Crazy right? I was surprised by a lot of these answers in a good way. While this is a pretty positive batch of perspectives, let me be clear, I wish AI image generators could be uninvented. Enormous ethical outrage aside, I just hate what a bummer this all is and how it can sap all of the fun and excitement out of our favorite thing if we let it. Personally, (whether it’s logical or not) I’m just kind of over it - it could be false confidence, hopium or just worry exhaustion - but I say let’s get back to work and make things that are cool. I wanted to make this as a quick check in video and I wanted to show up with some actual industry perspectives and not just my opinions and future gazing. I hope you found this helpful and encouraging. All art/painting/tutorial videos from me in the months ahead! Now go paint something awesome :)
I don't think AI image generation should be uninvented in general. It can be useful to do a lot of things like create architectural designs or medical charts etc. The problem is that corporations don't want to just use it as a tool but rather as a replacement for labour. In art creation I, personally, don't really think that AI is needed, so I wouldn't be sad if it stopped in that regard.
This was really encouraging to watch, thank you for taking the time to interview these art directors and give us a taste of what's going on out there in the industry.
Love to hear about the state of the industry in this series, looking forward to hearing about more topics as well. I can't stomach the idea of using AI for idea and concept generation for a multitude of reasons but it sounds like the industry agrees and is favouring using it to communicate ideas to the client faster. The ethics of the entire business model make it a no-go for me but it does sound like the industry is settling into a fairly artist-first relationship with AI which I can definitely support. Ultimately I think artists positioning themselves as an 'AI artist' are going to enter into the biggest race to the bottom in the industries history whereas people who develop their traditional skillset will always be more valuable.
Really great points about market forces and general attitudes. I've been really encouraged by that - those who make games and movies are often artists themselves (at least at heart) and seem to genuinely care about that human connection. Thanks for your perspectives
Thanks for this one Hardy, appreciate you taking the time to craft a thoughtful and well-researched answer to this. These days I pretty much stick my head in the sand and just keep drawing but in doing so I also miss a lot about what isn’t going wrong. I can confirm that I have seen it creep into the mood boarding side of things at my job, but the attitude is still largely dismissive, and certainly none of it is directly used in production. And having learned more about art direction over the past couple of years, I can say with confidence that it requires a clear vision of mood, concept, and world-building that AI cannot, and perhaps never will be able you provide.
Hey Eben! Man, I'm really glad to hear that from you. I'm with you on the head in the sand plan - for a while there I was doom scrolling on this topic way too much and I decided to step back from that which has been a great move BUT I feel like we need to keep one eye on it as you said. I hope you're doing well and thanks for checking this out.
I'm starting watching. Been waiting whole week for this - seriously, can't wait to see what's inside. Its hard to find an art director take on this - it's great you got round to it and interviewed them. That's rare material in our community (or I just don't know where to look) And so weird it has only 2k views so far, it's a HOT DAMN TOPIC
I'm glad this was helpful! Yeah, these talking head videos are less enjoyable to watch in my opinion so I didn't expect this one to get a huge view count. Back to the fun stuff for my next video :)
I was hoping that AI would be useful for creating new pipeline tools. All I want is a "Make Animatable Topology and UV's" button and that's it! I love everything else about the process and live to do it myself. Outside of that, AI is a hard NO for me.
As someone who went to art school in the 90s and watched the classic illustration profession crater in the face of the digital revolution (so many of the big illustrators of the time had to move to gallery work), I think AI will be an even bigger threat to the human touch in art. Once they work out the licensing and mitigate the threat of IP litigation, it will be off to the races for large AAA/creative companies. I think it's tragically unavoidable.
this is actually dangerous - a young person might watch this video and decide to purpose their career in this area when actually it's has no future, we all know that. most of the human work will be outsourced to cheap locations
I'm really sorry you feel that way. For what it's worth, my own son is an artist and wants to have an art career when he grows up - I fully encourage that and I believe he will achieve that goal. The world he grows into will be different from today but I believe there will be a value and a need for artists and craft even in the most dystopian of AI predictions. Hoping for the best for artists and the many other intellectual careers that this will disrupt.
