because it’s only good at improving upon things lot of people have a recorded history of doing. So yes it would be more accurate at things more widespread and loved :)
I like rigging 😂 AI helped me to create custom scripts to make the process so smooth and satisfactory... Obviously you need to know what you need and the steps to take. No black magic whatsoever.
@@WyvernZu It doesn't look great, but you must know by now that if they keep working on this it'll eventually become an incredibly useful tool, right? Why are people so short sighted when it comes to AI?
Seems very useful for generating base meshes for sculpting. Instead of generating a whole character u can generate each part separate , leg , arm , torso , head.the end result is a lot more detailed
@@FreakazoidRobots you get a lot more detail , try making just a male torso without arms and or legs and the anatomy and detail are pretty impressive , esp when u see jusy the mesh , no texture. Vs a whole body its just a general human figure with some textures but no detail when it comes to the actual mesh .
@@FreakazoidRobots the models being generated seem to be capped around 10k -12k polys. So i can Get a 10k poly character or 10k on just one body part with decent anatomy and detail. So yeah it is kinda kit bashing, but free, custom and actually kinda fun.
And once again, it is a system without source code which you can't run locally, which wants your login information, and definitely wants to become a SaaS.
If you can run SD Zero 123 go for it, but it right now the open source solutions need more VRAM than I have available. Considering that 24GB VRAM is pretty spendy, I think SaaS is the only way to go for now.
@@travissmith5994 I think the point is it's a toy that someone wants to keep ownership/control of so they can sell out to some other huge company, rather than truly contribute to the world's knowledge.
@@timmyanimations8321 🙄 No one could possibly store a 10GB file on a personal computer... Pruned AI models are smaller than modern video games. VRAM is what makes them hard to use locally, not disk space.
@@Akumei_1 it doesn’t matter. It is price to quality. If I can get bad quality for free. It beats good quality for 3k per 30 second advert haha. Especially when you can get a primitive model, do the animation, and reskin with control nets through something like flux. That will turn it into a high quality animation. More the control nets and prompts become controllable and predictable. The more it pushes artists out.
The real milestone will be when it can do things like hands, faces, knees, etc. For hard surface models like an apple, topology doesn't really matter very much anyway.
I generated some human skeleton parts and other basic objects, and the results were absolutely atrocious. That said, the meshes could serve fairly well as reference images for amateurs like myself. I generated a skull, and got a 25K vertex blob. However, there was enough of a visual suggestion of the basic skull shape, features, and ratios that I can use it to create a 3-view reference sheet for use in creating a useful skull.
i mean the writing is in the wall for a while now, probablly some basic drawing and storybording is the best place to invest your time if you want to create stuff, things that help communicate the end result you want, those people that studied art history will all of the sudden be usefull.
I saw twitter threads crapping on this tech just yesterday. Having seen how quickly text to 2D generation has improved and knowing how much more money is in the AI space now than there was just ~1 year ago when that was being done (literally orders of magnitude more, and more $ = more gpus to train on), I fully expect text to 3D to improve at a similar pace, even being harder to do. People saying this looks too blobby or the retopo is bad etc are missing the the point. This literally didn't exist not that long ago and now it's where 2D generation was roughly a year or so ago. Its getting better and will continue to do so. I don't think its replacing AAA character modelers anytime soon but acting like its dumb/useless seems super short sighted. I agree with you that it's coming (whether people want it to be or not) and I think knowing about it is better than people sticking their heads in the sand, so thanks for showing where it's at currently in this video.
Just tested it myself. I had it create 2 things I've recently created to compare my stuff to its. First a heart shaped pink waffle. All 4 looked like something that came out of a toaster from 1987. 2nd a lemon, which many had 3 green nipples and were a horrible shape. So far it's hot garbage stuff. But I'm sure eventually it will kill us off.
As an environment artist, I don't see it as a threat. I see it as a very useful tool. With the current quality and lack of artistic control, it's still good for small assets, base meshes and baking. I can imagine using it for baking foliage maps, like a cherry tree, instead of modeling the cherries just use AI. In the bake you won't even see the difference.
After Inspecting a large amount The uv map is not very well defined, but it's great to prevent anyone to edit it afterwards. No Symmetry 1 for 1 mapping The quads were good but the symmetry was not there across the usual dimensions Sometimes errors in geometry would appear No consideration for stretching in animation
@@-Tayden It's coming along for sure and taking over at different stages has benefits. I want to see a rigging and animation sequence feature based on the mass of mesh muscle shapes
@@bonecircuit9123 why wouldnt i downvote such an outragous comment? .. so becasue it doesnt create hollywood vfx quality models for muscle system rigging , its shit by default ?
