"A.I." (Machine Learning Algorithms) To Generate Art

[Deleted User]

@NeuM said:

@monz0id said:

@NeuM said:

@knewspeak said:

@NeuM said:

@knewspeak said:

@NeuM said:

@knewspeak said:

@NeuM said:

@monz0id said:

@NeuM said:

@monz0id said:

@Tamir_Raz_Mataz said:
What do u think the ai can’t do and a only a real illustrator can?

Make precise alterations based on commissioning editor feedback. And/or adhere to a strict brief.

If publishers just want weird random stuff then they can bypass illustrators and get someone in the office to do it. Usually though their requirements are very specific.

@NeuM said:

@Tamir_Raz_Mataz said:
What do u think the ai can’t do and a only a real illustrator can?

These cannot do hands well yet, so...

Faces aren’t great either - there’s more often than not strange bits stuck on, and distortions that don’t adhere to generally acknowledged concepts of anatomy. Then there’s the extra legs that seem to crop up in group renders…

I find the deliberately weird stuff entertaining, but most of the illustration attempts I’ve seen (and Instagram is chock full of the stuff at the moment) have all the artistic integrity of an Athena poster on acid.

The good stuff is generally a result of using the AI as a starting point, and using other digital tools to hoik it into place. So, human illustrators, then.

I fully expect those bugs will be worked out of these systems within a year, possibly less.

Maybe, not if it ‘learns’ by scraping images from its own output though, which are now properly taking over the internets.

Things might get even weirder….

As I said, I like the weird, incredibly detailed trippy stuff, and I can see it being used by fantasy illustrators as part of an existing toolkit, but in its current form it’s still a long way from replacing traditional illustration work.

To get a sense of how fast these systems are advancing, I recommend this YouTube channel.

https://www.youtube.com/user/keeroyz

The really big leaps have come in the last several years and that curve is accelerating, not decreasing.

And when advanced machine learning systems are married with robotics, suddenly most of the world will find themselves replaced with tireless workers who never stop working, except to be replaced with another better, faster, cheaper robot.

I saw what was coming and got out of my line of work and instead chose to invest in companies best suited to take advantage of these huge advances. If I was a younger person, I'd plow every extra dollar I had into companies leveraging machine learning/A.I. and retire early.

A couple of questions what will humanity be doing and where is all the energy to keep the multitude of AI machines and humanity going.

I have no illusions about all work suddenly being handed over to robots, but it’s now happening much faster than even I originally thought it would. And why is it happening? The massive demographic shift that is happening worldwide demands it. There are too many people leaving the workforce and too few people replacing them. Thus, the rise of the robotic workforce.

That may be so, but what of the implications, the two questions I raised, a massive redundant population of humanity and implications of resources to humanity and the machine’s.

At some point, all skilled work will be performed by robots and artificial intelligence. It's inevitable (unless there is something like the "Butlerian Jihad" of Frank Herbert's Dune universe coming around the corner). People will just have to figure out what is important when such a scenario comes about. Remember, there didn't used to be a YouTube or Instagram economy built on personalities before, but there is now. Markets change and people adapt.

YouTube is largely human’s producing content for other human’s using technology as a tool in the creative process, what happens eventually when AI produce nearly everything, the role of most of humanity becomes mute and economically mute too, therefore power and wealth will be consolidated with the AI owner’s, a small fraction of the world’s population. Yes I agree that would be a market change.

I'm not going to be pessimistic about these developments at this stage. People are clever and adaptable. And it's Elon Musk's opinion that people will choose to merge with A.I. rather than be replaced.

He also thinks Ukraine should give in to Russia, but he’s wrong about that too.

Robots won’t take over all work because we are at the beginning of a total, climatic breakdown, along with economic and energy crises, and so technology resources - from research, to materials - are going to have their work cut out dealing with that stuff instead.

Who’s going to rebuild Ukraine, is it Tesla robots? No, it isn’t. It’s people, isn’t it Elon.

This is not a political thread. Don’t try to make it one.

You introduced him into the thread to back-up your claim. I’m pointing out he doesn’t know it all. Nothing ‘political’ in that.

Oh, and don’t tell me what to do.

NeuM

@monz0id said:

@NeuM said:

@monz0id said:

@NeuM said:

@knewspeak said:

@NeuM said:

@knewspeak said:

@NeuM said:

@knewspeak said:

@NeuM said:

@monz0id said:

@NeuM said:

@monz0id said:

@Tamir_Raz_Mataz said:
What do u think the ai can’t do and a only a real illustrator can?