Nice video! It's good to see how the use of AI in all stages of production is almost seen as a disease by most directors. There's an ethical, moral and philosophical problem which AI will not get rid of anytime soon. I shared this on X(but could't tag you).
As a new-ish Freelance illustrator (working since 2021) I haven't noticed a significant dropoff in work. That said, rates have been the same, and money doesn't go nearly as far these days.
I really appreciate you taking the time to make this video, Hardy. AI is a controversial topic these days, and I am very glad to have you shed light on a realistic, professional outlook of AI in the industry and whether it is even realistic to pursue Art as a career. While this video is extremely reassuring and makes me feel more optimistic about the future of artists still being at the forefront of the industry, it still leads me to question the vague aspect of some of the art directors' responses using AI as a tool for mood-boarding and iteration purposes. Do you see AI in that aspect still replacing artists as that work would have likely been done by a concept artist in the past, or is that something that we will have to worry less about as, as you stated, those companies still use artists for the final production result? Thank you again for the video, it is very motivating to see as a pursuing artist.
Great question and that one caught my eye too but that is an example of an area where work that might have gone to a hired concept artist was done by AI. However, when budgets are very limited mood boards and even pitch decks often have art grabbed from the internet just to get very early ideas of what they want to build. In those cases, it doesn't really seem like a job was lost, just AI stuff replacing google images. I really can't say for sure how disruptive this will end up being but I'm enormously encouraged by my own experiences and the artists around me in these first two years and I mainly wanted to share that. There is plenty to be concerned about with AI (honestly, I think all desk jobs will eventually be having their existential crisis like we have been recently) but so far so good - still standing!
Hey Hardy, we've used AI in some of the work I've been a part of. It never makes it to the final product and almost serves as placeholder or to quickly iterate on compositions. Like the other AD mentioned, either mood board generation or helpful composition reference for me to create something. For example, quickly putting together a series of compositions in a fisheye or worm's eye view, it was helpful to have some AI images handy for reference. I'm sort of at the mercy of my own AD and CD to do their bidding, but I've pushed back at times and said "I'm going to start with basic black and white sketches by hand" to make sure 1. I don't get rusty and 2. Keep the artist hand present in the work I help with.
I want to point out that a few months ago Treyarch posted a job listing on ArtStation that specifically called for artist to fix AI generated illustrations and assets.
Perhaps the more subtle threat of AI is not the replacement of the Artist as 'technician' in terms of the actual production of the images but the replacement of artistic judgement throughout the entire process. Who, after all, are the real advocates for quality in almost all creative pipelines? That would be the Artists themselves. It's those employed to create assetts that usually push the envelopes of quality and originality, sometimes to the frustration of those more focused on the bottom line. But what happens when those creatives are replaced with AI? It's not just their technical expertise that is lost, it's also their broader aesthetic influence on the entire project. What makes AI imagery so seductive to the non creative is that it seems to offer a way to more quickly and cheaply achieve the same results as using human creatives- you just type in a prompt and like magic a fully realised image appears in seconds! Now there may in fact be all sorts of problems with that image, in terms of it's relationship to other assets, or how it may or may not contribute to the story being told by the project as a whole- but who is now left to point these things out? In dispensing with the artist as technician you may have also thrown away something else, something that you may not have even realised you needed- which is that broader overview of what your images were intended to express. If Art is ultimately about communication and we replace artists with machines, who then, is really in charge of what is communicated?
There’s two sides to this answer. 1. AI usage isn’t regulated (yet) so a lot of companies still feel they’re in shaky waters when it comes to the usage of others IP’s or Content cause if the law changes they could get sued and lose millions. The other one is, for now, AI is mostly used on the pitch or moodboard phase of a project but the reality is voice AI is now tumbling increasingly on videogames and other media (ence the backlash by VAs). I feel in the future, just as any technology, it’s going to be (hopefully) ethically and morally incorporated in the workflow of a project but I don’t think it’s going to kill artists just by existing. We have stop motion pictures when 3D is here, we have artisans who make stuff with their hands when we can buy pretty much anything pre-made at IKEA. The human component is and always will be the main ingredient to any creation and we as humans will always feel empathy for any human artist more than a factory, a machine or any software. Great video, thanks for sharing!
