There is billions the best can earn... thats why you got massive companies investing so heavily in AI and the like, as any investment will be offset by the final product. Don't think it will work perfectly in the future, but likely for small studios, it will be a massive help, or even for people needing models to better describe or display their product or service; imagine architects making models from floorplans or a DIY programmer making a stylized shell for their project, simply using a Core model with the thread holes and air vents as a reference to be built into the core.
I loved your video. Can’t believe how good the best version was. Also you gave me an excellent introduction to tools I was completely unaware of, given my 3D printing journey only began two weeks ago. Excellent work. Thanks.
6:32 this already seems impossibly good to me wtf. It actually interprets what the image SHOULD be in 3D exactly as you'd expect. 12:50 THIS is more what i thought would be the best we could get. Blobs, random wrinkles, dudes melting into chairs, etc. I'm always astonished with how fast progress is being made in this field
really impressed with rodin, i have been waiting for some time to have good high quality 3d models from text and/or images. i will be playing with this for certain.
Thanks for testing Meshy! We've recently updated the Hard Surface Mode in Image to 3D. This enhancement significantly improves the mesh quality of hard surface models, making them ideal for 3D printing. We invite you to try it out and share your feedback!
Hey Meshy, thanks very much for getting in touch. I've loved your setup since I started using it in prep for my previous video on 3D Generative AI. I'll definitely be playing with the updated to your service and I'll be covering it in future videos. It's great that it sounds like you're actively working to improve your tool for use with 3D printing so I look forward to hearing more about that.
Thanks, very nice test and you wend through the complete process including printing the models. I learned a lot from this, and I am sure going to give these AI generators a try.
I'm not surprised they struggled with the word stetson, it's not a word we use much in the US, and if in theory their training data was mostly US -english centric then it would be a word not represented much.
At least on Cura and probably other slicers you don't need to cut objects that aren't flat to get a flat surface for printing. Simply set the orientation to the way up you want then set the z head to a small negative, any part of the model below the xy plane gets cut at that point when slicing and will have a flat bottom...yet I've seen so many people say to get round this by cutting models in mesh mixer...unless you're sharing it on for others that don't know this trick don't waste your time!
Wonderful video - again! Thank You! I love your tempo, your style, your way of explaining… It is •Just Very Good• ! And… the content is very interesting and has top value as well. Subscribed ☺️
I think you're underestimating how much it's already being used in big productions. The only difference is that right now it's still less effort to dive in the free library models to make shovelware games. That's why they're commonly Known as asset dumps. But really that doesn't matter. They're not real games. It's generally just a form of scam. Those games don't hire anyone to do anything anyways. Big studios have already done a ton of their layoffs a couple years ago in reaction to this tech becoming available to them... It's not the future, it's now. My personal hope is that we see indie become strongly empowered by these tools and can fill the gap left by increasingly uninspired big studio productions with some bigger scale gems despite not having all the resources. Because that's how the market is supposed to work. A small nimble newcomer is supposed to be able to displace large slow moving existing companies
My neighbour asked me the other day "Why would I spend money on any AI-created product when I know I can generate mine just as easily?". Watching this video, those words are starting to sink in. Can't wait for these tools to get even better.
I appreciate the intro so much. The matter of originality has been bothering me in the back of my mind as I was thinking of using AI generation for my 3D model generation.
I can't wait for the inevitable rain of garbage AI Model flooding printables and thingaverse with comments like. I made this cool thing. Give me points for free filament.
Can't wait for the flood of regurgitated gen AI content on Fortnite & Roblox custom maps.Easy access to so much content that took no actual input or creative thought behind it is something i find incredibly boring.
I agree to disagree: Symbolic AI (of Alan Turing type) manipulates symbols according to rules or constraints to generate valid original combinations (new output)… Anyway your video is original and quite interesting!
Thanks for the video . Question.. do any of the products tested have the ability to separate objects. Like you robot on the throne. So you could separate the throne from the robot for better printing. Thanks.
So Rodin goes: text to image -> image to text, but then what model is it feeding into to make the actual 3d object? The first 2 parts are very easy but that last bit is something I haven't seen. Some sort of photogrammetry engine would be my guess. Anyone know?
Kleenex is also a brand, but it is also a generic word for 'thinly sliced dead tree for making funny nose noises'. Everyone knows what is meant, even if you aren't literally talking about the specific brand. Where I live in the US, if you say a 'Stetson', everyone will assume you mean 'cowboy hat'. I would see where that it is leap of logic for some contexts, but on a trained model, I would think it would pick up on the colloquial use of the term.