Make precise alterations based on commissioning editor feedback. And/or adhere to a strict brief.

If publishers just want weird random stuff then they can bypass illustrators and get someone in the office to do it. Usually though their requirements are very specific.

@NeuM said:

@Tamir_Raz_Mataz said:
What do u think the ai can’t do and a only a real illustrator can?

These cannot do hands well yet, so...

Faces aren’t great either - there’s more often than not strange bits stuck on, and distortions that don’t adhere to generally acknowledged concepts of anatomy. Then there’s the extra legs that seem to crop up in group renders…

I find the deliberately weird stuff entertaining, but most of the illustration attempts I’ve seen (and Instagram is chock full of the stuff at the moment) have all the artistic integrity of an Athena poster on acid.

The good stuff is generally a result of using the AI as a starting point, and using other digital tools to hoik it into place. So, human illustrators, then.

I fully expect those bugs will be worked out of these systems within a year, possibly less.

Maybe, not if it ‘learns’ by scraping images from its own output though, which are now properly taking over the internets.

Things might get even weirder….

As I said, I like the weird, incredibly detailed trippy stuff, and I can see it being used by fantasy illustrators as part of an existing toolkit, but in its current form it’s still a long way from replacing traditional illustration work.

To get a sense of how fast these systems are advancing, I recommend this YouTube channel.

https://www.youtube.com/user/keeroyz

The really big leaps have come in the last several years and that curve is accelerating, not decreasing.

And when advanced machine learning systems are married with robotics, suddenly most of the world will find themselves replaced with tireless workers who never stop working, except to be replaced with another better, faster, cheaper robot.

I saw what was coming and got out of my line of work and instead chose to invest in companies best suited to take advantage of these huge advances. If I was a younger person, I'd plow every extra dollar I had into companies leveraging machine learning/A.I. and retire early.

A couple of questions what will humanity be doing and where is all the energy to keep the multitude of AI machines and humanity going.

I have no illusions about all work suddenly being handed over to robots, but it’s now happening much faster than even I originally thought it would. And why is it happening? The massive demographic shift that is happening worldwide demands it. There are too many people leaving the workforce and too few people replacing them. Thus, the rise of the robotic workforce.

That may be so, but what of the implications, the two questions I raised, a massive redundant population of humanity and implications of resources to humanity and the machine’s.

At some point, all skilled work will be performed by robots and artificial intelligence. It's inevitable (unless there is something like the "Butlerian Jihad" of Frank Herbert's Dune universe coming around the corner). People will just have to figure out what is important when such a scenario comes about. Remember, there didn't used to be a YouTube or Instagram economy built on personalities before, but there is now. Markets change and people adapt.

YouTube is largely human’s producing content for other human’s using technology as a tool in the creative process, what happens eventually when AI produce nearly everything, the role of most of humanity becomes mute and economically mute too, therefore power and wealth will be consolidated with the AI owner’s, a small fraction of the world’s population. Yes I agree that would be a market change.

I'm not going to be pessimistic about these developments at this stage. People are clever and adaptable. And it's Elon Musk's opinion that people will choose to merge with A.I. rather than be replaced.

He also thinks Ukraine should give in to Russia, but he’s wrong about that too.

Robots won’t take over all work because we are at the beginning of a total, climatic breakdown, along with economic and energy crises, and so technology resources - from research, to materials - are going to have their work cut out dealing with that stuff instead.

Who’s going to rebuild Ukraine, is it Tesla robots? No, it isn’t. It’s people, isn’t it Elon.

This is not a political thread. Don’t try to make it one.

You introduced him into the thread to back-up your claim. I’m pointing out he doesn’t know it all. Nothing ‘political’ in that.

Oh, and don’t tell me what to do.

Stop it.

Michael

Yes. Stop it! If you want to start a political thread, take it to Recent Events.

[Deleted User]

@Michael said:
Yes. Stop it! If you want to start a political thread, take it to Recent Events.

Where was the politics?

MrStochastic

Here's an interesting recent video. ( I hope I'm not reposting something already here; I tried to check quickly).

The composer David Bruce gives a good overview I think:

AudioGus

auxmux

@AudioGus said:

Lol

Krupa

https://www.theguardian.com/film/2022/oct/16/dune-subreddit-group-bans-ai-generated-art-for-being-low-effort?utm_term=634cd3433e76017551c418c601afcd96&utm_campaign=GuardianTodayUK&utm_source=esp&utm_medium=Email&CMP=GTUK_email

Early shots fired in the Butlerian Jihad

echoopera

Hrmm maybe we should ban all Generative and Conditional based sequences called Music by their Creators on the AB forum to follow suit. 🤨😝

Time to ban the use of all algorithm based drum sequencers.