As a storyboard artist that work mostly for animation, I experiment with AI image gens but so far I have not incorporated them in my workflow. The output they make is not what my job requires.
Hi, I have an unusual question, I was wondering what do you use when it comes to payment system when ppl hire you? I know it's maybe a bit personal or sth but Paypal seemes not to good, it takes a portion of your money
Also, I think it destroys collaboration. How is a traditional artist going to bounce ideas off of someone if they can't even conceive of it? Imagine if everything had to grind to a halt for someone to generate some ideas because they're too insecure of their own. Exercise the meat in your head, you need that thing!!
I don't even see the point in using AI for ideas when it can only give a worse, messy version of other artist's work... why would you not just go look yourself? And be inspired for way more, instead of taking a lazier, worse route? I would be very embarrassed if I worked my ass off developing my own skills, to have my input shrugged off in favour of whatever the AI threw together. And of course, spend all my time fixing what it messed up.
Right on. The tools seem really clumsy currently too (which is likely to change). It seems to give you loads of weird stuff that is ok but not really what you need
Yes, the Concept Art Association is leading the charge there, and they have a Go Fund Me page that is going towards legal action and supporting artists
Eben beat me to it. They have great resources and advocacy options. It is led by Karla Ortiz who is a legit hero for our community and you can donate to support their efforts. They are hiring lobbyists and helping to drive legislation to govern the ethical use of image generators. I've heard Karla and Steven Zapata refer to this situation as a boulder rolling down a hill - we can't stop it, but we can nudge it in a less harmful direction. This is definitely the cause to back if you're inclined.
Companies are FOR PROFIT. As soon as AI can do the same job with minimal input, they will begin hiring less and less artists. It's the same as manufacturing: if you want your company to be competitive nowadays, you do manufacturing in countries like China.
Thankfully I don’t need to use my art to make money. I don’t post anything either. I do feel bad for artists that do work in the field. Ai will only continue to grow. It’s not going anywhere.
I watched a demo from Craig Mullins at Lightbox and he was largely of the same mind. Not really because of AI but he was telling everyone not to use art to make money because it would turn something you love into something ugly. He was a really interesting speaker. You're totally right, it's not like this tech is going to go away - hoping for a future where it doesn't make human input atrophy completely
@ real art made from humans will always have a place. I don’t think it will be the mainstream stuff. There’s still people doing claymation and other old forms of art.
Everyone says no to AI because it's not good enough yet and they have an option to pay to real people to do the job. Every company outhere is chasing AI and how to replace positions with it to cut costs and flex to investos, in few years these AD might not be given option to pay to real people to do the job because their boss said so. I noticed that a lot of sci-fi 2d concept artists I follow on Artstation started to do more designs in 3d because because it's more valuable for production and AI can't do 3d.
That's fair - my understanding is pretty surface level and is limited to my particular corner of an industry and how it affects us. If you want to elaborate, I would welcome your perspectives
@@fowlerillus When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used. There are many tools so lets start by choosing one to talk about: Stable Diffusion. Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory. Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution. Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px: There are a total of 4,096 (64 x 64) pixels on a canvas. Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space. For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000). Total images = 16,777,215×4,096 = 68,719,476,736 We can store every single image into the vault with a unique address. What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics. Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters. To generalize the process: 1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form. 2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art. 3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make. So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process. Now when it comes to ethics, that is personal choice. For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one. Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things. Is this a decent start. This is a pretty huge simplification of one tiny subset of AI. But I think it's a fair starting point to bridge the gap and remind artists that mathematics and all the digital tools you used were created by people who were artistic in their own, different sense. Mathematics is as artistic as art is mathematical. 💯
@@fowlerillus When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used. There are many tools so lets start by choosing one to talk about: Stable Diffusion. Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory. Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution. Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px: There are a total of 4,096 (64 x 64) pixels on a canvas. Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space. For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000). Total images = 16,777,215×4,096 = 68,719,476,736 We can store every single image into the vault with a unique address. What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics. Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters. To Summarize the process: 1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form. 2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art. 3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make. So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process. Now when it comes to ethics, that is personal choice. For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one. Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things. Remember, Mathematics is as artistic as art is mathematical.