I think there is a long way till 3d ai model have a real life usage but im looking forward to having an ai add-on in blender like the adobe firefly in photoshop
Unfortunately Rodin Gen1 needs a subscription to be able to add multiple images for multiview or fusion. It also limited the polygon count for free accounts.
I covered Tripo quite a bit in my previous video on 3D generative AI. As for CSM, I find the user interface is glitchy and problematic. Plus, the free version quotes over 50 hours wait for a single generation, and even the paid version quotes 2 minutes which is slower than the free version of most of the ones featured in this video. It's something I will address in the future but I didn't think it had anything to contribute for this video.
Thanks for the review of the different applications. It looks like Rodin generates significantly better models. Does anyone know of a photo ai app that can generate multiple views of the same object so it can be used in Rodin?
Thanks for the video! Great info. One thing is that you erroneously call the hat a Stetson. Stetson is a brand and they make cowboy hats and other styles too. ;-)
For a feature they haven't done yet. It seems like there is big room for improvement on the quest home environment. Right now these are just pointless spaces, it would be nice to have portals and interfaces that take the place of boring menu mechanics ported from tablets and phones. Especially with AI integration. Perhaps I want to ask a magic well or mirror to summon a game or a Horizon space or maybe I want to use hand gestures to cast spells to open portals into different apps and games both horizon and 3rd party. Let us customize the space and have persistance. Also move apps which aren't currently installed into a menu within the apps screen you must open on purpose if they aren't going to let us remove them!
Amazing video comparing the models! I would say if you still have your account 3D AI Studio does have a Beta version of their new model that is available under tools! Its better and more on par with Rodin :) Think you should give that a go as well!!
Really glad you enjoyed the video. Yes I am going to be playing more with 3D AI studio as I've got another couple of weeks on my account with them. However, interestingly it looks like the reason their output is getting better (and "more on par with Rodin) is because it seems that 3D AI Studio are now using Deemos Techs API: x.com/DeemosTech/status/1813634728316268868
You can. You just have to have hardware capable of running the models. In the case of Rodin they likely have servers running some H100's because that generation was very fast.
That's not actually a box, it's a Google TV remote holder. However the concept can be adapted for anything which is what I did when creating that. The concept of a push, click system is nothing new. I did a dedicated video on that a few years ago that you can find here ; ruclips.net/video/K9tCMbt6WkA/видео.htmlsi=rJzWw8dhgQ87XNwl And you can download the STL from here: makerworld.com/models/66021
In the past there has been services that have claimed to be ai model generation but actually is a real person outsourced to model the object. I’m suspicious about some if these.
I'd be impressed if they had enough people skilled enough to manually design pretty detailed 3D models of anything a user requests in a matter of seconds, and capable of keeping up with the level of users using their service.
I tried all of them and none of the mesh that were generated are usable to me. The images that are being generated are getting very good but the mesh are so bad. I have not seen any improvements in the results since your last video. Try with prompt "Medieval round shield with detailed engravings" and you will see.
Great quote by Mark Twain. Evolution itself is literally copying. Plus if you believe in gods and deities they evolved their creations as well. It's not like everyone was created at all at once. Hell the universe itself is constantly evolving and largely somewhat copying previous iterations of closely the same things with minor adaptations along the way.
I am definitely looking forward to this being released to, though I think it's interesting and worth noting that in Metas own paper on this subject (ai.meta.com/research/publications/meta-3d-gen/), they reference Rodin Gen1 a lot, in comparison to their own platform and other AI mesh generators. Even in the first table detailing the capabilities of these platforms (top of page 5), they identify Rodin Gen 1 capable of generating clean topology, where as none of the others, including their own Meta 3D Gen, are capable of this. Meta 3D Gen does win out over Rodin Gen1 in that specific table in time taken however. As I say, I'm excited to get stuck into this. Though being in the UK, I know there will likely be a delay before we receive it. Meta have said they will be releasing their various AI platforms in the UK immnently, unlike the rest of Europe, but there is still going to be a bit of a wait I think,
@3drevolution I've had issues with it not generating the original image. Has anyone run into this? Example: I created a text input similar to the video with a figure on a throne. Looks great and hit generate. It would then leave out the whole throne or a portion of it or even a portion of the figure. Could never get it to generate the whole original image it created. I'm using the free account to start so not sure if that's why.