Well, I have tried it out, it is very rough on the finished model, so, until they get it to do better, we have a small reprieve! It might be usable for a background object viewed from a distance.
Glad you seem to have comments back. I totally get not chasing views and interactions - had that whole deal myself ages ago. But its good to be able to ask questions and that kind of thing.
Would be fun to run this in Blender directly in a while, as Stable Diffusion did with seamless textures in Blender. Krita has become really good at Stable Diffusion and Control Net last months. Fun stuff.
just the first iteration; i thought it would take a bit longer to even get to this point; im not worried now, but this stuff will advance so incredibly fast... guess its time to look into a different career.
I use it. I have a room and I want to populate it with props. I get those props from this tool and then model on top of them. I helps me speed up the process of creating new props for a new scene.
I would prefer an auto rigging and auto weight painting auto remove intersecting meshes non mandibles ,limited dissolve auto retopology and instant baking stuff . Or some kind of auto lod system you create a high poly and it makes different low and mid polys of the mesh though some calculations. If it could generate a basemesh for unreal model , mixamo or something then it could work
oh wow this seems very interesting, ai is evolving so fast just recently i saw one that can turn a image to video but this one seems more useful for game developers and 3d designers
Oh, yeah, right. Image to 3D has been around for 5 or 6 years. And it's still abysmal, horrendous-the worst thing ever, even for generating a high poly with terrible topology. Text to 3D seems even worse. I've been following NVidia (love their audio2Face), and it was the most promising... and still crap. I'm not a 3D artist, despite having used it for 15 years (in animation, ads, and comic books). So, I'd love to have a program that could automatically translate a character sheet into a 3D model. But nope, there's no such program, and I doubt it'll exist in the next 7 years. The same goes for 2D. I'm a comic book artist (my bread and butter), but I couldn't find any AI or AI+addon combinations that could genuinely help the process. AI still can't even draw straight lines on a goddamn building, for example. AND I've been complaining about the same issues over and over since the beginning. Nobody fixes this. I'd like to speed up my personal work to turn it into an IP, but no. Until now, AI is freaking useless for business. So, don't worry. AI won't steal our jobs. It's WAY TOO stupid for this.
@@kusog3 Yup, agreed. It depends on the goal. For industry standards, it's not good enough (not even close. Even my agent didn't like any results, mine or others. No editor gave a green light for this too. I agreed with them. Too amateurish results). But, for a person testing outside the entertainment industry without any intention to do anything serious, or just for fun, or even for a funny mobile game or webseries, it's completely good enough. :) "As for whether it can steal jobs, i don't really care about that part" Agreed. Job positions vanish and others are created. People just need to care about staying sharp to adapt to the environment. Industry == Jungle.
"Too amateurish results" Um...I don't know which one you are using but it certainly isn't the right one or you aren't using it properly. Use midjourney v5-6 or Krita with Stable Diffusion plugin.
I am a 3D Artist in the gaming industry since 15 years and i see it as a gift not a threat. It takes away the boring tasks. Feels like photogrammetry in the early stages
@@teka5509 haha have you ever worked in the gamesindustry? thats not how it works in reallife. AI is just a tool and what ever comes it will just speed up the workflow. I worked on AAA Games most of my life and thats what most of the industry thinks. Nothing to fear. The only fear there is, is that the young generation will die out. The bummer is, we wont use it until the legal question is solved
@@davebohm464 Games created by AI will come pretty soon, enjoy your 2-3 years that you have left in the game industry. I work in other 3d fields which soon will be gone too.
Those results are insane. I remember when AI generated 3D models still looked like absolute nightmare fuel like it was yesterday, and all those objects in the background look pretty much spot on. It seems to still have a hard time with complex objects like characters or cars but that's only going to get better. I expect text-to-3d to be pretty much solved within the next 4 years.
I'd really like to see the 3D model equivalent of inpainting at some point. That way when we get outputs that are almost perfect but have a few flaws, they can be fixed. Maybe it could be implemented by letting the user use a shape as a mask by placing it over a section of the model, or highlighting a group of vertexes to modify.
Ughh I better not tell my friend about this I have been working a simple horror game with a friend of mine( He is a software engineer/ who also knows blender) if he sees this he will give up and not wanna work on it anymore...... 😂😢
For anything more complicated than an apple the results are not great, though can still get some reference ideas from what it generates. It's impressive that AI is progressing this fast and hope to see improvements to this in future.