@@fowlerillus I've tried replying twice, but the reply isn't showing for me and I don't want to spam you. When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used. There are many tools so lets start by choosing one to talk about: Stable Diffusion. Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory. Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution. Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px: There are a total of 4,096 (64 x 64) pixels on a canvas. Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space. For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000). Total images = 16,777,215×4,096 = 68,719,476,736 We can store every single image into the vault with a unique address. What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics. Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters. To Summarize the process: 1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form. 2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art. 3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make. So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process. Now when it comes to ethics, that is personal choice. For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one. Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things. Mathematics is as artistic as art is mathematical. 💯
When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used. There are many tools so lets start by choosing one to talk about: Stable Diffusion. Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory. Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution. Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px: There are a total of 4,096 (64 x 64) pixels on a canvas. Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space. For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000). Total images = 16,777,215×4,096 = 68,719,476,736 We can store every single image into the vault with a unique address. What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics. Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters.
![Lethal Company Music Video 🎵 "Follow The Orders" [VERSION A]](http://i.ytimg.com/vi/QeooZB4Z7RA/mqdefault.jpg)
Crazy right? I was surprised by a lot of these answers in a good way. While this is a pretty positive batch of perspectives, let me be clear, I wish AI image generators could be uninvented. Enormous ethical outrage aside, I just hate what a bummer this all is and how it can sap all of the fun and excitement out of our favorite thing if we let it. Personally, (whether it’s logical or not) I’m just kind of over it - it could be false confidence, hopium or just worry exhaustion - but I say let’s get back to work and make things that are cool. I wanted to make this as a quick check in video and I wanted to show up with some actual industry perspectives and not just my opinions and future gazing. I hope you found this helpful and encouraging. All art/painting/tutorial videos from me in the months ahead! Now go paint something awesome :)
I don't think AI image generation should be uninvented in general. It can be useful to do a lot of things like create architectural designs or medical charts etc.
The problem is that corporations don't want to just use it as a tool but rather as a replacement for labour.
In art creation I, personally, don't really think that AI is needed, so I wouldn't be sad if it stopped in that regard.
This was really encouraging to watch, thank you for taking the time to interview these art directors and give us a taste of what's going on out there in the industry.
Thank you my friend! I'm so glad this was helpful :)
Great video glad to hear all this
Thanks Tyler. I appreciate it!
Love to hear about the state of the industry in this series, looking forward to hearing about more topics as well.
I can't stomach the idea of using AI for idea and concept generation for a multitude of reasons but it sounds like the industry agrees and is favouring using it to communicate ideas to the client faster.
The ethics of the entire business model make it a no-go for me but it does sound like the industry is settling into a fairly artist-first relationship with AI which I can definitely support.
Ultimately I think artists positioning themselves as an 'AI artist' are going to enter into the biggest race to the bottom in the industries history whereas people who develop their traditional skillset will always be more valuable.
Really great points about market forces and general attitudes. I've been really encouraged by that - those who make games and movies are often artists themselves (at least at heart) and seem to genuinely care about that human connection. Thanks for your perspectives
Thanks for this one Hardy, appreciate you taking the time to craft a thoughtful and well-researched answer to this. These days I pretty much stick my head in the sand and just keep drawing but in doing so I also miss a lot about what isn’t going wrong. I can confirm that I have seen it creep into the mood boarding side of things at my job, but the attitude is still largely dismissive, and certainly none of it is directly used in production. And having learned more about art direction over the past couple of years, I can say with confidence that it requires a clear vision of mood, concept, and world-building that AI cannot, and perhaps never will be able you provide.
Hey Eben! Man, I'm really glad to hear that from you. I'm with you on the head in the sand plan - for a while there I was doom scrolling on this topic way too much and I decided to step back from that which has been a great move BUT I feel like we need to keep one eye on it as you said. I hope you're doing well and thanks for checking this out.