Altering the text after it generated the image does not make any difference here. "a elephant on a motorbike" changed to "a duck on a motorbike" generates the exact same image...
As someone who sculpts my own character prints, these things produce such godawful meshes that most of the time it is faster to genuinely build them from scratch. I've used generated images as references and been able to produce some great stuff, but these meshes barely even get the proportions right, let alone any details. More often than not you're better off just taking an image and building the model with primitives. The "robot king" is still 95% just rectangles, cylinders and spheres stacked on top of each other, with a little bit of proportional editing. I'm sure at some point these tools will get around to a point where they'll actually create some useful geometry for printing, but these products are nowhere near that. The 10-20 hours it takes to learn the basics of any free modeling software will leave you with a lot more options and far better models than any of these generated dumpster fires.
While some of what you say is true. Making your own from scratch would be in some ways easier. A LOT of people don't have the 10 hours to put in just to learn how to use a modeling program much less to actually make the models. I think for a lot of people this is a dream of "I have this idea in my head and I want to see it come to life." that's pretty much what 3D printing is all about. If people have the brain power and the time to learn how to 3D model their own things it is a 100% worth it if you're a 3D printer enthusiast But recognize that this is not something the is in everyone's wheelhouse. And people are looking for a tool that will help them overcome this.
The very first result on Google when I search "mark twain no such thing as a new idea", is a page on the 'Good Reads' site (www.goodreads.com/quotes/843880-there-is-no-such-thing-as-a-new-idea-it) which states that full line I read in the video as being a quote directly from Mark Twains autobiography: www.goodreads.com/book/show/7526366-mark-twain-s-own-autobiography. There's also thousands of references in every other google result I get, I'm not quite sure why you're only seeing recent references.
I think your attitude to copyright and IP needs a lot more careful thought. There is nothing wrong with derivative works but there is a lot that is wrong in how the data used to train these AI has been obtained. The whole tech industry needs an ethics overhaul. Everything from privacy, copyright and T&C that continuously change needs close attention from government and lawmakers.
I don't believe I ever specified that for this one. My previous video that compared AI 3D model generators was exclusively using free ones, but as I state in this video, that's not the case here. At the time of recording, as I say in this video, all but one of the tools I use here are available to use for free in some capacity, and the one except for that was simply to see if paying for a premium exclusive platform would give significantly better results.
3d ai studio is worthless. cant even see what it can do without forking over 15 bucks. all of these failed my voronoi balogna test. copilot passed but it wont give you a model
Honestly, give it about 5-10 years, and we'll see the AI 3D models look so good that game developers and filmmakers will be capable of using them in video games and cinematics.
Maybe it’s me but all these ai tools no matter the specialization are crap. They all seem like unfinished early stage prototypes that produce lackluster results but somehow feel entitled to charge for their services. Finish and polish your damn products ai evangelist. All these weird companies just rushing to get to market without realizing crappy products will lose more customers than being early to market will attract. Maybe I’m old and surly but all the ai is overhyped and gimmicky at best…
Im not against AI to help buuuutttt be real be real about it! I draw anime and i used AI to draw my image for my profile why? i dont have the time and 2ed the creativity for a drawing just doesn't come... ill literally sit there on my graphics tablet and i hit the undo button so many times that i would save so so much time for my projects and my everyday life. Im not a machine 😂 nore do i ever wanna be.
I have a bit of a look at Luma in one of my earlier videos (ruclips.net/video/GVd5vcMDfVI/видео.htmlsi=woeo5vSoBjpoipVA) but I'll be taking a closer look at it's gen abilities soon.
Using prompts doesn't make you an artist. AI is for the "idea man". people who have ideas but lack the skill to follow through with them. I lack skill there for rely heavily on other people's files. This technology takes real peoples art and remakes it.
AI like this will remove any meaningful human work except for physical labor. Why would a company hire an artist when they can just type what they want into a computer and generate what they need. Programs like this should be banned for the good of humanity.
@@TheOfficialOriginalChad I'm sure they do disagree but once AI has reached a point where someone can just enter a sentence to get what they want all humanity will be lost from art and design. At that point what purpose will it have? Unlike most advances in technology where one profession is replaced by another AI technology has the potential to remove almost every white collar job if it is given the opportunity to advance. Why would a company hire an artist, industrial designer, website designer, musician, graphics designer, etc when a relatively cheap AI tool could do the job for a tiny fraction of the cost?
