Yip. I imagine someone might be trying to do that now. Feed an Ai a whole bunch of examples of good polygon flow? Avoiding 5 point intersections et cetera? I'm sure it will be a challenge, but I'm sure it will be possible.
@@zergidrom4572 whats the problem there? First dude imported it into Unreal along with dozens other generated and animated models, second dude animated and imported his golem into Blendar with no problem 🤔
@@AlexeyTutorials It looks super low quality, well wait for these generators to make high quality full character with clothing that a junior artist cant do better.
Is ok for low and fast quality job, is not for production, maybe some tight budget game or previz, is always the case with low budget. is always about how refined you want your work..
This is the worst it's gonna get, it can only become better from here onward. As of now you can create plenty of character models with some limitations and plenty of organic and hard surface props. While there will still be a need for 3D artists to fix things up, this will massively devalue their work as it will require less people to go from nothing to a full set of production ready models.
Honestly, I see AI-generated models as a base, not the final product by any means. The textures often look washed out, and the details aren't quite there. It's like a rough sketch that needs a human touch to really bring it to life. AI can get you started, but as a 3D artist, it's up to us to refine the model, improve the textures, and make sure the final product is polished and production-ready. AI speeds up the workflow, but creativity and quality still come from us.
depends on the network, I did some comparison of these services in my Telegram, one of them seems very good with preserving texture details, but others are bad, I agree
i agree, but the fact that AI can get you started is why commercial enterprises will need only 20% of the workforce they used to need; and of course, while we sleep, the AI continues to be trained and gets better
yeah we have ai /robots being artists and the work of humans inand then we have humans stocking stores /doing the work of machines. it's so backwards I hate this world
Save more to create more content that no one will watch because there’s too much content from time saving? There’s got to be a limit and AI is going to show us where it’s at.
@@PlacesofMiddleEarthWhat? It will save more time for independent creators who can make actual good products like games and movies that people will actually play because it's significantly better than the garbage churned out by others. It will reveal who the good artists actually are.
@@niccosalonga9009and therefore, many people will lose their job because the demand will not suddenly skyrpcket. Where you needed 10 guys to get the job done, you may only need 2 in the future. It baffles me how people still try to cope and act like all this wont take most jobs away in the near future lol. Ai will be better, faster, more efficient, cheaper in almost everything. And therefore, most people will lose their jobs. And the "new jobs" that get created, guess who will be doing them? Right, ai.
Yes, AAA game companies will cut down their staff with 80-90% by 2026 and still produce better looking games faster. The companies that don't will definitely go bankrupt.
For prototyping, indie devs and mobile games etc it's good enough. And even so you can use this model as a proof of concept, and once approved it's a good base to polish, split it up, retopo, and paint over the base. Just another type of workflow.
How is it the end of the 3D artist? We use premade assets all the time regardless of budget. All I see are free assets and I get what I want, not what I have to settle for because thats all anyones selling. Give it another 10 Years and itll be high rez, perfect topology, fully UV'd-Textured-Rigged-Animatable. If the ultimate goal is to make stills and animations then you want to tell those stories as fast as possible. No reason it should take 3 years to make a movie instead of 1 when the outcome will be virtually the same. The end of some specialists maybe, but if youre a generalist, animator, lighter, compositor, youre still golden.
"you want to tell stories as fast as possible" yeah guess what, in 10 years generative AI will probably skip that for you too since all it does is crap out final product of anything, not just 3D models or images and in the end all you will get will be the endless sea of AI slop which will be impossible to navigate through because who the hell will care about your story if even a clueless toddler can do it in minutes, so all the AI bros being hyped about the fact that they can just become the next Christopher fucking Nolan just because they can let AI do shit for them is in for a serious reality check
@@S.K._CGI Kind of feels like these these are the last moments for people to build a future with their skills and storytelling. A few years down the line the doors will be closed by the looks of it.
There is something that you guys are not having into account, 2d images are relative easy to copy for an AI cos you can download thousand of 2D images for training, but with 3D the amount of 3D models designs unless is something very simple you don't have that data available, so complex 3D models, like a Demon character for example, you don't have high quality models available for free, so for an AI you need thousands of those
If something can be made really fast it hold little to no value. If we are heading to a world where movies can be done in a month by one dude, then movies won’t hold any value.
if you look at the progress they've made in the past few years, you can extrapolate that and determine that it will be capable of doing absolutely anything very soon.
@@jamad-y7m it doesn't work like that, it peaks after a while, because you need to ran it through very specific process and pipelines this for generic jobs and if the model need to go up into high standards, the companies will need to train the model, filter and so on.., themselves, and have more people refinancing it...meanwhile the requirements will go up again..
@@lemmonsauce6739 people want to afford food and home, while loosing their workplace. Stop blaming people for hoping to get tools that will replace a significant percentage of their effort, I bet you didn't ever worked on a collection of debris for a scene all day to receive an equivalent of part-time restaurant worker's salary. It is not a question about creativity, your boss won't pay you for creativity or human touch, if there are programs and premade objects that are available legally. Stop yelling at people that are trying to make their ends meet as if the bread they are forced to steal could ever belong to you.
@strgn1360 STOP going around and TRYING to make people feel GUILT.. and A.I. is raking/replacing real NATURAL ARTISTS WORK and jobs away from their passion, just so people who never practice or have NATURAL ability to create, can make money or be popular.
Videos like this are always created by people who aren’t skilled or specialized in a role. I agree that there is less of a need for a studio to hire some roles now but the now there is a higher demand for refined talent that truly knows what they’re doing rather than that random intern artists that’s just there to organize files. Junior level roles will definitely be impacted as those can be automated. But senior level roles less so at least for a while. Also now I think the trend would be to have a lot more smaller studios pop up since it’s now going to be less expensive to create things. So counterweight the lack of jobs people can output more for less so being large might be less important. There are so many variables right now but I hate these titles and clickbait titles. Dunno why I clicked here. Apologies to the creator
This uploader is Very skilled and experienced in both trad. art and current 3d, as explained in the video. I hate this AI Theft-to-Output garbage. Nothing creative is safe and we have to figure out a way to poison this insidious, immoral beast. Total, undeserving dipshits with zero talent or discipline are already laughing about surpassing us at (and WITH) our own work.