No, I don't think so. The capability of those text-to-3d generators will evolve with time, and in a year or two they are going to be noteworthily more capable than this current iteration, similarly to how AI generated images have improved from Midjourney version 1.0 to the current iteration. Seriously, look it up if you haven't yet seen how far those have come already. I'd reckon proper 3D models are a bit more complex than 2D images when it comes to how much data you need to train a model, but I believe that, given more time and more powerful computers, the text-to-3d generators will also improve drastically. Give it a year or two, this is only an early preview of what's to come.
AI can only imitate. It's not capable of creating. All of these 3D models needed training data from somewhere. Plus a lot of AI companies are facing lawsuits due to copyrighted works being used as training data. Midjourney is in a lot of trouble since their CEOs were caught compiling lists of famous artists to scrape their work.
I really don't know what the fuck to do, I am 34, only recently got a job as a 3d artist after having done freelance for a decade, still in debt from my education, put all my eggs in the 3d graphics basket (originally wanted to do 2d, but did 3d because it seemed like a solid option for employment) and soon I will be utterly useless in terms of employment, this shit makes me so fucking depressed
Nahh no worrys this AI 3d artist will take the cheap clients no one wants. Its like mc donalds exist on side with gourmet kitchen. The cheap hunters go for the trash and the people with class and money go for gourmet. Same with art
i put in lich lord , lich god , crowned beast ... and the results were lowres and shabby but the overall designs were amazing .. really good for creating basemeshes for sculpting or for drawing comics too
Just a matter of time before we can see perfect models out from AI. (I remember decades ago I saw Existenz a movie made by director D. Cronenberg, it was about a reality were people spend most of they time in games, and games creators where the stars, they were just writing complex story lines, I imagine that the creation was done by AI (in the movie no big game studio was mentioned).) Well it's sad, but eventually AI will bring more freedom of creation without the hassle of technicalities.
Why do you turn comments off for all videos? I can't ever get help thru the comments because of it.... I'm very used to most videos having some comments indicating what's changed in the case that a node is removed from blender, or if something changes between versions... Either way, thanks for all that you do with these and I hope you understand where I am coming from!
Just remember, these look like Midjourney generations looked a year ago ! Give it some time and it will spit perfectly made models in seconds. I think we´ve got to rethink our jobs...
I’ve been explaining this to my colleagues for almost 6 years now prior to the AI boom that we were already seeing early signs that asset creation (2d and 3d) would have little to no demand. I think we had those discussions early enough because looking at what leads at blur studio who left were saying “we are sculpting less and less and are just moving parts around”. Now I believe it’s coming to the point where bigger questions should be asked by the individual. “Why are we making art?” “Is income the largest stressor when it comes to seeing AI take our work”. Things of that nature.
Ever since the AI boom began to pick up, I keep hearing word about tech companies firing half of their staff/employees. I am sure glad I decided to go to school to become a millwright instead of an IT guy! (I was thinking about going into computer stuff at first)
They were gonna fire people, AI or not. It's got more to do with the economy than with AI. It's in a 'bust' right now. Just a normal part of the craziness that is capitalism.
@@szkrukowski I doubt that would happen, but if it did, physical labor would definitely still be practical since "robot construction & maintenance" would become a very in-demand job.
@@Lar_me Lol it's been happening for decades and it's going to accelerate. You don't need a lot of people for maintenance, and besides, you can just make a robot to handle most of it.
I always download fbx and import it to my empty project Instead of using a ready blend file because those tend to be too customized by other person and messy for my use
Soon you can do "text to 3d program" and combine houdini/blender/softimage ice to one super app. Then of course you will not need it since ai does all them task anyways.
They took our jobs… But really I hate this time right now. Everyone thinks they are some artists just because the write a prompt and this makes me angry!
And whos going to use this tool? Thats right, 3d artists. This AI thing is just a tool, sure everyone can use it but a professional will always produce a better result, AI merely speeds things up.
I think AI wont be used in big companies because there's the regulation thing going on to protect artists whose art was stolen and companies don't want to get sued for that. Will definitely be interesting how AI's gonna change the art world
Not goid enough to be considered as a tool or a resource, fix things is harder and take more time than create a new one, I tried of course for me time is money, but atcless at this point in time AI is very far to can even be considered a tool, we already have GOOD tools available, CC, iClone, MetaHuman, Blender geometry nodes, many generative add-ons, huge 3d assets libraries with hundreds of thousands of models and textures. Sorry but they came TOO LATE to the party that at this point I will not use their stuff not even for free.