I'm starting watching. Been waiting whole week for this - seriously, can't wait to see what's inside.
Its hard to find an art director take on this - it's great you got round to it and interviewed them. That's rare material in our community (or I just don't know where to look)
And so weird it has only 2k views so far, it's a HOT DAMN TOPIC
I'm glad this was helpful! Yeah, these talking head videos are less enjoyable to watch in my opinion so I didn't expect this one to get a huge view count. Back to the fun stuff for my next video :)
I was hoping that AI would be useful for creating new pipeline tools. All I want is a "Make Animatable Topology and UV's" button and that's it! I love everything else about the process and live to do it myself. Outside of that, AI is a hard NO for me.
I hear that, and I share your mix of hopes and concerns.
Thank you for this video Hardy. Last couple of years ive been in a real art slump and one of the major reasons is AI. Maybe the future is not so grim.
I'm really glad - that's what I was hoping the takeaway would be here. Keep making cool stuff! :)
As someone who went to art school in the 90s and watched the classic illustration profession crater in the face of the digital revolution (so many of the big illustrators of the time had to move to gallery work), I think AI will be an even bigger threat to the human touch in art. Once they work out the licensing and mitigate the threat of IP litigation, it will be off to the races for large AAA/creative companies. I think it's tragically unavoidable.
Thanks for the perspectives. That would be a sad result indeed.
Imagine AI + advanced and monetarily approachable 3d printing.
AI creates the image and the 3d printer produces it in oil for example.
This is definitely encouraging, thank you for taking the time to ask these questions to art directors. It helps a lot 💜
You're welcome! I'm glad this was helpful!
thank you for the hope 💜
this is actually dangerous - a young person might watch this video and decide to purpose their career in this area when actually it's has no future, we all know that. most of the human work will be outsourced to cheap locations
You're welcome. Glad this was helpful
I'm really sorry you feel that way. For what it's worth, my own son is an artist and wants to have an art career when he grows up - I fully encourage that and I believe he will achieve that goal. The world he grows into will be different from today but I believe there will be a value and a need for artists and craft even in the most dystopian of AI predictions. Hoping for the best for artists and the many other intellectual careers that this will disrupt.
Nice video!
It's good to see how the use of AI in all stages of production is almost seen as a disease by most directors. There's an ethical, moral and philosophical problem which AI will not get rid of anytime soon. I shared this on X(but could't tag you).
Thanks. It's an interesting time for the industry to be sure and I've been more encouraged than I was expecting back in 2022.
As a new-ish Freelance illustrator (working since 2021) I haven't noticed a significant dropoff in work. That said, rates have been the same, and money doesn't go nearly as far these days.
Well I'm glad that work has remained steady but sorry that wages are stagnant. That is a tough thing these days
I really appreciate you taking the time to make this video, Hardy. AI is a controversial topic these days, and I am very glad to have you shed light on a realistic, professional outlook of AI in the industry and whether it is even realistic to pursue Art as a career. While this video is extremely reassuring and makes me feel more optimistic about the future of artists still being at the forefront of the industry, it still leads me to question the vague aspect of some of the art directors' responses using AI as a tool for mood-boarding and iteration purposes. Do you see AI in that aspect still replacing artists as that work would have likely been done by a concept artist in the past, or is that something that we will have to worry less about as, as you stated, those companies still use artists for the final production result? Thank you again for the video, it is very motivating to see as a pursuing artist.
Great question and that one caught my eye too but that is an example of an area where work that might have gone to a hired concept artist was done by AI. However, when budgets are very limited mood boards and even pitch decks often have art grabbed from the internet just to get very early ideas of what they want to build. In those cases, it doesn't really seem like a job was lost, just AI stuff replacing google images. I really can't say for sure how disruptive this will end up being but I'm enormously encouraged by my own experiences and the artists around me in these first two years and I mainly wanted to share that. There is plenty to be concerned about with AI (honestly, I think all desk jobs will eventually be having their existential crisis like we have been recently) but so far so good - still standing!
Thank you so much for this video!
Glad it was helpful!