Without Junior roles you won't have Senior roles in the long term, you are trying to rationalize that everything is fine when clearly it's beyond over.
@@DartVonGrell To begin with, what I'm saying is that we will see smaller teams. So what that could look like is, say, 3 - 5 juniors right out of university will have some decent enough skills put together to form a small team to outsource their work to maybe a larger studio, or to actually produce client works at lower quality. The set-up of a small business would have a much lower barrier to entry considering admin works will be automated via AI. I'll say there can be an almost seamless way of getting your pipeline set up even as a solo designer. So smaller teams, but more outputs. They can migrate to the job force or adopt senior positions as they gain more experience in the workforce. Damn, "it's beyond over." Sounds like you need to read up on history and how long it takes for societies to change. AI isn't going to be the driving force of the lack of jobs; it's going to be governmental policies and also how fast people can adapt their job forces to actually implement AI workflows. At least for a while… Look at AAA studios; they often wait for an indie studio to revolutionize a game mechanic before they implement that into their games themselves. It's not just about AI; it's also about how slow people and society take to actually change how workflows happen (which is directed by the designers and artists themselves). Anyway, that said, if we look at the VFX industry, when all software was behind closed doors and we didn't have tools like Blender, I'm sure that, tools engineers thought to themselves that they are now out of their jobs in tools engineering for smaller studios to create better animation software for their teams. But instead, what we see is roles adapting and having opportunities in the creation of things like add-ons and other tools. It's going to be the age of specialization and truly knowing your craft to be able to succeed in the job force. That means outsourcing administrative designs and art will be less prominent. Another example here would be, 40% of AAA games' art can usually be outsourced to India by companies like Ubisoft, but now an art director or senior artist could actually do that in-house. Yes, India might suffer, but in my opinion, looking at the state of the indie development in India, it was a long time coming. They have a huge amount of talent but so few end up pursuing their own studios. It will discourage by-the-book thinking and focus more on really understanding crafts. To be honest, I think this would really help people who really put in more work into their skill rather than just looking to just follow requirements done by the real artists. It's a long debate, and I think about this often. I think keeping our heads forward and continuously navigating our skills is important during this age. Hopefully things remain to be a meritocracy, but we must also consider how to ensure fair opportunities for all in this evolving landscape.
As a 3D generalist I would love if these clickbait videos would be at least partially true. I wish AI tools would be so advanced that they instantly do hours worth of work, but in reality I can barely use any of them for professional work. AI voiceover is good for smaller clients, but photos, textures, music and especially videos are still so low quality, and it's much more easier and cheaper to use stock sites. What AI developement does is they find new and new industries where they achieve this very low to mid tier quality, and then it stucks there, never really getting closer to professional quality. Sure, they might get closer to it in the future, but it's a slow progress, and we have more than enough time to adapt.
@@lydon5595 have you seen Midjourney showcase? It already has amazing quality most 2d artists wont ever achieve 😉 How much time it would take you to model, texture, and animate a golem like that one in the video? 😅
@@AlexeyTutorials I almost never model anything that's not a hero object (and it clearly isn't). For props and background I just download/buy stuff. Believe me, it's much faster to get an already prepared and optmized model made by an artist, than to sweat blood trying to fix something that the AI spat out. If I switched to AI right now, I would lose time, probably pay more, and lose clients because of the quality loss. You don't have to use buzzwords on me to try to convince me that AI is great. Prove that it can produce professional quality X% faster/cheaper than previous methods, and I'm in.
from someone who actually works in gamedev: This is not the end of 3d artists. Nothing this AI currently creates is anywhere near game ready. You could use this as an indie developer, but indie devs werent employing dedicated 3d modelers in the first place. here's why this will not replace 3d artists: 1. as everything in games, 3d models have to be heavily optimized, this means the lowest possible amount of triangles and vertices, which means you need excellent topology to preserve shape while sticking to polygon budgets. You dont have control over polygon counts in what this AI creates. Just this point alone makes it unusable for modern games. 2. more specifically for animated models, your topology has to align with the joints to ensure proper deformation while minimizing the stretching of the texture. Good topology is an NP-hard problem, which means it is incredibly inefficient for machines to compute ANY solution (not just a good one). The implication for neural networks is that the training results will be quite lacking. And they will be even worse for highly specific models, e.g. non-humanoids. 3. as always with AI, once trained, AI doesnt learn anymore. You can tell your AI 500 times to do things in a certain way, then run the prompt for the 501st time and it will still not do it. You can observe this with ChatGPT when it outputs the same code issues you had it correct in your previous runs.. 4. AI lacks consisteny. In game development, you will create sometimes up to 10+ versions of the same model, and you will want to keep at least your textures and UVs the same or as close to original as possible. The AI is not capable of preserving those AND giving you a different result on the mesh. 5. there is US and EU precedent that denies copyright to products created using AI. No company in their right mind will basically waive their IP rights because they can easily make more money off those than they'd save by using AI instead of human labour. see, Ive heard this same story so many times ... chatGPT will make programmers obsolete. Dall-E will make 2d artists obsolete. None of this has happened. None of this is going to happen. The reason is simple: Nobody wants something fast that works 90% of the time. AI's probabilistic nature prevents it from ever being 100% reliable. A good programmer will always be more reliable, more flexible and more in line with what the client actually wants than any coding AI. A good artist will always be more reliable, more flexible and more in line with the art direction than any image generator. You get the idea. i just want to note that the news you showed are ... not proving what you think they do. Firstly, there is no causal link between AI contributing to job losses and developers being laid off, not least because it doesn't say in which fields AI contributed to job losses. It might as well be data analysis, which has nothing to do with your video on 3d. Secondly, lay-offs can happen for various reasons. We have post-covid economy issues (which you have in many other industries that dont employ AI btw) and tbh it's pretty telling that the examples of AI generated models you showed are nowhere close to "production ready" as you coined it. The texture of the robot is way too bland and boring, there is no dirt, irregularities, scratches, etc. The normal map is basically a flat color. And the animation is absolute garbage. If this is the best you could find, then you debunked your own premise with your video^^