I'm not sure I'd incorporate this into my workflow yet for anything more complicated than an apple... but it looks like the time when it could actually replace me is coming fast. I kind of don't mind. If I can make anything I want in a few minutes, I won't cry about it.
❌i dont develop ai but i would assume 3d ai is far more complicted then 2d plus 2d art generation has been around for atleast a couple years now and it hadnt replaced artist and most people hate the people who use it@@joelt7869
I dunno, guess it's me, I must be too old or something, but I just don't like this trend to all these web based apps like this. I'm used to old fashioned locally installed software that I can continue to use offline. yeah, it's free, but that' snot my issue. It's the web/network dependency and whatever they are doing on the other side. So much centralization. I don't think it's good. The world is worse now.
ngl, if they did auto rigging or auto uvw mapping they'd be hailed as a godsend.. but no. ai only does what people love doing.
I like UV unwrapping :)
because it’s only good at improving upon things lot of people have a recorded history of doing. So yes it would be more accurate at things more widespread and loved :)
It does auto rigging since years ago, its not the best, but it does the job.
I like rigging 😂 AI helped me to create custom scripts to make the process so smooth and satisfactory... Obviously you need to know what you need and the steps to take. No black magic whatsoever.
@@kunemann Click this text box to prove that you are human. Lol. Cheers mate.
One thing is for sure, I'm NOT cleaning up those things.
This! Only someone with zero experience with 3d programs would look at those melted models and think its a good idea to fix it.
a cleaning up AI will cleanup
I told it to draw me a machine that turns peanut butter back into peanuts. common problem
Reminds me of the Earthbound joke where a kid tries to invent a machine to turn hard-boiled eggs back into regular eggs.
next u'll hear they trained it using sketshfab and used them with no permission lol
This seems great if you want to create molten clay figures.
The apple looked great. I’d image it’d eventually get better at bigger objects too
Hope we'll see this comment again in 5years..
@@kidtoucher302 You think it looks good?
@@WyvernZu It doesn't look great, but you must know by now that if they keep working on this it'll eventually become an incredibly useful tool, right? Why are people so short sighted when it comes to AI?
i knew that moment will come sooner or later
Seems very useful for generating base meshes for sculpting. Instead of generating a whole character u can generate each part separate , leg , arm , torso , head.the end result is a lot more detailed
How does separating out the body parts help?
@@FreakazoidRobots you get a lot more detail , try making just a male torso without arms and or legs and the anatomy and detail are pretty impressive , esp when u see jusy the mesh , no texture. Vs a whole body its just a general human figure with some textures but no detail when it comes to the actual mesh .
@@4D_art
Interesting. At some point, I'm not sure what the difference between this and kit bashing are, though.
@@FreakazoidRobots the models being generated seem to be capped around 10k -12k polys. So i can
Get a 10k poly character or 10k on just one body part with decent anatomy and detail. So yeah it is kinda kit bashing, but free, custom and actually kinda fun.
And once again, it is a system without source code which you can't run locally, which wants your login information, and definitely wants to become a SaaS.
If you can run SD Zero 123 go for it, but it right now the open source solutions need more VRAM than I have available. Considering that 24GB VRAM is pretty spendy, I think SaaS is the only way to go for now.
@@travissmith5994 I think it's easier to run this thing from Google Collab
@@travissmith5994 I think the point is it's a toy that someone wants to keep ownership/control of so they can sell out to some other huge company, rather than truly contribute to the world's knowledge.
There's no other way to do it really. The AI needs a ton of storage to know what all the words mean, so no one could just download it on a PC.
@@timmyanimations8321 🙄 No one could possibly store a 10GB file on a personal computer...
Pruned AI models are smaller than modern video games. VRAM is what makes them hard to use locally, not disk space.
These models can be used as a base for further modify them. I'd cut them in half, use a mirror modifier to guarantee symmetry and start sculpting
I don't care what "AI" can do, it is still dumb as heck. I don't want it to replace the fun part of my projects
Adapt or become homeless.
@@zachary3603 these models are extremely primitive
@@Akumei_1 it doesn’t matter. It is price to quality. If I can get bad quality for free. It beats good quality for 3k per 30 second advert haha. Especially when you can get a primitive model, do the animation, and reskin with control nets through something like flux. That will turn it into a high quality animation. More the control nets and prompts become controllable and predictable. The more it pushes artists out.