@@smithyboats283And what makes you believe the tech will reach that point? Because what you’re describing is not technological advancements, not that AI will create better art, but a world where humans no longer value original works of art. Look around. You’re not the only person to have this believe. In fact, you’re in agreement with more than half of the general public. Now if half the general public hates AI art, please elaborate on how companies are going to make higher profits selling it to you.
@smithyboats283 @TheOfficialOriginalChad I actually agree with you both and understand both points of view. Smithy, I absolutely get what you mean, and you can to an extent already see similar impacts in other fields. As this technology (generative AI in general) advances, there will be a smaller distinction between what can be produced from a single text prompt, and what can be produced by a human who's dedicated their life to learning the arts of a trade. As someone who loves to learn new skills and techniques in creativity, that is absolutely scary. However, I also see Chad's point of view. I think we're in for quite a ride over the coming years and it's going to be turbulent, but currently (and this viewpoint changes regularly), I think it will eventually settle in an equilibrium where these tools allow lower budget/personal productions to be able to create whatever they want for free (other than the high monthly subscription fee these AIs will charge). Whilst that will of course result in a reduced level of commissions for lower cost work in those fields, due to the saturation of AI content out there, we will not only maintain our appreciation for that which is a reflection of human experience, but the desire for man made content will actually grow. I believe (or at least somewhat naively hope) that we will manage to utilize the benefits of these tools, alowing everyone to better portray and explain their thoughts, and immaginations through an array of artistic mediums that would have been out of their reach before, but that as a society we grow to revere and respect man made content far more. That said, I walk the line on whether AI is a good or bad net positive for the future of humanity, and half the time I am in the doomsday camp. However, even on that side, I very much feel that pandoras box has been opened. These tools can't be uninvented, and even if half the world outlawed it, the other half wouldn't and they would then begin producing the content that the first half consumed. In other words, there's nothing we can do to stop whatever is coming from coming now, so we may as well learn to make the most of the tools we have available to us whilst we can.
@@3DRevolutionWhat a wonderfully response. Not only because I agree with it, but I really appreciate that you have thought about this and took the time to articulate it. Thanks for tuning in to the conversation and bringing the value-add (no pun intended)
This is degrading, quoting mark twain does not absolve your complicity of the numerous data sets that have been made by scraping artists work without artists consent.
IDK, I'm going to try that Rodin one. Even if it just gets me to the base mesh stage faster it'd be worth using IMO. But seriously give it 6 months and it'll be completely different.
I think it is safe to say that I’ll be sticking to Rodin
Im speechless how fast this is progressing.
I just finished reading a series of Reddit posts from last year claiming this "might happen in our lifetimes."
There is billions the best can earn... thats why you got massive companies investing so heavily in AI and the like, as any investment will be offset by the final product. Don't think it will work perfectly in the future, but likely for small studios, it will be a massive help, or even for people needing models to better describe or display their product or service; imagine architects making models from floorplans or a DIY programmer making a stylized shell for their project, simply using a Core model with the thread holes and air vents as a reference to be built into the core.
I loved your video. Can’t believe how good the best version was. Also you gave me an excellent introduction to tools I was completely unaware of, given my 3D printing journey only began two weeks ago. Excellent work. Thanks.
6:32 this already seems impossibly good to me wtf. It actually interprets what the image SHOULD be in 3D exactly as you'd expect.
12:50 THIS is more what i thought would be the best we could get. Blobs, random wrinkles, dudes melting into chairs, etc.
I'm always astonished with how fast progress is being made in this field
really impressed with rodin, i have been waiting for some time to have good high quality 3d models from text and/or images. i will be playing with this for certain.
Thanks for testing Meshy! We've recently updated the Hard Surface Mode in Image to 3D. This enhancement significantly improves the mesh quality of hard surface models, making them ideal for 3D printing. We invite you to try it out and share your feedback!
Hey Meshy, thanks very much for getting in touch.
I've loved your setup since I started using it in prep for my previous video on 3D Generative AI.
I'll definitely be playing with the updated to your service and I'll be covering it in future videos.
It's great that it sounds like you're actively working to improve your tool for use with 3D printing so I look forward to hearing more about that.
Thanks, very nice test and you wend through the complete process including printing the models. I learned a lot from this, and I am sure going to give these AI generators a try.