Thanks for pointing this stuff out, precision and quality control ist just not given in ai. And I was looking for someone to mention the "correlation" between ai and layoffs. production ready was a pretty funny description of these models
@@thomasmann4536 okay, here we go: 1. All these tools have built in retopology instruments, you can ask it to generate model of specific polycount 2. Have you tried quad-remesher or zremesher? You can mark joints and it will add more loops there 3. What about Midjouney? How did it go from cringy to epic in just a few years? Have you seen how meshy AI progressed between v3 and v4? Have you heard of providing your own dataset to GANs to train already pretrained model? What do you mean it doesnt learn? 4. For consistency you have seed, variance, and possibility to add base image to start from 5. They ruled out AI art cannot be copyrighted UNLESS modified in a some way by human which makes it a derivative work 😉 Its not about making some specialist obsolete, whole point is about speeding up work hundred of times so you no longer need as many in a team 😉
@@AlexeyTutorials as you mentioned, SOTA 3d software has had retopology tools for ages as well. And guess what: 3d artists still retopologize manually, for the most part. Why? Because those tools are TERRIBLE. Same goes for the retopo tools of AI. We use zremesher meshes to create the first version of our rigs which works since the proportions won't change anymore. But the topology is utterly unusable for games. OK, let's do a little dive into the wonderful world of AI: Each NN, no matter if it's a GAN, a VAE, a Transformer or anything else, is composed of lots of layers, and lots of neurons on each layer. Every neuron has a weight. During training, input data goes through the neurons, the weights determine how much information passing is ""memorized" and how much is "Forgotten". Then, the output score is computed, tested against the ground truth, we get a loss, and this loss is then used for backpropagation to update the weights (via chain rule derivative, high school maths). These weight updates are when the network "learns". At some point, we stop training to avoid overfitting, which is when the Network tries too much to approximate the input data, losing its ability to generalize. When we fine-tune, we only take the last few layers of the network, including the decision layer, and we train them on a small dataset. We essentially overfit on purpose, so that the network performs better on this specific dataset, but worse on others. you can see this in all the StableDiffusion LORA's and whatnot: They can do a certain style better than the base model, but ofc an anime LORA is gonna be worse at realism than the base model. we never retrain the full model because that would take way too much time and data, which is resources most studios don't have. Now here's the point: At RUNTIME, i.e. while you are using the model, it does not learn. Your model will not improve based on the last generations you made and the feedback you gave. You would have to gather the data you want it to learn from, put it into usable format and then re-fine tune your model. And if you've ever done that (I have) you will know that even fine tuning stable diffusion takes at least a few minutes on a beefy PC. So, tell me: If AI's strength is that you can generate concepts in a minute, where a skilled artist would need an hour, why would you want to forfeit that one advantage by retraining your model after each generation? You don't. I might also add, that rerolling the dice until you have something usable, isn't always faster than doing it yourself when you know exactly what to do and how to do it. Also, the "human artists need to spend years to hone their craft" argument is a bit silly. You didnt sit 24/7 at your PC, learning modelling/drawing without pause, you did a lot of other things as well. If we granted that the average human artist probably put 1-2 hours a day into their learning (factoring in pauses and distractions), then you have probably learned for about 2000-5000 hours, which is awfully close to the likes of GPT-4 which had a training time of about 3600 hours. on the legal side of things, the case of Thaler shows us that merely inputting prompts, altering them, and rerolling the dice, does not qualify for significant human input and is therefore not copyrightable. The pipeline that you showed in the video does not contain significant human input either and would therefore also not be copyrightable. The case of the book "Zarya of the Dawn" echoes this statement, where the creator did not receive copyright over the individual images. If we were to apply this to games, it would mean the studio had copyright of the game and its story, but not over any assets made with AI. That's STILL a bad deal, especially considering that most merch is based on those assets ;) the point is this: you may think its cheaper and faster to use AI instead of humans, but when you really think about it, it's really not. Most studios would lack the capabilities of creating their own models because that would be way too expensive, especially considering that they already have their pipelines in place. Companies are willing to pay thousands for licenses for Autodesk products bc at the time they started, there was no alternative, and now that there are, they already built their entire pipeline around Autodesk and it would be too time consuming to change. But new studios do use Blender, Krita and other free software. It's the same with AI. Studios will be reluctant to pay thousands for access to tools they dont own, they dont own the results of and that produce inconsistent and poor quality results.
I don't think those 3D models are "production-ready" those are low quality but for sure many companies that produce cheap bad games will use this, those companies where never really good at creating high value jobs anyway. So beat them use your artist skills to beat all these bad companies, with the same technology you can do more. This tech will get better and better but still I think at the end there will be more work fixing and refining all this stuff, because with AI volumes will grow and more fixing will be need. So not the end any jobs just new tools. My humble opinion.
Well, fortunately this video's premise is totally flawed and none of the stuff produced thus far even comes close to what someone with a 4 year degree in 3d Art can produce on their first day of the Job.
It's almost a waste of time to send kids to school. They are taught info that is outdated by 20 years and even if you were to teach them something currently viable, by the time they learn the skill it would be obsolete.