@@zachary3603 you would need tremendous clean-up, which actually would take more time than making it from scratch. and probably cost you more. lol
@@zachary3603 the models are awful, you'd probably have to wait 10 years to see any type of growth.
I was shocked by how decent the topology is now. Last time I saw it topology was messed up. Now it's looking ok.
The real milestone will be when it can do things like hands, faces, knees, etc. For hard surface models like an apple, topology doesn't really matter very much anyway.
I generated some human skeleton parts and other basic objects, and the results were absolutely atrocious. That said, the meshes could serve fairly well as reference images for amateurs like myself. I generated a skull, and got a 25K vertex blob. However, there was enough of a visual suggestion of the basic skull shape, features, and ratios that I can use it to create a 3-view reference sheet for use in creating a useful skull.
Probably have integrated a quad remesher pass in the background.
Cause it’s all stolen
@@USBEN.I was thinking the same. So nothing actually incredible
" The time has arrived'' . lt was a great run .Guess I will become a monk ,live out rest of my life on the montains.
I love this stuff. It's like it gets it to 80% so we can do the rest. And for background assets that are far away or blurred it's perfect.
i mean the writing is in the wall for a while now, probablly some basic drawing and storybording is the best place to invest your time if you want to create stuff, things that help communicate the end result you want, those people that studied art history will all of the sudden be usefull.
I saw twitter threads crapping on this tech just yesterday. Having seen how quickly text to 2D generation has improved and knowing how much more money is in the AI space now than there was just ~1 year ago when that was being done (literally orders of magnitude more, and more $ = more gpus to train on), I fully expect text to 3D to improve at a similar pace, even being harder to do. People saying this looks too blobby or the retopo is bad etc are missing the the point. This literally didn't exist not that long ago and now it's where 2D generation was roughly a year or so ago. Its getting better and will continue to do so. I don't think its replacing AAA character modelers anytime soon but acting like its dumb/useless seems super short sighted. I agree with you that it's coming (whether people want it to be or not) and I think knowing about it is better than people sticking their heads in the sand, so thanks for showing where it's at currently in this video.
It also wouldnt have existed with out all of the scraped data from websites like sketchfab.
Just tested it myself. I had it create 2 things I've recently created to compare my stuff to its. First a heart shaped pink waffle. All 4 looked like something that came out of a toaster from 1987. 2nd a lemon, which many had 3 green nipples and were a horrible shape. So far it's hot garbage stuff. But I'm sure eventually it will kill us off.
Someone must stop the AI developers
Yeah It’s inevitable, but hey. There’s always a chance this’ll just make our lives easier
I honestly wonder if in 10-15 years, we'll look back at the 2020;s and laugh at how everything needed AI for some reason.
1-2 years*
As an environment artist, I don't see it as a threat. I see it as a very useful tool. With the current quality and lack of artistic control, it's still good for small assets, base meshes and baking. I can imagine using it for baking foliage maps, like a cherry tree, instead of modeling the cherries just use AI. In the bake you won't even see the difference.
Indeed, been using and customizing tree generators just to have some variety. If 3D AI can help with that, that would be nice
Sometimes the luma textures suck, but if you screenshot it -> magnific it, reproject that texture, looks way better.
the nuke goat, i tip my hat to you
After Inspecting a large amount
The uv map is not very well defined, but it's great to prevent anyone to edit it afterwards. No Symmetry 1 for 1 mapping
The quads were good but the symmetry was not there across the usual dimensions
Sometimes errors in geometry would appear
No consideration for stretching in animation
Retop was quick though after re raster
@@-Tayden It's coming along for sure and taking over at different stages has benefits. I want to see a rigging and animation sequence feature based on the mass of mesh muscle shapes
@@bonecircuit9123 why wouldnt i downvote such an outragous comment? .. so becasue it doesnt create hollywood vfx quality models for muscle system rigging , its shit by default ?
Um ok mate
@@alexmehler6765
Well, I have tried it out, it is very rough on the finished model, so, until they get it to do better, we have a small reprieve! It might be usable for a background object viewed from a distance.
Glad you seem to have comments back. I totally get not chasing views and interactions - had that whole deal myself ages ago. But its good to be able to ask questions and that kind of thing.
Would be fun to run this in Blender directly in a while, as Stable Diffusion did with seamless textures in Blender.
Krita has become really good at Stable Diffusion and Control Net last months.
Fun stuff.
FFS making all the super rich richer and putting all the little people out of work.
