to match colour exactly and to put the right colour in the right place is unusual in our current times.  If one's ever said "mixing colour is hard" or "why doesn't this look like how I want it to?" then take a month off and learn this method.

For many people, this method is the goal they aim towards.  For me, I'm treating it as a skill to add to my toolbox.  A stepping stone because it's a bit too much like being a human camera with a brush stroke filter.  


This guru, who I greatly admire, stumbled the other day.  Stumbled hard.  It doesn't mean his teaching isn't less valuable.  In fact, I think his stumble is the perfect example of why his teaching is so important.  



The video begins with the idea that we should always paint with the goal of making something great.  Try our best every time so we can learn faster.  And above all else, make sure the reference photo you are working from is the best of all possible photos.

He goes on to describe the painting he's working on now as quite possibly the best painting he ever painted.  

The number one thing to painting your masterpiece, he says, is finding your source material.  It's important to put everything you have into getting this absolutely right, we are told, or else the painting is doomed (I'm paraphrasing this last bit).

He goes on for quite some time about how important the source/reference photo for the painting is.  

Looking at history, this source material can be from life, sketches, studies, all sorts of things.  For DrawMixPaint's method, he uses from life or from photos and teaches us how to copy this and make it into a painting.

Are you board yet?  Here I am yammering on about reference sources as if it's important or something.  Well, his online course spends the first five chapters (there are 10 total) on some variation of this theme about getting the source/reference material perfect.  

And the more I learn about painting, the more I believe him.

Here is his source photo for his Masterpiece and best painting ever!



It took over one thousand tries on AI, he tells us in the video.  He even did some editing to get it perfect.

Is it perfect?

He's been painting long enough and is way better than me, so his word has weight.  But perfect?  

The city in the background doesn't match the rules of perspective that I learned - but maybe I haven't learned enough yet.  Some of the buildings repeat or are cut off weirdly.  The windows themselves defy logic.  The light bounces in the room weirdly.  That's just what I spotted.  If you want to learn what else is wrong with it, the comment section in the video goes way deeper.   Mostly, for me, it's the sense of wrongness that I get when looking at this.  A feeling like I ate too much cheese.  All these big and little errors really bug me.  

Errors that this guy should be good enough to spot and fix.  But he's happy with this as a source?  

Has AI art trained us to become happy with mediocre reference images?  

"If you are going to choose mediocre artwork, don't even paint."  (from the video)

This video changed the entire way I looked at this guy and his teaching.  If he accepts this source as a perfect masterpiece reference - then I worry that I can't trust the other things he teachers.  

Maybe it's a case of do what he says, not what he does.

---

At this stage, I'm feeling downhearted.  These amazing living artists that I admire for not just their painting skills but also their ability to put together ideas and find or make sources to get the lighting right and all that.  To make the art believable.

The moment they start using AI, it exposes how flawed these artists are.  They look like nothing but photocopiers.  Sure their painting skills are amazing.  But I've got a camera that can do that.  Art to me is more than copying what we see.  That's like stopping after the tutorial of a game and saying "yep, I'm at level one.  I've totally mastered this."

I know there is more to art than that, because I see it in the old works before cameras and AI.  So where does this fit in todays world?  Are we just overpaid cameras?  


Does that mean that the furthest I can ever go with my art is their level?  'Cause that's not good enough for me.  I want better.  I expect better from myself.  

---

AI AND ART DONE RIGHT

There is hope.  Small.  But I've been very, very lucky to find it in a local painter.  There is a way to use AI to benefit  both artist and viewer. But it takes a lot of work.

This is how it went:

An idea happened.

Thumbnail sketches were made with pencil and paper.  Maybe 20, maybe 100.  These were drawn at about 1 inch at the largest side.

From there, one or two were chosen to experiment with larger thumbnails.  Again pencils and paper.  At this stage, slightly playing with AI to get some more ideas on things to try.  Not sticking with any idea yet.  Still doing the larger thumbnails.  Maybe a dozen.

A composition is chosen.  A small sketch (again, paper and pencil) is fed into AI and several options are generated.  

Two or three of them are chosen and edited in something like photoshop to take the best from each.  The lighting corrected.  etc.

Feed the results back into AI.

This goes on for a few days.  AI helps, human corrects and improves.  

Eventually, they arrive at a place where they can get a final sketch and do it with a pencil and paper to get the values and light logic right.  Using the edited AI images for colour and things that one would normally use a photograph and model for.  

And even still, there was need for a photo shoot with a model to get the hands and some of the face correct.  

The results are not finished.  But what I've been blessed to see - they are stunning!  There is none of that uncanny valley kind of feeling like I got with the above examples.  It feels right and it feels real.  

...

The point is, this artist knows enough and CARES ENOUGH to put the effort in to getting it right.  

They have lived a life among humans, light, and hands, and paid attention.  For them, the AI image generator is just a tool to help them along the way.  To save money or time on early photo shoots.  But they don't allow the AI system to tell them what is good and what is "real".  They trust themselves to correct the mistakes AI made.

That said, from watching the way they struggled with the AI software, and from the ignorance of someone new to art, I can't help but wonder if a photo shoot near the start would have saved them so much time!  It feels like the current errors of AI slowed them down and held them back.  But once AI improves, this method of caring and putting in the work to get the source correct will give artists like this one, an edge in the art world.  An edge lost by those who are technically good painters but trust AI too much to do the work for them.  

---

This gives me hope that we can move beyond just copying a reference image.  

But also, it worries me how much work it takes.  And how much most of my living art heroes don't care enough to put in that work.  

I have a feeling that AI's mark on the art world will be to divide the photo copiers from the artists.  And that doesn't seem so bad to me.

I just need to keep improving my own art skills and move past the photocopier stage as quickly as possible.  I wonder how I do that.  

---

An interesting question sprung to life during a conversation earlier today.

Can a person own the copyright to a painting they made from an ai generated image?

For the most part, no.

As it sits in spring 2025, focusing on usa, canada and international copyright law, and with the disclaimer never trust a random person on the internet when it comes to legal advise...

When a human creates something unique like writing, photography, art, it is automatically copyrighted (although protecting your copyright is a different kettle of sharks).  Original is the key here.  And human (not person, that would include several large corporations that have personhood).

If someone creates something that isn't original, things get complicated.  Just by singing a song someone else wrote, we don't automatically have the rights to that song.  Master copies are common in painting, but copying van goghs almonds doesn't give me the rights to that composition.

There has to be a significant transformation to the original work for the new artist to have full rights to their work.

This is a massive oversimplification in an attempt to get past the boring bits.

Significant transformation would look something like



into



And even still, Dali still gives credit to where he got the idea (Millet).

So copying an image for a painting is considered fair game.  Ish.  But it doesn’t mean you own the full copyright to the new image.  Singing a song may give limited rights to the specific recordings,  but someone else owns the rights to the words and music.

But what about copying an ai image?  

Well, at this snapshot in history,  AI generated content is not protected by copyright.

Like a bear taking a selfie, these non human images are public domain.  Copying something in the public domain doesn't give them the rights to that composition. Anyone using these ai generated source images without significant transformation to paint their paintings is, well, cooking up a recipe for trouble.  

Here is a small example of what I mean by repairing ai errors.