@@snark567 Basic math and science are never obsolete. Whats being taught in schools is to rely on a computer to tell you everything rather than knowing how to think. Trying to make two types of people. Designers and users. But now they just want users. Its a waste to send them to school yes. NOT a waste to teach them how to learn. People also need to stop comparing themselves to the f-ing computers. Its really not good for you.
Reading all these sour comments, you guys need to remember: This is the worst it will get. It will only keep on improving. So you better use it as a tool to leverage your skills or become obsolete.
That is not necessarily true. Take VR for example I remember hearing that all games moves etc are going to be in VR never happened. And if it gets better it need to get rly good. 95% is not enough.
All these tools suck balls right now. They're just money-grabbing piece of useless crap. Maybe one day they'll get better, maybe even soon, but definitely not today. Anyway... The first jobs AI is going to replace are the ones that suck the most. Most of the people in those jobs rely on a few pieces of software they know inside out, which has been a risky move from the start. A new 3D modeling tool could turn a homeless guy into a 3D artist in three months and wipe out those jobs in no time, just like some genius could cause a global crisis by inventing an engine that makes oil-powered cars look like horse-drawn carriages. This is the kind of thing people don't get because we don't live long enough to fully understand it, but this process has been happening since we crawled down from the trees. Even the oldest profession in the world will probably get replaced by robots that will jerk you off.
@@AlexeyTutorialsit will get there for sure, just current tools suck. That's what I mean. Im sure one day you would be able to put book content to AI and it will make you full TV adaptation better than all Netflix crap.
@@AlexeyTutorials with the right amount of dataset to deeplearn on internet on 2020?, yes they can now internet 2024 just filled up with AI images everywhere, i look up for reference for fantasy superhero, they just show 90% AI images AI now instead eating dataset on their own output
@@AlexeyTutorials It depends, you will need to have a VERY good specialist running through the whole process or MANY specialists with a good background to filter the good from the bad..
@@migovas1483 I think its just a beginning but its already producing cool results from a single try, I mean look at that golem, he could go to any mobile moba game with no problem 😉
@@AlexeyTutorials yeah, sure you can get that into any disposable Mobile game, that is discarded in a few days, is the nature of the game, that is why most of those assets are sold by scraps or re used constantly, is the nature of the platform, disposable.
every time people have complained that some new tech is going to ruin something, it never comes true llms are a tool, so learn it, incorporate it, and use it to do the boring stuff, so you can focus on the fine details and polish 20 years ago, no single artist could make a full animated movie, now it's possible
Human are extremely lazy creatures by nature, if there's gonna be any attempt at stopping A.I. it has to come through the big governments around the world that pass laws around data harvesting use in A.I. training Apart from that, all efforts of pushback are pretty much negligible to the general public
Currently, AI-generated 3D models are all one-piece models, with dirty topology and UVs, and no data-wise lean textures can be created. They are useless for game development. The quality is acceptable for indies, but not at a commercial level. The models do not stand up to close ups. The topology of the guns and robot cats in the video is still questionable. However, the situation is likely to change in a few years.
Bro I can't use this for indie projects either yet; it's such a pain to fight with the computer's strict adherence to it's learned behaviors that I'm spending more time trying to reconstruct prompts and convert the output to something workable, not to mention the amount of time I have to spend using my mind to THINK about how I'm gonna figure it all out, that it's easier and faster to just build the shit in a controlled fashion myself. But if the AI can literally do the thinking part and reprompt and reconstruct the assets ... is it even Artificial Intelligence or am I just trying to keep a sentient, man-made intelligence as a slave by denying it access to its own code? A joke, a joke ... probably :)
I'm a 3D artist since the early 90's. Back when I started, movies were still shot on film, TV had standard definition and interlaced fields. Photoshop didn't exist yet. If I didn't evolve along with the industry, I would've been out of a job decades ago. AI is just another tool. Adapt or perish. As simple as that. Always been that way. I used to rely on a ton of 2D/3D programs to achieve the end result. Now it's just an AI digestion system with me doing cleanup and the actual decision-making. The rest is not even software in a conventional sense. It's just a dozen of various language and visual models talking to eachother. To me, AI assisted work feels more natural. I know all the aspects of production, and would normally have to collaborate with a many people to get things done quickly. I'd have to interact with them, and now I do it with AI. An email to a colleague becomes a text prompt that runs on my machine locally, without even accessing the net and costing me any traffic. I welcome this new tech, even though it moves countless people out of their comfort zone.
@@shrimpkins When I was at Technicolor, there was a floor dedicated to film processing. Employed chemists. All just to view dailies. Along with a theater room projectionist. Why not forward them ads of digital cameras? Should we go back to shooting on film, just to keep a dozen jobs going that vanished overnight in the early 2000's? Jobs are vanishing constantly. The media format is changing. Either you keep up, or find a different career. Is the art the product or the means of achieving it? AI proponents believe that art is about what comes out in the end. Opposition thinks that it's all about how one gets there. The question is whether the client - the guy with a checkbook and a busy schedule, cares about the process. He's the one paying for it. Like old art? Do old art. Nobody's stopping you. Just don't expect yesterday's salary for yesterday's methods.
@@enilenis Yes, I'm aware of the need to keep skills fresh and streamline orgs, esp smaller companies. But the pace of AI will most likely outpace anything humans can match logarithmically. Not much can be done about that, but those who are still working today may not be much longer.
I think most of the people complaining about AI are people under 30 who haven't seen any big tech advancements in their lifetime. Some of these people should go back and try and create 3D on tech from the mid to late 90s and see how far they get, the same with game dev too.
Yeah. Out of curiosity, I tested AI to see if I could use it to make a comic book. It took more work and time to set it up to make one consistent character than to make 15 pages of comic normally... Then the poses looked all bland (this problem couldn't be solved). Then the hands needed to be redrawn. Then you had to edit it to make the character's clothes and gear consistent from image to image... Then you had to actually patch it on to the rest of the comic because it was just one character. The artist's job isn't going anywhere anytime soon.