F Knows where all this is heading.
just the first iteration; i thought it would take a bit longer to even get to this point; im not worried now, but this stuff will advance so incredibly fast... guess its time to look into a different career.
Cinematic ahh video
Soon enough people will be able to make their own 2-hour long action movies
in less than 5 mins
😐
I use it. I have a room and I want to populate it with props. I get those props from this tool and then model on top of them. I helps me speed up the process of creating new props for a new scene.
I would prefer an auto rigging and auto weight painting auto remove intersecting meshes non mandibles ,limited dissolve auto retopology and instant baking stuff . Or some kind of auto lod system you create a high poly and it makes different low and mid polys of the mesh though some calculations.
If it could generate a basemesh for unreal model , mixamo or something then it could work
oh wow this seems very interesting, ai is evolving so fast just recently i saw one that can turn a image to video but this one seems more useful for game developers and 3d designers
Yeah honestly I welcome it. Hopefully it keeps improving
Oh, yeah, right. Image to 3D has been around for 5 or 6 years. And it's still abysmal, horrendous-the worst thing ever, even for generating a high poly with terrible topology. Text to 3D seems even worse. I've been following NVidia (love their audio2Face), and it was the most promising... and still crap.
I'm not a 3D artist, despite having used it for 15 years (in animation, ads, and comic books). So, I'd love to have a program that could automatically translate a character sheet into a 3D model. But nope, there's no such program, and I doubt it'll exist in the next 7 years.
The same goes for 2D. I'm a comic book artist (my bread and butter), but I couldn't find any AI or AI+addon combinations that could genuinely help the process. AI still can't even draw straight lines on a goddamn building, for example. AND I've been complaining about the same issues over and over since the beginning. Nobody fixes this. I'd like to speed up my personal work to turn it into an IP, but no. Until now, AI is freaking useless for business.
So, don't worry. AI won't steal our jobs. It's WAY TOO stupid for this.
It didn't work out for you, but i'm sure it worked well enough for others. As for whether it can steal jobs, i don't really care about that part
@@kusog3 Yup, agreed. It depends on the goal.
For industry standards, it's not good enough (not even close. Even my agent didn't like any results, mine or others. No editor gave a green light for this too. I agreed with them. Too amateurish results). But, for a person testing outside the entertainment industry without any intention to do anything serious, or just for fun, or even for a funny mobile game or webseries, it's completely good enough. :)
"As for whether it can steal jobs, i don't really care about that part"
Agreed. Job positions vanish and others are created. People just need to care about staying sharp to adapt to the environment. Industry == Jungle.
wow you are a lookin a wise man :)
@@twinfrozen1421 he aint
"Too amateurish results"
Um...I don't know which one you are using but it certainly isn't the right one or you aren't using it properly. Use midjourney v5-6 or Krita with Stable Diffusion plugin.
Time to use nightshade on those LLMs.
I am a 3D Artist in the gaming industry since 15 years and i see it as a gift not a threat. It takes away the boring tasks. Feels like photogrammetry in the early stages
Soon all aspects will be ai generated, so enjoy the enjoyable tasks while you can 😁
@@teka5509 haha have you ever worked in the gamesindustry? thats not how it works in reallife. AI is just a tool and what ever comes it will just speed up the workflow. I worked on AAA Games most of my life and thats what most of the industry thinks. Nothing to fear. The only fear there is, is that the young generation will die out. The bummer is, we wont use it until the legal question is solved
@@davebohm464 When AI can create a full game by itself who is needed? Give me a list of tasks you think AI will not be able to do.
@@teka5509 at that point 95% of humanity will be unemployed. We will have bigger questions at that point then just the gameindustry haha.
@@davebohm464 Games created by AI will come pretty soon, enjoy your 2-3 years that you have left in the game industry. I work in other 3d fields which soon will be gone too.
ok i had a look at this program. what's with all this bad low res blurriness in the textures?
What's going on there?
this looks great if you like garbage
Those results are insane. I remember when AI generated 3D models still looked like absolute nightmare fuel like it was yesterday, and all those objects in the background look pretty much spot on. It seems to still have a hard time with complex objects like characters or cars but that's only going to get better. I expect text-to-3d to be pretty much solved within the next 4 years.
let's be real. you can photoscan a youtube video of whatever thing and get comparable results :) So far we're safe
Temporary 😅
so it's either computer vision or generative AI
We are not safe bro
I never thought I'd be alive to witness what is essentially the emergence of holodeck style technology. (sans energy to matter)
"nice topology", "a russian tank"
I'd really like to see the 3D model equivalent of inpainting at some point. That way when we get outputs that are almost perfect but have a few flaws, they can be fixed. Maybe it could be implemented by letting the user use a shape as a mask by placing it over a section of the model, or highlighting a group of vertexes to modify.
your new lense is fantastic
thanks, there goes my evening!