I'm not surprised they struggled with the word stetson, it's not a word we use much in the US, and if in theory their training data was mostly US -english centric then it would be a word not represented much.
"cowboy hat" would have been the verbiage to use instead of Stetson which is a brand of hat not a style in the US, at least.
Unfortunately it's not a Texas-centric vocabulary. 😢
Great video! I’ve been dealing with ai tools to generate stls a bit, Rodin is really amazing!
At least on Cura and probably other slicers you don't need to cut objects that aren't flat to get a flat surface for printing. Simply set the orientation to the way up you want then set the z head to a small negative, any part of the model below the xy plane gets cut at that point when slicing and will have a flat bottom...yet I've seen so many people say to get round this by cutting models in mesh mixer...unless you're sharing it on for others that don't know this trick don't waste your time!
Wonderful video - again! Thank You!
I love your tempo, your style, your way of explaining… It is •Just Very Good• !
And… the content is very interesting and has top value as well.
Subscribed ☺️
Rodin is insanely impressive
The fact that one of them offered a riggable character makes me wonder how many low budget games will use AI generated characters in the near future
We might even skip past that straight to playing real time AI generated video as a game.
@@jaredf6205 you're completely missing my point. Congratulations
@@catdisc5304 how
@@jaredf6205He's not positive about the outcome...
I think you're underestimating how much it's already being used in big productions.
The only difference is that right now it's still less effort to dive in the free library models to make shovelware games. That's why they're commonly Known as asset dumps. But really that doesn't matter. They're not real games. It's generally just a form of scam. Those games don't hire anyone to do anything anyways. Big studios have already done a ton of their layoffs a couple years ago in reaction to this tech becoming available to them... It's not the future, it's now.
My personal hope is that we see indie become strongly empowered by these tools and can fill the gap left by increasingly uninspired big studio productions with some bigger scale gems despite not having all the resources.
Because that's how the market is supposed to work. A small nimble newcomer is supposed to be able to displace large slow moving existing companies
My neighbour asked me the other day "Why would I spend money on any AI-created product when I know I can generate mine just as easily?". Watching this video, those words are starting to sink in. Can't wait for these tools to get even better.
Wow dude just the opening mechanism is awesome.
Wow, that was a great comment about AI and humans. I totally agree!
I appreciate the intro so much. The matter of originality has been bothering me in the back of my mind as I was thinking of using AI generation for my 3D model generation.
I can't wait for the inevitable rain of garbage AI Model flooding printables and thingaverse with comments like. I made this cool thing. Give me points for free filament.
Can't wait for the flood of regurgitated gen AI content on Fortnite & Roblox custom maps.Easy access to so much content that took no actual input or creative thought behind it is something i find incredibly boring.
If it's cool, it's not garbage.
@@Kotka1986 I'd rather use AI to make some simple things for a roblox game than just use some random shit from the library like most games do.
They should make a separate website for ai print like an ai thing verse or something
@@lexdysic416 Don't see you making better models. 👀
Points just for quoting Twain!
Thank you very much for sharing. I really appreciate you and your videos/tutorials!
I agree to disagree: Symbolic AI (of Alan Turing type) manipulates symbols according to rules or constraints to generate valid original combinations (new output)…
Anyway your video is original and quite interesting!
Very cool. So glad you made this!
I might give Rodin try
Absolutely loved your video. Thank you. Liked, commented and subscribed!!!!!
Awesome! Thank you! Really glad you liked the video.
Rodin took all the golds here, it created the best-looking models, the best textures and neater meshes. Plus it did it faster.
Thanks for the video . Question.. do any of the products tested have the ability to separate objects. Like you robot on the throne. So you could separate the throne from the robot for better printing. Thanks.
this video is gold❤
So Rodin goes: text to image -> image to text, but then what model is it feeding into to make the actual 3d object? The first 2 parts are very easy but that last bit is something I haven't seen. Some sort of photogrammetry engine would be my guess. Anyone know?
Name suggestion for your horse with a mask: Mascarpone
Please compare it to Tripo 2.0, I think its even better
Stetson is a brand not a type of hat. They make many different types of hats. Maybe you meant fedora which is a type of hat.
You don’t even need to write fedora, you can just use a descriptor.
Kleenex is also a brand, but it is also a generic word for 'thinly sliced dead tree for making funny nose noises'. Everyone knows what is meant, even if you aren't literally talking about the specific brand.