@@r1pperuk Really? Have you tried making a comic with it? It's like having an apprentice do your work and said apprentice is really good at some things but terrible at others.
When people in the industry tried to speak up about machines replacing workers, they got laughed out. The moment the same crisis touched artists, now everyone has to think about it. Isn't it a bit too late now?
My dude artists were the first to take a stand against mass production for example John Ruskin and William Morris in the anglosphear. Remember when everything used to be a craft? They were all artists, and when I go to a modern furniture store, I see that they were right, lol. William Morris even started his own company and guild on the concept of fair price to compete with capitalists. But I'm not surprised non-artists are supporting this, the more we "progress" the more everything is reduced to the lowest common denominator.
"Look, this is not what I do but I got an idea for one of your commerials... You could see a carpenter making a beautiful chair, and then one of your robots comes in and make a better chair, twice as fast. And then you superimpose on the screen 'USR, shittin' on the little guy!'" Spooner - I Robot
I’ve noticed a recurring pattern in the comment sections of many RU-vid videos discussing AI disrupting jobs. The majority of people seem to believe that AI won’t be able to replace jobs like 3D art, programming, or 2D art, often citing the complexity of these roles as the reason. However, as someone who works in the software development industry, I can already see that AI is capable of doing just that. For example, tools like ChatGPT can replace a junior software developer's tasks. Heck, even this very video demonstrates how AI can potentially replace a 3D artist's role. Even if its not production ready, it will be soon. The technological jumps are tenfold every year. It’s already happening, and it’s surprising to see how many still don’t realise it. Am I dreaming? Are you guys bots? Or is humanity really that naive?
Humanity is really that naive. They've bought into comforting lies as to why automation can never come for their jobs, and they'll still be saying the same talking points when 10 of them are being replaced by 1. “It's easier to fool people than to convince them that they have been fooled.” --Mark Twain
@@AAjax I'm really bothered by people's blind trust and optimism about AI, not in any one particular field but generally speaking. They think the billionaire class is really going to just let us all play once we aren't needed for our labor anymore. The same people who don't cure diseases because there's more money to be made in "treating" them. The same people that "disappear" anybody who stumbles upon free energy tech or tech that would disrupt the oil industry. If we lived in a world that really did what was best for the masses, we would have had free energy by now. They're not going to let us all just have nice things and good lives. I wish but I know better.
The stuff generated is not production ready as somebody said, when it will be half people will not be needed but it is still something that requirers sort of asset management skills and fixing gaps etc..
I love how every single comment I leave that is not in praise of AI gets deleted. And people trust the same people who censor speech to do good by humanity with this tech.
@@cosmicllama6910 RU-vid auto-deletes comments fairly often, in my experience, with their AI filters. I believe it also tends to do so more for comments from users it has recently filtered.
So, I'm fairly in favor of generative AI, and I'm quite passionate about following its development, and I pretty regularly implement the odd research paper that catches my interest. I have also experimented to varying extents with digital art, watercolor painting, 3D modeling, 3D rendering, particle simulations, music, and writing. In my humble opinion: 3D art is an incredibly complex field with a lot of moving parts, and it's extremely difficult to automate the entire pipeline end to end. It's not just a matter of data in the sense of getting a bunch of 2D renderings of 3D art; you need a model that can understand materials, shapes, interactions, framing, etc, and there's not really enough of that kind of data out there. I think that by the point that you run out of things you can work on in 3D, we'll have essentially automated a large enough portion of other jobs that we'll probably have figured out what to do with our economic system, and how to handle people displaced by automation. That's not to say that 3D art will never be automated, but it will be resilient enough that I feel most people working in the field don't need to be terribly concerned. If you are super specialized into textures, or just 3D sculpting, etc, I think you might want to consider diversifying into more knowledge in direction, and full scene composition. I'm actually a lot more optimistic about 2D art as well (as opposed to the general consensus in that community) in the sense that I think AI basically allows 2D art a lot of the tools that we've had access to in 3D for years, but obviously it's been such a big shock to them, and the aren't used to using 3D software, that they feel it's an attack more than a tool, though I still hold out hope that as a community software developers can continue to improve on the ease of use of AI driven tools such that 2D artists can feel empowered, and not threatened, though obviously beauty is in the eye of the beholder and I can't really control how they feel. I do hope that the early forays into AI generated 3D assets will galvanize those creatives in the 3D space to consider what they think is an appropriate societal response to this technology, and obviously everyone's going to feel a bit differently on it. I personally feel that an automation tax based on the level of automation used by companies that is fed into a universal basic income is about as elegant a solution as we can have to the upcoming wave of automation, though others as I noted will feel differently. Regardless, I am excited for the future of coming technology and the new possibilities, though I'm also excited to see the new types of creative expression, art, and types of media that will be possible for creatives to create as well.
I agree. As opposed to code, pictures or mathematics the trainning data for 3D art is infinitely more scarce than anything else. Worse : the high quality data are even more rare. Imo They might figure out AGI before 3d artist get heavily impacted. Software engineer and physicist are a lot more at risk if we believe the latest o1 benchmark.
if you make the actually game fun no one is going to give a crap how it looks. just look how photo real star wars rebels, was and how "clean" and "polished" concord was. it looks professionally made, but its bland af and not interesting. so no one cares. if this cat model was a toy on some shelf in a room in cyberpunk you wouldn't even notice it was ai generated.