Ughh I better not tell my friend about this I have been working a simple horror game with a friend of mine( He is a software engineer/ who also knows blender) if he sees this he will give up and not wanna work on it anymore...... 😂😢
syncproblem detected :) love your vids anyway :)
For anything more complicated than an apple the results are not great, though can still get some reference ideas from what it generates.
It's impressive that AI is progressing this fast and hope to see improvements to this in future.
3D modellers look out! theyre coming for you!
5 min later: Impressive but also not impressive.
Pretty sure we'll be just fine, guys.
No, I don't think so.
The capability of those text-to-3d generators will evolve with time, and in a year or two they are going to be noteworthily more capable than this current iteration, similarly to how AI generated images have improved from Midjourney version 1.0 to the current iteration. Seriously, look it up if you haven't yet seen how far those have come already.
I'd reckon proper 3D models are a bit more complex than 2D images when it comes to how much data you need to train a model, but I believe that, given more time and more powerful computers, the text-to-3d generators will also improve drastically.
Give it a year or two, this is only an early preview of what's to come.
@@TheGeefriendAI will never replace the professionally employed
AI can only imitate. It's not capable of creating. All of these 3D models needed training data from somewhere.
Plus a lot of AI companies are facing lawsuits due to copyrighted works being used as training data. Midjourney is in a lot of trouble since their CEOs were caught compiling lists of famous artists to scrape their work.
@@Whalester It will
@@teka5509 You are a clown if you think it will.
This is actually a great tool to for making good looking previs and imo .
that apple looks like it got crystal meth infused, but I don't expect anything less from an AI. Iconic.
Just a reminder to everyone, but this is the worst it will ever be…
I love tutorials from Pedro Pascal
I really don't know what the fuck to do, I am 34, only recently got a job as a 3d artist after having done freelance for a decade, still in debt from my education, put all my eggs in the 3d graphics basket (originally wanted to do 2d, but did 3d because it seemed like a solid option for employment) and soon I will be utterly useless in terms of employment, this shit makes me so fucking depressed
Happens, good luck o7
We have a maximum of 2-3 years to find a different job ..
Nahh no worrys this AI 3d artist will take the cheap clients no one wants. Its like mc donalds exist on side with gourmet kitchen. The cheap hunters go for the trash and the people with class and money go for gourmet. Same with art
i put in lich lord , lich god , crowned beast ... and the results were lowres and shabby but the overall designs were amazing .. really good for creating basemeshes for sculpting or for drawing comics too
Just a matter of time before we can see perfect models out from AI.
(I remember decades ago I saw Existenz a movie made by director D. Cronenberg, it was about a reality were people spend most of they time in games, and games creators where the stars, they were just writing complex story lines, I imagine that the creation was done by AI (in the movie no big game studio was mentioned).)
Well it's sad, but eventually AI will bring more freedom of creation without the hassle of technicalities.
Thanks for the heads up. I just made a totally cursed Robocop model and I love it.
Why do you turn comments off for all videos? I can't ever get help thru the comments because of it.... I'm very used to most videos having some comments indicating what's changed in the case that a node is removed from blender, or if something changes between versions...
Either way, thanks for all that you do with these and I hope you understand where I am coming from!
im terrified
more ai sewage. yay
In 2 years this stuff is gonna be almost human like, I'm kinda hyped but also makes me a bit sad.
Just remember, these look like Midjourney generations looked a year ago ! Give it some time and it will spit perfectly made models in seconds. I think we´ve got to rethink our jobs...
I’ve been explaining this to my colleagues for almost 6 years now prior to the AI boom that we were already seeing early signs that asset creation (2d and 3d) would have little to no demand. I think we had those discussions early enough because looking at what leads at blur studio who left were saying “we are sculpting less and less and are just moving parts around”. Now I believe it’s coming to the point where bigger questions should be asked by the individual. “Why are we making art?” “Is income the largest stressor when it comes to seeing AI take our work”. Things of that nature.
Makes my life easier lmfao
I found these very usefull as a base mesh and then to add futher procedual detail for example for rocks and adding them in houdini
Fascinating to play with but blobby is the word for anything with any detail at all.