Where I live in the US, if you say a 'Stetson', everyone will assume you mean 'cowboy hat'. I would see where that it is leap of logic for some contexts, but on a trained model, I would think it would pick up on the colloquial use of the term.
I think there is a long way till 3d ai model have a real life usage but im looking forward to having an ai add-on in blender like the adobe firefly in photoshop
Rodin is OP
Unfortunately Rodin Gen1 needs a subscription to be able to add multiple images for multiview or fusion. It also limited the polygon count for free accounts.
good for random props
no blinking in this video
SF3D , TripoSR , Unique3d , InstantMesh all open source can run locally on gpu with 24gb vram
you also should have used CSM and TripoAI
I covered Tripo quite a bit in my previous video on 3D generative AI.
As for CSM, I find the user interface is glitchy and problematic.
Plus, the free version quotes over 50 hours wait for a single generation, and even the paid version quotes 2 minutes which is slower than the free version of most of the ones featured in this video.
It's something I will address in the future but I didn't think it had anything to contribute for this video.
@@3DRevolution you bring a fair point, just found your content will have too look in to your Tripo cover ;)
I dont understand why the image to 3D generated are not accepting MULTIPLE images.
Why wouldn’t they want more accurate images to build upon?
Even limiting multiple images to 5 seems unusual.
Thanks for the review of the different applications. It looks like Rodin generates significantly better models. Does anyone know of a photo ai app that can generate multiple views of the same object so it can be used in Rodin?
speach: „for your dnd game“
subtitle: „for your daddy game“
Yah that gave me a giggle too.
kaedium actually blew me away .. thought rodin was the best
Thanks for the video! Great info. One thing is that you erroneously call the hat a Stetson. Stetson is a brand and they make cowboy hats and other styles too. ;-)
For a feature they haven't done yet. It seems like there is big room for improvement on the quest home environment. Right now these are just pointless spaces, it would be nice to have portals and interfaces that take the place of boring menu mechanics ported from tablets and phones. Especially with AI integration. Perhaps I want to ask a magic well or mirror to summon a game or a Horizon space or maybe I want to use hand gestures to cast spells to open portals into different apps and games both horizon and 3rd party. Let us customize the space and have persistance. Also move apps which aren't currently installed into a menu within the apps screen you must open on purpose if they aren't going to let us remove them!
SUBSCRIBED!!! 👍👍👍 IT SHOULD OF HAPPENED SOONER.
That's it, I have no more job.
Amazing video comparing the models! I would say if you still have your account 3D AI Studio does have a Beta version of their new model that is available under tools! Its better and more on par with Rodin :)
Think you should give that a go as well!!
Really glad you enjoyed the video.
Yes I am going to be playing more with 3D AI studio as I've got another couple of weeks on my account with them. However, interestingly it looks like the reason their output is getting better (and "more on par with Rodin) is because it seems that 3D AI Studio are now using Deemos Techs API: x.com/DeemosTech/status/1813634728316268868
I just wish I could get a standalone software instead of that credit and monthly payment stuff
You can. You just have to have hardware capable of running the models. In the case of Rodin they likely have servers running some H100's because that generation was very fast.
I really want to be able to use a 3D model for the style.
I got some mods I want to make to some stls I already own.
hey dude you logo looks like yeah... turn it sideways might think a rebrand or a minor tweak to the logo its buggin me
Wheezy the safe breathing horse
that hat stayed the same :P
Went from hylics to steamworld
😂
where can i find the push latch box in the video at 1:15
That's not actually a box, it's a Google TV remote holder. However the concept can be adapted for anything which is what I did when creating that. The concept of a push, click system is nothing new.
I did a dedicated video on that a few years ago that you can find here ;
ruclips.net/video/K9tCMbt6WkA/видео.htmlsi=rJzWw8dhgQ87XNwl
And you can download the STL from here:
makerworld.com/models/66021
In the past there has been services that have claimed to be ai model generation but actually is a real person outsourced to model the object. I’m suspicious about some if these.
I'd be impressed if they had enough people skilled enough to manually design pretty detailed 3D models of anything a user requests in a matter of seconds, and capable of keeping up with the level of users using their service.
The speed of generation makes your suspicions impossible. That's not a person, that's an H100 gpu.
I tried all of them and none of the mesh that were generated are usable to me. The images that are being generated are getting very good but the mesh are so bad. I have not seen any improvements in the results since your last video. Try with prompt "Medieval round shield with detailed engravings" and you will see.