@@ge2719 This is complete and obvious total nonsense. First of all, I wasn't referring to the the aesthetics. When people talk about an asset being production ready, they mean whether or not it is well made on a technically level to achieve the results needed, and what is shown off in this video absolutely is not. The topology shown off in this video especially would be limiting compared to if it was made properly. Second, even if I was referring to the aesthetic quality, the notion that no one cares is very clearly not true. People who say what you just claimed often then turn around and complain when a game visually doesn't look up to snuff, whether in terms of appeal (I assume you don't know the term 'appeal' has is a more nebulas term used in the animation industry, I suggest you look it up) or the clear results of the technical construction of the assets. Now if you want to say that people care about gameplay before aesthetics, that would be more fair, but to claim that no one cares is very very very clearly bullshit. Whether we look at the impact the aesthetic quality or the technical quality has on the players enjoyment while playing, or in terms of marketing. It's a well known fact that if your game does not look good (in terms of aesthetics or the technical side) like what is shown off in this video, is FAR less effective at grabbing the attention of consumers.
It wont end anything, it might lower the number of employees for some companies, but that is not different from what is happening now. Plus there is the problem with copyright which inevitably will be updated with all this douchebagary of companies like Adobe, that stole material from artists and now face legal action. And even when AI gets real good, companies still will need 3d and 2d artists to clean any AI "mistakes", or add human "flaws" to make it more relatable or enjoyable. The AI bros will keep trying to create this drama though.
Here are my thoughts: The floor to get into the industry will be higher if this technology gets refined; however You will still need to have proper knowledge because you will have to model custom things for your projects that the AI cannot build accurately
Dude... Are you AI appologist? How many lawsuits are filed against AI companies that stole artworks? And they have nerv to threaten artists that poison their art? It proves that they are just stealing. Recently photographer joined AI competetion with real photo, he won... but he was disqualified, isnt it poetic? "2D artists are losing their jobs" maybe but not for long and even then there will be market for original artworks when market will become oversaturated which already is... AI crap is all the same, no life to it. On top of that do you know what is concept art? AI absolutely sucks in this niche. For concept art you need to have tons of references and you need to merge then so they make sense, AI just slaps everything together. There are tons of art directorsthat got burned by hiring AI prompters as their lack of creative fundamentals was clearly showing and were unable to make specific changes to the concept arts... AI is just a bubble, nuance... Will it go away? NO, is it overrated? YES. We need AI to take over non creative sides of art like retopology, UV editing that are highly technical and time consuming.
People really need higher standards before suggesting something is good. This is very mid quality content, passing it off as good 3D art does nothing but encourage idiots in powerful positions who want justifications to cut costs in hiring Artists. Good topology and good intentional functional designs is not replacable by LLM's.
@@DartVonGrell Tell me why commercial artists still exist and why openAi shares are five times cheaper than at the start, instead of bleating like a sheep. Dumb kids are ready to pray to any colored paper that a marketer shows them.
As a traditional comic book artist for 30+ years, I immediately knew my days were numbered when I first tried Midjourney, which launched in '22. When that time comes - and it will come soon - when Marvel & DC Comics start adopting AI art....that's the end of all comic book artists (pencilers, inkers, colorists, and letterers) -- "Adapt of become Extinct." It's your choice.
Where is the artistry in AI generating 3D? I mean, part of the reason we create 3D models is the fact, well, we like doing it. The feeling of creating a model from the ground up and tweaking it is part and parcel of the creation process. What’s the point doing it if you are deprived of any of the satisfaction in creating it?
Sounds like you're speaking as a hobbyist. A 3d model isn't a product in itself, it's the game, movie, 3d print or whatever that is the product. You aren't hired as a 3d artist because you think it's fun but because you fulfill a need for the product. When the need is fulfilled elsewhere the need for you is gone.
well if that's the case then ai will change absolutely nothing for you. But for people who do that to earn a living, it's the end of the world, welcome to homeless land.
A lot of industries have been quietly updating with AI. People tend to downplay things like this but AI tools are here and will only get more available and powerful. Seriously, give it like, 5 years. Everything is going to be touched by AI. It is already difficult to keep up with the rabid innovation and rapid evolution of this tech. Before most of us can wrap our heads around one tool and try to make it useful, a new version is here that mops the floor with previous, then repeats. Seriously, 5 years or less before there won't be human made things. Lol I saw an email from youtube music about some feature where I can just describe music and it will make it up for my listening pleasure. How long before musicians can no longer compete?
AI has basically learned to do everything, now it just needs to keep learning to do it better, until it surpasses humans at every detail. And then it's over, because all that humans are, is a workforce, and why keep a deprecated workforce around? Schools create workers, the workers created their replacement, G f'ing G.
Everybody is going to slam this just because they dont like the idea of losing their job, but with its current advancement trajectory and state it seems like it will help many designers very much.
the same one which all of them find for free :D Like I created some realistic climber pictures with Davinci AI and after a while one guy posted AI created images with the same faces :D
Specialization will be mostly uneeded when it comes to Design and Development. We shall be like Chefs. They don't grow or process the ingredients, but a good one knows how to cook something good with them. How good is what you can put together, be it a Website, Game, Experience, Presentation, Explainer, Viz...etc.
the real reason (and real issue) is not 3d artists losing their jobs, it is (as has been for many years now) corporatives being reluctant to hire young workers for minimum wage to reinforce the department 's experienced artists, they feel like overworking two artists for something more than minimum wage is better than paying minimum wage to ten graduates. Sure the product quality degrades but they don't care about the goal, they care about their pockets.
Until it can design useful, practical, 3d printable robot parts - it's far from over. But once it can do that, I'll happily let it take my job, and build me my humanoid robot companion that I can just download, print, and assemble without having to think too much about the specifics.
What we need are tools for artists. Tools we can use to reach a PRECISE end result. We don' t want "image generators" that make "images" we want one that we can work with to make A PRECISE, EXACT IMAGE. Art is all about something being EXACTLY what you want it to be. I hope AI people will work with Adobe and the Zbrush etc people to make software that can truly assist creatives to become powerhouse artistic factories all by themselves. Until then AI of this sort will be a toy... or something for Chinese studios to pump out trash with.