Hey Hardy, we've used AI in some of the work I've been a part of. It never makes it to the final product and almost serves as placeholder or to quickly iterate on compositions. Like the other AD mentioned, either mood board generation or helpful composition reference for me to create something. For example, quickly putting together a series of compositions in a fisheye or worm's eye view, it was helpful to have some AI images handy for reference.
I'm sort of at the mercy of my own AD and CD to do their bidding, but I've pushed back at times and said "I'm going to start with basic black and white sketches by hand" to make sure 1. I don't get rusty and 2. Keep the artist hand present in the work I help with.
Thanks Joe, great to get perspectives from a pro like yourself and I totally get that - we have to go with the flow of the team.
Art Director 5 is a champ 💪 lemme support those projects
Yes, I really liked those answers too :)
Thank you Hardy👍🏻🫡
My pleasure. I'm glad you found this useful!
I want to point out that a few months ago Treyarch posted a job listing on ArtStation that specifically called for artist to fix AI generated illustrations and assets.
I didn't see that one but I'm sure that will be a relatively common thing. Thanks for the heads up
Great content! Thanks mate!
Glad this was helpful!
From one artist to the next thanks for your view on AI .
Thanks, I appreciate that
I’ve lost hope in both the human element as much as the a.i element in the art industry. I’m going to do my own thing with personal projects
Sorry to hear that but I get it. Make art for yourself - this stuff is meant to be good for the soul after all.
Perhaps the more subtle threat of AI is not the replacement of the Artist as 'technician' in terms of the actual production of the images but the replacement of artistic judgement throughout the entire process. Who, after all, are the real advocates for quality in almost all creative pipelines? That would be the Artists themselves. It's those employed to create assetts that usually push the envelopes of quality and originality, sometimes to the frustration of those more focused on the bottom line.
But what happens when those creatives are replaced with AI? It's not just their technical expertise that is lost, it's also their broader aesthetic influence on the entire project. What makes AI imagery so seductive to the non creative is that it seems to offer a way to more quickly and cheaply achieve the same results as using human creatives- you just type in a prompt and like magic a fully realised image appears in seconds!
Now there may in fact be all sorts of problems with that image, in terms of it's relationship to other assets, or how it may or may not contribute to the story being told by the project as a whole- but who is now left to point these things out? In dispensing with the artist as technician you may have also thrown away something else, something that you may not have even realised you needed- which is that broader overview of what your images were intended to express.
If Art is ultimately about communication and we replace artists with machines, who then, is really in charge of what is communicated?
Amazing points and perspectives here. Thanks Paul. I can definitely imagine this problem affecting all kinds of AI implementation into our lives
There’s two sides to this answer. 1. AI usage isn’t regulated (yet) so a lot of companies still feel they’re in shaky waters when it comes to the usage of others IP’s or Content cause if the law changes they could get sued and lose millions. The other one is, for now, AI is mostly used on the pitch or moodboard phase of a project but the reality is voice AI is now tumbling increasingly on videogames and other media (ence the backlash by VAs).
I feel in the future, just as any technology, it’s going to be (hopefully) ethically and morally incorporated in the workflow of a project but I don’t think it’s going to kill artists just by existing. We have stop motion pictures when 3D is here, we have artisans who make stuff with their hands when we can buy pretty much anything pre-made at IKEA. The human component is and always will be the main ingredient to any creation and we as humans will always feel empathy for any human artist more than a factory, a machine or any software. Great video, thanks for sharing!
Great perspectives - thanks for this.
As a storyboard artist that work mostly for animation, I experiment with AI image gens but so far I have not incorporated them in my workflow. The output they make is not what my job requires.
Good to know! Thanks for your perspectives on this - I hadn't heard anything from that side of the industry and I'm glad to hear it
Hi, I have an unusual question, I was wondering what do you use when it comes to payment system when ppl hire you? I know it's maybe a bit personal or sth but Paypal seemes not to good, it takes a portion of your money
From personal experience it's already widely used in board games with overpainting slop outputs and fantasy book covers, plenty of those sadly.