I'll definitely give it a try. Prompt engineering is definitely an important aspect of it.
Great quote by Mark Twain. Evolution itself is literally copying. Plus if you believe in gods and deities they evolved their creations as well. It's not like everyone was created at all at once. Hell the universe itself is constantly evolving and largely somewhat copying previous iterations of closely the same things with minor adaptations along the way.
Meta 3d Gen laughing at this. Just waiting for meta to release it
I am definitely looking forward to this being released to, though I think it's interesting and worth noting that in Metas own paper on this subject (ai.meta.com/research/publications/meta-3d-gen/), they reference Rodin Gen1 a lot, in comparison to their own platform and other AI mesh generators.
Even in the first table detailing the capabilities of these platforms (top of page 5), they identify Rodin Gen 1 capable of generating clean topology, where as none of the others, including their own Meta 3D Gen, are capable of this. Meta 3D Gen does win out over Rodin Gen1 in that specific table in time taken however.
As I say, I'm excited to get stuck into this. Though being in the UK, I know there will likely be a delay before we receive it. Meta have said they will be releasing their various AI platforms in the UK immnently, unlike the rest of Europe, but there is still going to be a bit of a wait I think,
Name: Mask-a-Mane or Mas-K-Raid.
Another "prompt engineer". cool
@3drevolution I've had issues with it not generating the original image. Has anyone run into this? Example: I created a text input similar to the video with a figure on a throne. Looks great and hit generate. It would then leave out the whole throne or a portion of it or even a portion of the figure. Could never get it to generate the whole original image it created. I'm using the free account to start so not sure if that's why.
Altering the text after it generated the image does not make any difference here. "a elephant on a motorbike" changed to "a duck on a motorbike" generates the exact same image...
Video started 3.45
As someone who sculpts my own character prints, these things produce such godawful meshes that most of the time it is faster to genuinely build them from scratch. I've used generated images as references and been able to produce some great stuff, but these meshes barely even get the proportions right, let alone any details. More often than not you're better off just taking an image and building the model with primitives. The "robot king" is still 95% just rectangles, cylinders and spheres stacked on top of each other, with a little bit of proportional editing.
I'm sure at some point these tools will get around to a point where they'll actually create some useful geometry for printing, but these products are nowhere near that. The 10-20 hours it takes to learn the basics of any free modeling software will leave you with a lot more options and far better models than any of these generated dumpster fires.
While some of what you say is true. Making your own from scratch would be in some ways easier. A LOT of people don't have the 10 hours to put in just to learn how to use a modeling program much less to actually make the models. I think for a lot of people this is a dream of "I have this idea in my head and I want to see it come to life." that's pretty much what 3D printing is all about.
If people have the brain power and the time to learn how to 3D model their own things it is a 100% worth it if you're a 3D printer enthusiast But recognize that this is not something the is in everyone's wheelhouse. And people are looking for a tool that will help them overcome this.
a Giga Farm of 3d Printers for robots 🤣
:o
Oldest Mark Twain quote "There is no such thing..." I found internet is from 2016 and there is not much answears in search.. troll?
The very first result on Google when I search "mark twain no such thing as a new idea", is a page on the 'Good Reads' site (www.goodreads.com/quotes/843880-there-is-no-such-thing-as-a-new-idea-it) which states that full line I read in the video as being a quote directly from Mark Twains autobiography: www.goodreads.com/book/show/7526366-mark-twain-s-own-autobiography.
There's also thousands of references in every other google result I get, I'm not quite sure why you're only seeing recent references.
I think your attitude to copyright and IP needs a lot more careful thought. There is nothing wrong with derivative works but there is a lot that is wrong in how the data used to train these AI has been obtained. The whole tech industry needs an ethics overhaul. Everything from privacy, copyright and T&C that continuously change needs close attention from government and lawmakers.
I thought this videos was for FREE softwares
I don't believe I ever specified that for this one.
My previous video that compared AI 3D model generators was exclusively using free ones, but as I state in this video, that's not the case here.
At the time of recording, as I say in this video, all but one of the tools I use here are available to use for free in some capacity, and the one except for that was simply to see if paying for a premium exclusive platform would give significantly better results.
@@3DRevolution I will search your videos then in your home page. Thank you
You dont even blink
Blinking is so overrated!
Name for the horse - Horvid = Horse + Covid🤦♂
Haha nice suggestion thanks. Unfortunately the poll ended on the weekend and the winning name is SeaGasket.