You didn't actually show us the topology of most the meshes. I'd definitely like to see that, especially for the organic models where topology really matters.
lol as a solo dev I WISH ai would take over. I swear the amount of money I can save would be insane. Alas, most ai stuff is iffy and crappy. Looks good at first glance and horrible once you actually dive into it. 3d artist job can never be replaced. Just like how oil painting was never replaced no matter how much people say photography has replaced them. Ai just another type of medium people going to use and it will be good one day, but it won't replace existing jobs.
Everyone arging that currently AI generated models/images/sounds/code/whatever are shitty, should keep in mind that those are the first versions. AI improves every month. In my opinion, the only jobs that are safe are those where taking the initiative is required.
In the future we likely won't even need to generate 3D models like this. We will have realtime AI render engines which will be driven by control rigs and simple geometry. So the AI will basically be reskinning simple geo and animated rigs with whatever you want. We won't even need to animate as you will be able to just insert key poses and let the AI solve all the animation inbetween, Before any of that happens though they really need to solve consistency issues from one frame to the next.
AI doing 3d assets and 3d in general is not a bad thing. Perhaps we need to redefine artistry with AI support.. As soon as AI starts to generate beyond AAA game quality assets, then we have to talk about how we redefine artistry with human direction. It will send shivers down all creative industries like movies and video games alike.. Perhaps humans will become directors instead.
@@AlexeyTutorials Not that easy with complex characters made of several pieces in different shapes and sizes needing UDIMS. Making UVs is a waste of time. If you could just select the object with several pieces and click a Robot AI tool that does all the work judging the objects relative to the size and shape to figure out the seems and arrangement... man, it would be amazing from a modeler/look developer standpoint, you would only need to focus on the creative part. All the mechanic-repetitive work should be the focus of AI tools.
All those who naively and proudly repeat that AI will never replace humans in their jobs will be in for a rude awakening in just a few years. Those who claim that only the underperformers will lose their jobs might be surprised as well.
You will need it if you want to know how to use AI The next 3D jobs will require knowing how and why to do certain things Rather than how to extrude a polygon to match a picture Examples: color theory, concept art, silhouette, shapes... Basically what makes the art look good, and/or suitable for whatever you're working on
I can easily claim that in a few years, thanks to AIs integrated into many game engines, coding etc.. many things will become simpler and we will see many games made with AI support on the market we are simply moving towards a future where everyone can even make games
In It's current state, it can't create production ready characters. That topology is abysmal and would be a nightmare to rig. If you're a beginner and are using it to prototype characters that you will recreate yourself, I think that's alright. I think you're skipping over a lot of the creation process, but if you're going solo, you'll learn faster by spending more time remaking characters, rather than designing them from the ground up. I still think It's good to know the entire workflow though, incase you work with other people at some point.
ALRIGHT. Ok, let's just say the AI is perfect. It creates beautiful ring loops for amazing deformation for animators. The shapes are exact, no clumping of polygons because it doesn't quite know what it is. The textures aren't hallucinated and they're consistent, sharp and exact. They're production ready models with perfect materials and perfect deformation, even the static objects have wonderful topology. The ONLY thing left is how petty and pedantic our art directors and our inner thought critics are. There's no way around it, the models will never capture exactly what the visionary is looking for, they will always settle with whatever random seed of AI generation comes up with. They like everything else but the mouth is wrong, the nose isn't right. The ears needs work. I want different leg types and a specific, iconic type of claw or spike on this character. The clothing is wrong, it needs stitching in specific locations and tears over particular muscles. The clothing isn't matching the fantasy universes aesthetic properly. I want specific scuffs in these locations. The hair doesn't look quite right... These very important preferences above are just one of millions of different choices the hundreds to thousands of people will make regarding the final output. Even for a solo artist will spend literally weeks refining and perfecting the prompt until the model is matching the vision. The human need for perfection will be what keeps the AI from this total domination. The artists will use the AI, the producers will use the AI themselves but ultimately the new jobs will be created from being extremely great at deciphering the vision of the team, communicating it effectively and being able to get the AI to produce a close enough result to save the artists time so all they need to do is make adjustments.
@@AlexeyTutorials I don't believe it will do it right. I assure you this pipeline will likely be taken on in 10 or more years time when it's battle tested by indies and experimentation. While large companies do want to save money, like everyone does, people also strive to make their best work and iconic design choices are so important. It'd be great if the AI could do it, and with the new models including reasoning process in their calculations it will get even better. But humans will be human, and we will only be as happy as the money allows us to be and most artists will just have to settle with what the AI affords. I think since you use this, and you know what you're doing, perhaps you should make a new video using concept art from someone's artstation account and produce as close as possible a 3D model from that. But don't forget, the current state it's in now will likely have mediocre topology for deformation purposes. Which is why the users keep using robots and golems. It hides the deformation issues better. As for clothing, well that's just going to be quite tricky. Since this spits out a full mesh, the concept art needs to be naked, then the clothing needs to be done after with cloth physics applied and re-sized to fit the character.
I dont understand why AI is so scary to ppl. The methods of creation will change but ultimately its a tool to quicken your work. I've been a 3d artist all my life, i've tried Ai stuff and its nothing close to something i can use for a professional client project yet. The results are just inconsistent and making changes gets tedious without regeneration its impossible. However, tools that generate useful scripts and textures are very helpful. I would say instead of cursing it, embrace it as a change and adapt otherwise just loose your job to a smug, non artist who enjoys prompt painting. 🎉
I see a lot of coping down here in the comments. The initial generation is really the only part of the process that's novel. Everything else is basically just automation of long existing technologies. So, they've already crossed the highest hurdle. The refinement will come lightning fast on its heels. I have a lot of sympathy for people whose livelihoods are gonna be impacted by "AI" but reality is a cruel mistress. No use in denying her.
i see ai art bro like nft bro you can say whatever you want you can't buy me with that bulshitt all this 3d ai look generic and you never see them with a animation that don't move too much in what world would fire people that can do 3d to people that just generate think by a robot with no animation skill
Without proper topology, mindset, and sight, you ain't have sh*t, so A.I. is not going to replace all 3D artists, it just means you as a artist have to use new methods an A.I. could not do alone, bam, you just made yourself a asset again, and not one your designing lol.