Ah gotcha. Thanks for letting us know
Also, I think it destroys collaboration. How is a traditional artist going to bounce ideas off of someone if they can't even conceive of it? Imagine if everything had to grind to a halt for someone to generate some ideas because they're too insecure of their own. Exercise the meat in your head, you need that thing!!
Agreed! Thanks - great points.
I don't even see the point in using AI for ideas when it can only give a worse, messy version of other artist's work... why would you not just go look yourself? And be inspired for way more, instead of taking a lazier, worse route? I would be very embarrassed if I worked my ass off developing my own skills, to have my input shrugged off in favour of whatever the AI threw together. And of course, spend all my time fixing what it messed up.
Right on. The tools seem really clumsy currently too (which is likely to change). It seems to give you loads of weird stuff that is ok but not really what you need
Is there anything like an organization pushing back against ai "art" where we can participate or donate money to?
Yes, the Concept Art Association is leading the charge there, and they have a Go Fund Me page that is going towards legal action and supporting artists
Eben beat me to it. They have great resources and advocacy options. It is led by Karla Ortiz who is a legit hero for our community and you can donate to support their efforts. They are hiring lobbyists and helping to drive legislation to govern the ethical use of image generators. I've heard Karla and Steven Zapata refer to this situation as a boulder rolling down a hill - we can't stop it, but we can nudge it in a less harmful direction. This is definitely the cause to back if you're inclined.
Companies are FOR PROFIT.
As soon as AI can do the same job with minimal input, they will begin hiring less and less artists.
It's the same as manufacturing: if you want your company to be competitive nowadays, you do manufacturing in countries like China.
Sad but very plausible. Thanks for your perspectives
Oh so handsome..
Haha. Thanks 😎
Thankfully I don’t need to use my art to make money. I don’t post anything either. I do feel bad for artists that do work in the field. Ai will only continue to grow. It’s not going anywhere.
I watched a demo from Craig Mullins at Lightbox and he was largely of the same mind. Not really because of AI but he was telling everyone not to use art to make money because it would turn something you love into something ugly. He was a really interesting speaker. You're totally right, it's not like this tech is going to go away - hoping for a future where it doesn't make human input atrophy completely
@ real art made from humans will always have a place. I don’t think it will be the mainstream stuff. There’s still people doing claymation and other old forms of art.
Everyone says no to AI because it's not good enough yet and they have an option to pay to real people to do the job. Every company outhere is chasing AI and how to replace positions with it to cut costs and flex to investos, in few years these AD might not be given option to pay to real people to do the job because their boss said so. I noticed that a lot of sci-fi 2d concept artists I follow on Artstation started to do more designs in 3d because because it's more valuable for production and AI can't do 3d.
Interesting that you're seeing that shift. I had noticed that too but I hadn't made the connection. Thanks for the perspectives.
There seems to be a huge lack of understanding regarding what the tool is and how it works, when voiced by people who don't use and won't try it.
That's fair - my understanding is pretty surface level and is limited to my particular corner of an industry and how it affects us. If you want to elaborate, I would welcome your perspectives
@@fowlerillus
When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used.
There are many tools so lets start by choosing one to talk about: Stable Diffusion.
Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory.
Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution.
Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px:
There are a total of 4,096 (64 x 64) pixels on a canvas.
Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space.
For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000).
Total images = 16,777,215×4,096 = 68,719,476,736
We can store every single image into the vault with a unique address.
What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics.
Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters.
To generalize the process:
1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form.
2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art.
3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make.
So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process.
Now when it comes to ethics, that is personal choice.
For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one.
Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things.
Is this a decent start. This is a pretty huge simplification of one tiny subset of AI. But I think it's a fair starting point to bridge the gap and remind artists that mathematics and all the digital tools you used were created by people who were artistic in their own, different sense. Mathematics is as artistic as art is mathematical. 💯
@@fowlerillus
When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used.
There are many tools so lets start by choosing one to talk about: Stable Diffusion.
Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory.
Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution.
Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px:
There are a total of 4,096 (64 x 64) pixels on a canvas.
Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space.
For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000).
Total images = 16,777,215×4,096 = 68,719,476,736
We can store every single image into the vault with a unique address.