Who’s with me?

Yes. I’m being sarcastic.

It is a difficult debate for sure but as we all know who leverage auto generative tools…what they output serve as a good foundation to build upon. If we just accept what it gives us there is no true emotion in the piece. However, if we use our God given talent of discernment to build on top of what the machine created, then we witness the spark of life and spirit in the creation…and if we’re lucky, that creation will connect with another person and move them as well.

I know I’ve shown this before but it’s good to remind us not to settle for what the machine delivers.

MidJourney output:

A couple hours of post processing to make it my own:

It’s a tool at the end of the day and we’ll use it based on our motivation and intention.

We get out of it what we put in to it.

Krupa

I think I’ll be checking this out today

https://80.lv/articles/a-new-free-add-on-integrates-stable-diffusion-into-blender/

AudioGus

The End of Art: An Argument Against Image AIs

[Deleted User]

@AudioGus said:
The End of Art: An Argument Against Image AIs

Haven’t watched it all, but good that.

I think also one of the attraction’s for me about a particular piece of art, music, writing etc., is that it’s usually going to include an interest, appreciation, respect etc. of the person (or group) that created it.

So for example my favourite artists include Robert Lenkiewicz and Brian Froud, and the art they’ve created is an extension of their own personalities and surroundings, which are funnelled back into their art. Froud literally paints goblins that roll down the Dartmoor hills and pose on his desk. Right? Of course he does. Where do I even begin with Robert…

Musically bands like Faust and Nurse With Wound fascinate me - not just for the music itself (some of it I don’t even like), but the way they create it, their histories, and personalities. The interviews, videos of their mad studios, the cover art…the whole package and ethos.

Writing - Alan Moore is a good example. Again a fascinating, inspiring character, whose personality seeps into his work. Even though I’ve never been there, I reckon I can now easily find my way around Northampton city centre. He basically inhabits his own books.

However great the AI output is, some bloke tapping a few keywords into his laptop and sipping a campachoochoo is never going to inspire and hold my attention like the guys above.

Carnbot

Yeah, it's a good explanation of the bad practice of companies like Stability who have no interest in the well being of the artists they are leeching from in a bid to become the next Facebook style monster off the back of it...

Krupa

Playing with stable diffusion using a local setup on my Mac today I’ve found ways to minimise the randomness; feeding it simple block versions of what I want and getting it to just ‘paint’ the surface, ace, what a great toolbox… I can’t wait to get animated versions going, just such a liberation in film making, and hopefully I’ll be able to train it on my own supply it’ll both eliminate the ethical troubles I have with it, and satisfy my own need to feel somewhat self sufficient in my creative output. I’ma bit of a maddie like that, hence the homemade valve synth stuff and general mad scientist / art school vibes to everything I make. Dog knows how I’ve made a living for the last twenty years or more 😂

Carnbot

I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Krupa

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

I was looking at the GitHub for that earlier, I’ll probably dive into compiling it at some point, just got to get over the dependency hell fear first 😅

Tamir_Raz_Mataz

Anyone tried generating AI music with https://www.aiva.ai/ ?

NeuM

@Tamir_Raz_Mataz said:
Anyone tried generating AI music with https://www.aiva.ai/ ?

Please start another thread dedicated to "A.I."-generated music. This is an art-centric thread.

AudioGus

@monz0id said:

@AudioGus said:
The End of Art: An Argument Against Image AIs

Haven’t watched it all, but good that.

I think also one of the attraction’s for me about a particular piece of art, music, writing etc., is that it’s usually going to include an interest, appreciation, respect etc. of the person (or group) that created it.

So for example my favourite artists include Robert Lenkiewicz and Brian Froud, and the art they’ve created is an extension of their own personalities and surroundings, which are funnelled back into their art. Froud literally paints goblins that roll down the Dartmoor hills and pose on his desk. Right? Of course he does. Where do I even begin with Robert…

Musically bands like Faust and Nurse With Wound fascinate me - not just for the music itself (some of it I don’t even like), but the way they create it, their histories, and personalities. The interviews, videos of their mad studios, the cover art…the whole package and ethos.

Writing - Alan Moore is a good example. Again a fascinating, inspiring character, whose personality seeps into his work. Even though I’ve never been there, I reckon I can now easily find my way around Northampton city centre. He basically inhabits his own books.