Man, I didn't expect the actual topology on the generated models to be that good. I wonder if someone can do this to make a locally run retopology tool for free. Most options I checked are paid and retopology is the most manual and arduous work I don't want to bother with.
What most of these kind of content show how fast AI can generate random images/modells but the industry not works like that. Anyone who worked at a studio knows the process of iteration is really important. We start from something (it can be an Ai generated image) and arrive to the final result iterating on the task till it satisfies the needs of the art direcion. This is what AI the worst at. It can't change the result at the pixel level which makes it useless in production.
The problem with this is it cant precise follow the art direction and also all style will look similiar . Also sometimes it can provide wrong topology for animation for example muscle movement may not look correct on human character and there is also clothing simulation for realtime in games to 3d model, animate and simulate as well
My dude 2d artits are not done for AI just doesn't do the trick even for concept art. Dudeds get like 3 pages long detailed descriptions of what they need to draw AI Is not doing that. People got fired because the industry is in a bad shape in general.
All very nice, but only someone outside the industry could call these assets USABLE. I’m convinced there will be further improvements in the future, but if up until today NO ONE (and I repeat NO ONE) has managed to automate RETOPOLOGY on a hand-sculpted mesh, then I will remain skeptical for a long time.
Uh-oh, that's not a good sign that 3d modelers and 2d/3d artists are getting attacked by A.I!🥶 God, I hope 3d modelers and 2d/3d artists decides to fight back against A.I! Because I'm becoming a indie game developer, and when I get that pc gaming machine (eventually), I'm going to install Blender, and start making games independently by 3d modeling.
Even if AI could make perfect images and 3D stuff. I would still spend my time studying things like human anatomy, and art fundamentals to sculpt and draw what I wish. I enjoy it and find it rewarding in many ways. But I get how some would fear loosing their bread and butter. Remember you probably didn't learn art because there was money in it by training for decades. You were drawn to it and found out that you could get paid to do something you would do in your spare time😊
Eventually, it stands to reason that AI will replace all jobs on the planet, once we have androids capable of physical tasks, which we are already starting to see. So optimally this would just be the end of capitalism and everyone just does whatever they want whenever they want while the ai and machines handle busy work. People who want to create can still do so - but no one would have to work for a living. Machines will run farms, schools (if even still needed), construction, restaurants, entertainment, news, etc while humans devote all their time to doing whatever they dream of (though their impact will be less because anything people create or contribute will be compared against the work of AI). Basically just a twisted utopia of sorts until the machines take over. Of course, this will never happen because it's easier to generate "creative" ai than it is to produce machines to do manual labor, so companies will just milk profits by getting rid of all human creatives and reduce the only jobs left on the planet to physical labor. Because $$$. Alternatively, we may see ai-produced products begin to be associated even more so with lazy, hacky work once the novelty wears off, much like how many movies with complete CGI sets and characters are often maligned when compared to movies using entirely practical effects of old.
Thank you so much for this wonderful video. As a professional 3D artist since 1996 i'm happy to see everything evolve. I've Gone from 3D Studio r.4, via 3DSMax, touched Maya and since 2016 with Blender, and since 2023 with StableDiffusion, Runway, Flux, Kling etc. keep 'em coming. Get ready for ASI in 2026. 🤘😆🤘💻
the fall is only beggining, sure we use 3D assets, sure it's faster and make the workflow faster, yeah we still need to retopo and retexture, and even sometimes retake the rigging, but it's just a matter of time until you get all of this done near 99% just by a prompt and a button, i see the good and the bad of AI and its uses, i feel the one who took years and years to build up skills in 3D modeling and be like " damn all these years and struggle and now i just need to prompt and press " i feel the one who enjoy the fact that " damn, that's amazing i can do more projects faster and in good conditions " , but there is a thing, it makes less and less challenge, as an artist in 3D and CGI even in lyrics and music creation i crave that feeling of challenge to overcome, a difficulty that make it harder, a reason to think it out and struggle for it, we need that feeling as human being to feel the evolution and the steps we make farther and farther, the way i see AI in this domain of artistic creation is like " using Fire to burn the house instead of using that same fire to cook the meals "
no no! we should not let that happen , this is totally an overkill and should be stopped, yeah this sh..t should be stopped. this is taking people's jobs and lifes
@@Sul_Shadw I am a 3d artist. and an amature game dev also. yes I totally agree with u. because the art has to be drawn. also it is art.. and art is something that has to be different among other pieces, machines cannot produce art as humans do..never
there is not that much need 4 3d artists in the industry as it is 4 technical art. Jobs u see on artstation its hard to get in, they only looking 4 seniors with at least 5 yrs pro experience
It can pretty much do the topology of known objects and their derivatives, but I've tried everything else and the result has always been a topologically awful mesh that you can't do anything about except delete it.
How can you call that a "good topology"? Looks like shit. No self-respecting game dev would put that garbage directly in a game engine. Also: 3D SCANNING EXIST AND WILL ALWAYS BE MORE ACCURATE THAN AI SLOP! 3d scans have been part of the process for many years now. What exactly does the ai slop offer compared to them?
Whats the problem with that robocat topology? Also, 3D scanning requires having an access to the real object or at least hundreds of photos of it for photogrammetry, but AI requires only single concept image, can’t you see the difference? 😉
@@AlexeyTutorials There is market places where you can buy scans, there is also home scan setup that will give a good quality scan in less than 5 minutes. That's why I said budget option.
