[00:03] generative AI is everywhere, including in art schools. So, this is a post by Threads user Tivu San writing, "Can you spot the AI? Take a close look at the images below." So, here we have one image. This looks like a fan art piece
[00:17] of Miss Fortune from League of Legends. Here we have another one. Is this Riven from League? I could be wrong. Another one, another one, another one, another one, another one. There's a fair amount, you know. We we can just skip through
[00:31] real quick. This one looks quite AI, but there's a lot. There's a lot of art pieces here to look through. It reads, "Yes, they were all created using AI by mentors and students from a CG school that teaches how to generate art, in
[00:43] quotes, with AI tools, aka neural networks. This is a long thread. The school is CG Lab. They've been teaching artists for over 10 years, originally for concept artists. However, according to the school, in order to stay
[00:58] up-to-date with industry demands, students now need to study AI tools. They refer to AI as neural networks. Let's take a look at their approach and screenshots below." This is their website, cg-lab.pro. So, this is their
[01:11] Russian website, and they also have an English website, which is cg-lab.school. One thing that I find really interesting is on the school one for the US one, or tab on the left-hand side, courses, digital art, blah blah blah blah.
[01:25] There's no tabs on the right-hand side. But, if you look at cg-lab. pro, on the left-hand side we have the courses, on the right-hand side we have some extra tabs, including a tab for neural networks. And this is the one that OP is
[01:38] talking about. And here it reads, "The CG industry is changing. Neural networks artists. They've already become part of animation, illustration, and design. If
[01:50] skills, then this course on neural networks is for you." So, apparently in 8 weeks, you will learn to master Stable diffusion, mid-journey, Dali, and ChatGPT, speed up your workflow, and become a more sought-after specialist.
[02:02] Which is interesting. So, we have authors for the course. Here we have a section of image generations from the course authors. So, these I guess are their generations. Nice. Good for you. And we have the course overview. The
[02:17] artists. I find that a little bit contradictory, honestly, given the topic of the course, but the course authors are professional 2D artists with many networks as part of the workflow in the gaming industry, film industry, blah
[02:30] blah blah blah blah. They solve real-world work problems by speeding up the processes, dynamic compositions, and emotions because it's difficult to achieve dynamic, non-standard poses, and complex emotions in characters and
[02:43] Hmm, kind of like uh if you drew it yourself, you can draw whatever you want, including dynamic compositions and emotions. So, this is the program at install it. Sorry. How much hand-holding do you need
[02:57] don't know, boot up their computer, too? We install and create the first generations in and then a list of AI softwares, I guess. It's a 1-hour lecture in week one. So, week two, stable diffusion basics, complex
[03:10] prompting, text to image, blah blah blah blah blah. Uh now, week three, image to image prompting. Week four, control net and upscale and stable diffusion. So, they teach you how to color your own sketches. Nice. How to recolor the
[03:23] image, generate poses. Then week five, we have mid-journey nano banana and ChatGPT, overview of ChatGPT tools. Week six, you stable diffusion training where you create mini models to generate your own characters in your own style. You
[03:36] character. How do you How do you create the data set, exactly? Just curious, you know? Is it going to be off of your artwork that you feed in yourself? Which do whatever you want, I guess, with your own work." Do you feed other people's
[03:50] stuff into it? Hmm, you know, how is that made? Laura training method on models on your computer learn matte painting in the works of the school founder Lina Sidorova training in the watercolor
[04:03] style and week 7 plus 8. So some extra stuff for 2 hours. Wait, I just scrolled down. This is kind of absurd. It's 32,000 32,000 Russian Ruble, which is 409 USD. It's
[04:16] Russian Ruble, which is 409 USD. It's almost 410 USD for 17 and .5 hours video dollars per hour. Now, don't get me wrong. I think paying for courses of stuff that is worthwhile is good, you know, like from a established artist
[04:30] whatever. Personally, and this is just my opinion, I I wouldn't spend my 400 hard earned buckaroos on this. So the thread continues. Now, let's take a look at different AI generated images and how
[04:42] student fed a sketch into stable diffusion, then generated all the different tools. They combined everything into a collage. So this is the final image. So they placed the image in image to image stable
[04:55] diffusion. They chose the best result with Laura. They made a background in midjourney, combined it, and then edited the face, made a portrait, and used one. This student made their own sketch, but instead of finishing it, they ran it
[05:10] process with the character and turned it into a collage. Pay attention to how additional AI tools are used to upscale already generated images, removing AI artifacts in the process. So here we have the final result. Version 1, this
[05:22] sketch of the background. This person can draw. Bruh, they colored it through They repainted it through image to image on XL and upscaled it. Then they did a
[05:34] character sketch line and flat fill for the character, colored liner collage and correction, repainted it with AI. Three, this student used both stable diffusion the second step, she photo bashed the sketch using midjourney images from step