What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics.
Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters.
To Summarize the process:
1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form.
2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art.
3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make.
So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process.
Now when it comes to ethics, that is personal choice.
For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one.
Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things.
Remember, Mathematics is as artistic as art is mathematical.
@@fowlerillus
I've tried replying twice, but the reply isn't showing for me and I don't want to spam you.
When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used.
There are many tools so lets start by choosing one to talk about: Stable Diffusion.
Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory.
Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution.
Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px:
There are a total of 4,096 (64 x 64) pixels on a canvas.
Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space.
For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000).
Total images = 16,777,215×4,096 = 68,719,476,736
We can store every single image into the vault with a unique address.
What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics.
Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters.
To Summarize the process:
1. Mathematical Process: Just like solving an equation, Stable Diffusion uses complex algorithms to transform random noise into a coherent image. This involves a series of calculations and adjustments, much like how an artist makes decisions about composition, color, and form.
2. Decision-Making: The AI makes “decisions” at each step, guided by the text prompt and its training data. These decisions are based on probabilities and patterns learned from a vast amount of visual data, similar to how an artist draws on their experience and knowledge to create art.
3. Byproduct: The result of this process is an image that might seem “soulless” because it lacks the personal touch of a human artist. However, the underlying process involves a form of decision-making that mirrors the thoughtful and informed choices artists make.
So, while the output might appear mechanical, the process behind it is quite sophisticated and involves a series of weighted decisions, much like the artistic process.
Now when it comes to ethics, that is personal choice.
For example, computers can be used for good and evil, but how they are used is the choice of the user. It's a tool we use to make our lives and workflow easier, but it can also ruin lives. Now, for some who are disabled, it means having a life again where you may not have ever had one.
Stable diffusion does that for me. I have aphantasia and cannot envision things. I have great ideas but I cannot bring them to life. I love learning and took the time to learn art theory informally. I have the ability, but the moment I look at the page, I can't create anything, because I can't see and envision anything. Generating a reference gives me a chance to "Draw from this reference what I like" in my style. Would you consider it wrong of me to use inspiration, because of where it came from? If I copied it completely that's one thing, but I only draw inspiration from it, and will always be transparent with my process. Does that make me bad? I don't and won't sell AI, as I respect artists, but I believe I have a right, technically being disabled to use a tool to help me to be a creator to create and profit off of my own things.
Mathematics is as artistic as art is mathematical. 💯
When talking about Artificial Intelligence--which is often used as an umbrella term to refer to all things pertaining to the field of AI--it is important to make the distinction between ethics and functionality. That is to say, there is a difference between what a tool is and how the tool is used.
There are many tools so lets start by choosing one to talk about: Stable Diffusion.
Stable Diffusion has it's foundations rooted in Mathematics. It is a beautiful and artistic union of Statistics and Graph Theory.
Mathematics itself is all about using logic to arrive at a solution. This is a process that requires one to gather information, sort information into meaningful groups, categorize those groups by value, assign meaningful weight for those groups, determine outputs, reiterate on decisions and finally, arrive at a solution.
Let us consider an example to understand how math and art relate:
Given a theoretical vault for storing our art, and an infinite number of canvases of 64px by 64px:
There are a total of 4,096 (64 x 64) pixels on a canvas.
Each pixel can have 256^3 (16,777,216) possible colors in the RGB color space.
For each of the 4,096 pixels, you can change its value in 16,777,215 different ways (from 0xFFFFFF to 0x000000).
Total images = 16,777,215×4,096 = 68,719,476,736
We can store every single image into the vault with a unique address.
What this means is that every possible canvas of 64px by 64px can be chosen from an address in the vault, if only we knew the address that it exists. And this goes for every possible canvas of all sizes. This has nothing to do with AI, this is just theoretical thinking and pure Mathematics.
Scaled, this means all images in infinity already exist, if only we know how to reach them. Stable Diffusion is the instruction, or weighing of many options based on human decisions, to arrive at a solution to a problem, where the problem is finding the pixels that meet all the requirements that we, the human, desire, when we set up the generation parameters.