However great the AI output is, some bloke tapping a few keywords into his laptop and sipping a campachoochoo is never going to inspire and hold my attention like the guys above.

I never actually fell too much in love with a 'digital artist' as far as visuals (music yes). They have all been painters for sure, although Dave McKean did do digital work at one time that transfixed me but once everyone jumped on it things have faded on that front for sure. AI will always fall prey to this and even quicker. I have fallen in love with the digital medium though, partly because money/food is important.

I think though that people will be more apt to simply fall in love with their own prompts/generations now but yah there is a hollowness to it if you do not train the system with your own images and ideas.

AudioGus

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Krupa

@AudioGus said:

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Have you done this with mac, or would I be better using my windows box for this in your opinion?

[Deleted User]

@AudioGus said:

@monz0id said:

@AudioGus said:
The End of Art: An Argument Against Image AIs

Haven’t watched it all, but good that.

I think also one of the attraction’s for me about a particular piece of art, music, writing etc., is that it’s usually going to include an interest, appreciation, respect etc. of the person (or group) that created it.

So for example my favourite artists include Robert Lenkiewicz and Brian Froud, and the art they’ve created is an extension of their own personalities and surroundings, which are funnelled back into their art. Froud literally paints goblins that roll down the Dartmoor hills and pose on his desk. Right? Of course he does. Where do I even begin with Robert…

Musically bands like Faust and Nurse With Wound fascinate me - not just for the music itself (some of it I don’t even like), but the way they create it, their histories, and personalities. The interviews, videos of their mad studios, the cover art…the whole package and ethos.

Writing - Alan Moore is a good example. Again a fascinating, inspiring character, whose personality seeps into his work. Even though I’ve never been there, I reckon I can now easily find my way around Northampton city centre. He basically inhabits his own books.

However great the AI output is, some bloke tapping a few keywords into his laptop and sipping a campachoochoo is never going to inspire and hold my attention like the guys above.

I never actually fell too much in love with a 'digital artist' as far as visuals (music yes). They have all been painters for sure, although Dave McKean did do digital work at one time that transfixed me but once everyone jumped on it things have faded on that front for sure. AI will always fall prey to this and even quicker. I have fallen in love with the digital medium though, partly because money/food is important.

Oh definitely. Digital artworking has helped me in part to pay the bills for nearly 30 years, though the work is generally pretty dull. I even taught Photoshop/Adobe techniques to design professionals in a few jobs over years. Not many commissions for dystopian fantasy art worlds, unfortunately, so the weird stuff is kept for my own album covers.

I still see interesting AI stuff on the Instagram, though as it’s a relatively simple process, they appear randomly - scattered across thousands of accounts that were previously posting shots of their dinners, now pumping out crazy images of faces with bits missing from them.

I think it’s great that people have access to all this, and when there’s a free, easy app way of doing it I’ll probably join in.

The concerns I’ve posted earlier are similar to those mentioned in the video you posted - the effect on professionals in the existing market for fantasy/sci-fi/game artworking, and where this will all lead to. I don’t want to see those interesting, creative people that inspire me on a daily basis, lose their incomes and have to give up what they do, only to be replaced by millions of anonymous code tappers. It’s the whole creative ‘package’ that inspires me, not just the finished piece.

Tamir_Raz_Mataz

@NeuM said:

@Tamir_Raz_Mataz said:
Anyone tried generating AI music with https://www.aiva.ai/ ?

Please start another thread dedicated to "A.I."-generated music. This is an art-centric thread.

U mean visual art as music is art too

I will start a thread one day unless someone predates me 🦥

Krupa

@monz0id said:

@AudioGus said:

@monz0id said:

@AudioGus said:
The End of Art: An Argument Against Image AIs

Haven’t watched it all, but good that.

I think also one of the attraction’s for me about a particular piece of art, music, writing etc., is that it’s usually going to include an interest, appreciation, respect etc. of the person (or group) that created it.

So for example my favourite artists include Robert Lenkiewicz and Brian Froud, and the art they’ve created is an extension of their own personalities and surroundings, which are funnelled back into their art. Froud literally paints goblins that roll down the Dartmoor hills and pose on his desk. Right? Of course he does. Where do I even begin with Robert…

Musically bands like Faust and Nurse With Wound fascinate me - not just for the music itself (some of it I don’t even like), but the way they create it, their histories, and personalities. The interviews, videos of their mad studios, the cover art…the whole package and ethos.