[05:47] different AI models. So, this was the final image. So, here they prompted a lot in midjourney. They drew in a sketch. They used various models to try finished the details with in paint in stable diffusion and painted it in
[06:01] and they used different neural networks Another example, the student generated a background based on a quick sketch then the same process for the character as well. And then everything was combined
[06:15] they did a sketch and then they generated, up scaled it and then made quick sketch of the character. I tried for a long time to achieve the pose as in the sketch. It stubbornly turned the character around. There were many
[06:29] attempts. I left the most adequate ones and chose the last one. Again, almost as your character facing whatever direction you want. And then they added, I Photoshop. And they moved the character out of petals and they adjusted the
[06:42] size. OP ads, a few more examples. So, here we have one example, the MF example, another example, another one, another one. It reads, "For some reason finish their sketches and instead feed them into AI training the AI on their
[06:57] a good example of it. Sometimes they draw the sketches themselves and sometimes they photo bash ideas using AI. All these tools are evolving and it will become increasingly difficult to tell whether somebody is being dishonest
[07:09] look for AI artifacts in the artwork itself because an AI user can up scale parts together and combine many techniques. And it's also not enough to simply see the art creation process since we already know many examples
[07:23] would like to know the answer. How we can protect ourselves from AI scammers new artists joining the community. Honestly, I feel sad. I know this is not these schools are graduating hundreds of students every year. And it adds, to
[07:38] we're living in, I'll add these student reviews about the course and the course's response regarding whether it's ethical to use AI. They paid for this approximately 450 USD. And right now I checked it it was 410. So a student
[07:51] wrote basically, "I didn't have time to study neural networks on my own, but I course and it was a great decision. And before the course, I just used AI without a special system and at random." Another person said, "Of course, it's
[08:03] fast that a lecture can become outdated in a month and a half, but all info was relevant." Another person says, "I like the course. I started almost from Another person wrote, "I was looking on courses on AI for a long time. Why?
[08:17] keep up. And because studying it on my own is very time-consuming." And on a FAQ apparently on how to use neural networks ethically, it says, "Each artist decides for himself how exactly he uses neural networks or the employer
[08:30] is responsible for this. Ethical use option. For inspiration references and I Stable Diffusion. For final textures and image pieces, DALL-E and Adobe Firefly that their data sets did not involve the work of artists without permission. Even
[08:45] regarding that, using Midjourney and Stable Diffusion in the process of using AIs, you know, that took stuff without permission. And that's the frustrating part, right? It's frustrating that it's being taught in schools and it just kind
[08:58] of feels like a losing battle in some regard. Again, I truly believe that day. Nothing can be and replicate human artwork, but a lot of people in the comment section kind of feel very like dis- heartened about the situation,
[09:11] feeling really frustrated, feeling frustrated that it's being actively used in the gaming and other industries, that many companies are just using it no their employees to use it. And a really interesting comment says, "They're
[09:24] creating art and it's so upsetting." Another user says, "What a terrible time implemented all over the world stating which works were done using AI. At this point, it's is even a drawing anymore, even if the sketch was normally done,
[09:38] devastating that people who are willing to draw before turn to AI. For me somebody who never even held a pencil in their life. How can they be proud of thoughts. Like there's people in the comments who will be like, "Oh, you're
[09:51] blah. Your choice. You can do whatever you want with your life. I don't like know, it's built off the theft of artists. Not cool to be using a tool whatever they want. But, if you are going to choose to use such a thing, at
[10:05] least disclose it. And the issue is many, many, many, many, many times the in the process, whatever. And there's plenty of people who are larping as actual artists who are using generative AI to make their works. And there's a
[10:18] lot of people who do not want to purchase stuff that is made with disclosing it because for fear of it obviously hurting their customer base, right? So, the clear disclosure of it is at least the bare minimum, in my
[10:31] secondly, it might seem like it's all doom and gloom, but at the end of the day, real human art is a valued, real human art is appreciated, and people are still enjoying it. The other day there was like a post on Twitter that Steam
[10:44] made, that, you know, the game platform, and they used human artists and tagged viral on Twitter. People were like, "Yeah, W Steam. Go Steam for doing lot of demand for human art. People still love and support human art, and
[10:58] day. So, if you're an artist, keep making art, and up to you what you do with your money, but maybe don't support things that are made with generative AI. if you like and subscribe. Again, don't send hate to any parties involved. I
[11:11] out there, and a coffee with free resources, and I also stream on Twitch 3 6:00 p.m. PST and Thursdays at 7:30 p.m. where I do a podcast with my friend. I'm shoot soon, so I'm really excited for you guys to check it out if you guys
[11:25] check me out on my streams or in another video. Peace.