Ever since the AI boom began to pick up, I keep hearing word about tech companies firing half of their staff/employees. I am sure glad I decided to go to school to become a millwright instead of an IT guy! (I was thinking about going into computer stuff at first)
The layoffs in the the tech industry are because of overhiring during Covid.
Well it's only because the openai and tesla robots are not ready yet ;) but physical labour is next :D good luck with everythin
They were gonna fire people, AI or not. It's got more to do with the economy than with AI. It's in a 'bust' right now. Just a normal part of the craziness that is capitalism.
@@szkrukowski I doubt that would happen, but if it did, physical labor would definitely still be practical since "robot construction & maintenance" would become a very in-demand job.
@@Lar_me Lol it's been happening for decades and it's going to accelerate. You don't need a lot of people for maintenance, and besides, you can just make a robot to handle most of it.
For a reference 3d object that will be so convenient
Why did you download the fbx and not the blend file?
I always download fbx and import it to my empty project Instead of using a ready blend file because those tend to be too customized by other person and messy for my use
yo mama gonna do UV and Retopo on these shits
👑
nnnOOOOOOOOOO I JUST STARTED MY 3D JURNEY WTFFFFF BROOO THIS SHIT IS NOT FAAAIR FUUUCJ
give it 1 year
but can it do geometry nodes? i think not
give it one year
I wouldn't put a date on when it's 'good enough' at something, but give it time. AI picks up speed fast.
Soon you can do "text to 3d program" and combine houdini/blender/softimage ice to one super app. Then of course you will not need it since ai does all them task anyways.
Why is Ed Sheeran wasting his time on this?
They took our jobs…
But really I hate this time right now. Everyone thinks they are some artists just because the write a prompt and this makes me angry!
3:18 ima use this as a pick up line sometime
I swear your videos are getting wider every day
I mean i hate modelling and im terrible at it but this is kinda fucked up.
And whos going to use this tool? Thats right, 3d artists. This AI thing is just a tool, sure everyone can use it but a professional will always produce a better result, AI merely speeds things up.
I think AI wont be used in big companies because there's the regulation thing going on to protect artists whose art was stolen and companies don't want to get sued for that.
Will definitely be interesting how AI's gonna change the art world
You have too much faith in the American judicial system. It's not applicable to the whole world.
You have too much faith in the American judicial system. It's not applicable to the rest of the world.
Thx i can use them for retopology
the end is near
I tried 2 different browsers and it crashed on both when I tried to get a more detailed version - when I chose one of the four versions
It only crashed when I was behind my work computer firewall
not too worried for my life tbh
Behind scene props are becoming easy then
I surrender to our Overlords.
i can see using it for small background stuff
Can someone tell me if there is a way to get a paid version that gives unlimited prompts?
I've followed graphics since a text game called quest ... It was a text game .... the graphics got better .... 50yrs old in june
why the downvote button is disabled by owner? i upvoted the video, but it leaves a bad taste.
Not goid enough to be considered as a tool or a resource, fix things is harder and take more time than create a new one, I tried of course for me time is money, but atcless at this point in time AI is very far to can even be considered a tool, we already have GOOD tools available, CC, iClone, MetaHuman, Blender geometry nodes, many generative add-ons, huge 3d assets libraries with hundreds of thousands of models and textures. Sorry but they came TOO LATE to the party that at this point I will not use their stuff not even for free.
Capitalism never stops
Damn
I'm not sure I'd incorporate this into my workflow yet for anything more complicated than an apple... but it looks like the time when it could actually replace me is coming fast. I kind of don't mind. If I can make anything I want in a few minutes, I won't cry about it.
Definitely will become very useful and will make some peoples skills less useful.
❌i dont develop ai but i would assume 3d ai is far more complicted then 2d plus 2d art generation has been around for atleast a couple years now and it hadnt replaced artist and most people hate the people who use it@@joelt7869
I'd say a year or so
One day they will destroy turbo squid
after you can project mapping some ai image created from krea ai. into the model
Are you ok?
harrowing
Considering it can't make nsfw stuff people who make adult content are safe from ai
Wow! But I dont need this many 3-D models
why did you disable comments on older videos? it kinda sucks not being able to read what others have commented which can be useful a lot of the time
I dunno, guess it's me, I must be too old or something, but I just don't like this trend to all these web based apps like this. I'm used to old fashioned locally installed software that I can continue to use offline. yeah, it's free, but that' snot my issue. It's the web/network dependency and whatever they are doing on the other side. So much centralization. I don't think it's good. The world is worse now.