Writing - Alan Moore is a good example. Again a fascinating, inspiring character, whose personality seeps into his work. Even though I’ve never been there, I reckon I can now easily find my way around Northampton city centre. He basically inhabits his own books.

However great the AI output is, some bloke tapping a few keywords into his laptop and sipping a campachoochoo is never going to inspire and hold my attention like the guys above.

I never actually fell too much in love with a 'digital artist' as far as visuals (music yes). They have all been painters for sure, although Dave McKean did do digital work at one time that transfixed me but once everyone jumped on it things have faded on that front for sure. AI will always fall prey to this and even quicker. I have fallen in love with the digital medium though, partly because money/food is important.

Oh definitely. Digital artworking has helped me in part to pay the bills for nearly 30 years, though the work is generally pretty dull. I even taught Photoshop/Adobe techniques to design professionals in a few jobs over years. Not many commissions for dystopian fantasy art worlds, unfortunately, so the weird stuff is kept for my own album covers.

I still see interesting AI stuff on the Instagram, though as it’s a relatively simple process, they appear randomly - scattered across thousands of accounts that were previously posting shots of their dinners, now pumping out crazy images of faces with bits missing from them.

I think it’s great that people have access to all this, and when there’s a free, easy app way of doing it I’ll probably join in.

The concerns I’ve posted earlier are similar to those mentioned in the video you posted - the effect on professionals in the existing market for fantasy/sci-fi/game artworking, and where this will all lead to. I don’t want to see those interesting, creative people that inspire me on a daily basis, lose their incomes and have to give up what they do, only to be replaced by millions of anonymous code tappers. It’s the whole creative ‘package’ that inspires me, not just the finished piece.

I’m with you on that, though I hope/believe that people will quickly tire of the auto generated stuff, and it will become a tool that’s used by artists in new ways. I’m really jonesing to use it to make low budget animation possible in the way that I want to work…

[Deleted User]

@Krupa said:

@monz0id said:

@AudioGus said:

@monz0id said:

@AudioGus said:
The End of Art: An Argument Against Image AIs

Haven’t watched it all, but good that.

I think also one of the attraction’s for me about a particular piece of art, music, writing etc., is that it’s usually going to include an interest, appreciation, respect etc. of the person (or group) that created it.

So for example my favourite artists include Robert Lenkiewicz and Brian Froud, and the art they’ve created is an extension of their own personalities and surroundings, which are funnelled back into their art. Froud literally paints goblins that roll down the Dartmoor hills and pose on his desk. Right? Of course he does. Where do I even begin with Robert…

Musically bands like Faust and Nurse With Wound fascinate me - not just for the music itself (some of it I don’t even like), but the way they create it, their histories, and personalities. The interviews, videos of their mad studios, the cover art…the whole package and ethos.

Writing - Alan Moore is a good example. Again a fascinating, inspiring character, whose personality seeps into his work. Even though I’ve never been there, I reckon I can now easily find my way around Northampton city centre. He basically inhabits his own books.

However great the AI output is, some bloke tapping a few keywords into his laptop and sipping a campachoochoo is never going to inspire and hold my attention like the guys above.

I never actually fell too much in love with a 'digital artist' as far as visuals (music yes). They have all been painters for sure, although Dave McKean did do digital work at one time that transfixed me but once everyone jumped on it things have faded on that front for sure. AI will always fall prey to this and even quicker. I have fallen in love with the digital medium though, partly because money/food is important.

Oh definitely. Digital artworking has helped me in part to pay the bills for nearly 30 years, though the work is generally pretty dull. I even taught Photoshop/Adobe techniques to design professionals in a few jobs over years. Not many commissions for dystopian fantasy art worlds, unfortunately, so the weird stuff is kept for my own album covers.

I still see interesting AI stuff on the Instagram, though as it’s a relatively simple process, they appear randomly - scattered across thousands of accounts that were previously posting shots of their dinners, now pumping out crazy images of faces with bits missing from them.

I think it’s great that people have access to all this, and when there’s a free, easy app way of doing it I’ll probably join in.

The concerns I’ve posted earlier are similar to those mentioned in the video you posted - the effect on professionals in the existing market for fantasy/sci-fi/game artworking, and where this will all lead to. I don’t want to see those interesting, creative people that inspire me on a daily basis, lose their incomes and have to give up what they do, only to be replaced by millions of anonymous code tappers. It’s the whole creative ‘package’ that inspires me, not just the finished piece.

I’m with you on that, though I hope/believe that people will quickly tire of the auto generated stuff, and it will become a tool that’s used by artists in new ways. I’m really jonesing to use it to make low budget animation possible in the way that I want to work…

I think so too. I’ve been able to create weird and wonderful images via phone apps for years, the AI stuff is just taking this one step further. Or maybe ten. A lot of fun to be had.

But it’s rarely quite as good (or commercial) as what the real pro’s can do with their Wacom tablets and Cinteq’s, and I’m guessing a lot of these will be incorporating AI elements into their own artwork, and taking things to a higher level than the current batch of hobbyists (not knocking hobbyists, that’s a good description of my place in the music world) are currently pumping out. Steve McDonald (not that one) is a good example of an artist taking the genre to new levels of creativity.

Carnbot

@AudioGus said:

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Yeah, this is where the interesting part of the technology lies All those people simply trying to copy others work is mere distraction which people will tire of quickly. It's the artists with the skills combined with this tech which will create never before seen work which is interesting.

Also just training from the finished results will never create the same type or quality of work as training from the same source material. I think we will see artists keep their source material off line rather than sharing everything, which is probably a good thing as there is too much stuff online!

I've been using it to train some of my photographs to create some source material for animation for my project I am showing next month in Czech Republic. I think AI animation is not good enough yet compared with normal animation techniques, but animating some of the material with real time compositing/generative tools I find is a great workflow because there is already so much to do in creating the soundtrack and editing etc, that it is a helpful AI assistant.

waka_x

https://tinyurl.com/aistudies

AudioGus

@Krupa said:

@AudioGus said:

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Have you done this with mac, or would I be better using my windows box for this in your opinion?

For the training I use a google collab python notebook (subscribed to the 15$ tier) and then for rendering I use Automatic1111 web ui running locally.

AudioGus

@Carnbot said:

@AudioGus said:

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Yeah, this is where the interesting part of the technology lies All those people simply trying to copy others work is mere distraction which people will tire of quickly. It's the artists with the skills combined with this tech which will create never before seen work which is interesting.

Also just training from the finished results will never create the same type or quality of work as training from the same source material. I think we will see artists keep their source material off line rather than sharing everything, which is probably a good thing as there is too much stuff online!

Thing is, if done well, in a lot of cases the end results can look 95% like the training images in terms of style and if people feel compelled to only release derivatives that are not the pure source, that is kind of odd as the pure source is the most powerful emotional connection, assuming the artist made it themselves. But then even if people can only make pale knockoffs that is enough to muddy the waters of who pioneered it etc. if they don't make a strong impression out of the gate now.

I think a lot of people will be holding back on what they share and that doing the slow trickle of instratwit etc will be super unappealing. I think people are likely cooking up some amazing things using their own art and dreambooth, training their own images and wondering how to go about releasing it so that they do not give away their golden goose to the world. Releasing art may become more like releasing a movie back in the day where things are under heavy lock and key, no previews, no spoilers, and then a big spectacle / statement is made upon release. This slow drip of giving away one's incrimental developments on social media is probably now not even going to be an option for a lot of people. I definitely feel way more compelled to only show things in grand finished products now.

I've been using it to train some of my photographs to create some source material for animation for my project I am showing next month in Czech Republic. I think AI animation is not good enough yet compared with normal animation techniques, but animating some of the material with real time compositing/generative tools I find is a great workflow because there is already so much to do in creating the soundtrack and editing etc, that it is a helpful AI assistant.

Yah the AI animation now is fun but gimmicky and all a bit samey to really invest too much time into. A neat demonstration of tech for sure but have yet to see any personal statements that I found compelling. It does inspire some cool music though. For me bashing images and characters out into 3d works great and now there is more time and bandwidth for the normal animation techniques before losing creative steam. I just love how things feel more like music production now.

Krupa

@AudioGus said:

@Krupa said:

@AudioGus said:

@Carnbot said:
I recommend checking out Stable Diffusion webgui, it's great. I'm using it regularly to train my own custom models

Being able to train images with Dreambooth has been one of the most mind blowing experiences of my life. Making things with these tools via training that are impossible for anyone else to make though is such a mindscrew because the moment you share it online people can nab it and train with it and then your golden goose is gone. heh/sob, ahh well first world new world problems I suppose.

Have you done this with mac, or would I be better using my windows box for this in your opinion?

For the training I use a google collab python notebook (subscribed to the 15$ tier) and then for rendering I use Automatic1111 web ui running locally.

Ah cool, I think I saw a tutorial for that earlier. I’ll have another look at it as installing locally might be an absolute nightmare, thanks 🙏