3d ai studio is worthless. cant even see what it can do without forking over 15 bucks. all of these failed my voronoi balogna test. copilot passed but it wont give you a model
Maskretariat? Seagasket?
Honestly, give it about 5-10 years, and we'll see the AI 3D models look so good that game developers and filmmakers will be capable of using them in video games and cinematics.
Maybe less time, I would say 5 years as max in that span
Probably a good prototype in 5 years rather than being full scale
sad day to be a 3d modeler
Maybe it’s me but all these ai tools no matter the specialization are crap. They all seem like unfinished early stage prototypes that produce lackluster results but somehow feel entitled to charge for their services. Finish and polish your damn products ai evangelist. All these weird companies just rushing to get to market without realizing crappy products will lose more customers than being early to market will attract. Maybe I’m old and surly but all the ai is overhyped and gimmicky at best…
There is a mega hype, because companies are investing tons of money in AI, despite the not so good results.
I think you are just old and lack foresight
Rodin gen1 seems pretty good, assuming it keeps improving, it'll probably be usable in a few years.
Im not against AI to help buuuutttt be real be real about it! I draw anime and i used AI to draw my image for my profile why? i dont have the time and 2ed the creativity for a drawing just doesn't come... ill literally sit there on my graphics tablet and i hit the undo button so many times that i would save so so much time for my projects and my everyday life. Im not a machine 😂 nore do i ever wanna be.
Pls stop. I can feel my eventual career becoming less eventual
This is crap so far 😂 The texture cration on Meshy seems good though.
Any one test luma ai
I have a bit of a look at Luma in one of my earlier videos (ruclips.net/video/GVd5vcMDfVI/видео.htmlsi=woeo5vSoBjpoipVA) but I'll be taking a closer look at it's gen abilities soon.
all of these programs are based of stolen artwork and the end result looks like trash
tried that useless and waste of time
Those who complain about AI's are rocks that do not want to move to the future. It's just an assistant.. much like a slave in prehistoric time 😂
Using prompts doesn't make you an artist. AI is for the "idea man". people who have ideas but lack the skill to follow through with them. I lack skill there for rely heavily on other people's files. This technology takes real peoples art and remakes it.
AI like this will remove any meaningful human work except for physical labor. Why would a company hire an artist when they can just type what they want into a computer and generate what they need. Programs like this should be banned for the good of humanity.
Most AI researchers and developers disagree. In fact, it’s quite the opposite.
AI needs to be trained.
@@TheOfficialOriginalChad I'm sure they do disagree but once AI has reached a point where someone can just enter a sentence to get what they want all humanity will be lost from art and design. At that point what purpose will it have?
Unlike most advances in technology where one profession is replaced by another AI technology has the potential to remove almost every white collar job if it is given the opportunity to advance. Why would a company hire an artist, industrial designer, website designer, musician, graphics designer, etc when a relatively cheap AI tool could do the job for a tiny fraction of the cost?
@@smithyboats283And what makes you believe the tech will reach that point? Because what you’re describing is not technological advancements, not that AI will create better art, but a world where humans no longer value original works of art.
Look around. You’re not the only person to have this believe. In fact, you’re in agreement with more than half of the general public. Now if half the general public hates AI art, please elaborate on how companies are going to make higher profits selling it to you.
@smithyboats283 @TheOfficialOriginalChad
I actually agree with you both and understand both points of view.
Smithy, I absolutely get what you mean, and you can to an extent already see similar impacts in other fields. As this technology (generative AI in general) advances, there will be a smaller distinction between what can be produced from a single text prompt, and what can be produced by a human who's dedicated their life to learning the arts of a trade.
As someone who loves to learn new skills and techniques in creativity, that is absolutely scary. However, I also see Chad's point of view. I think we're in for quite a ride over the coming years and it's going to be turbulent, but currently (and this viewpoint changes regularly), I think it will eventually settle in an equilibrium where these tools allow lower budget/personal productions to be able to create whatever they want for free (other than the high monthly subscription fee these AIs will charge).
Whilst that will of course result in a reduced level of commissions for lower cost work in those fields, due to the saturation of AI content out there, we will not only maintain our appreciation for that which is a reflection of human experience, but the desire for man made content will actually grow.
I believe (or at least somewhat naively hope) that we will manage to utilize the benefits of these tools, alowing everyone to better portray and explain their thoughts, and immaginations through an array of artistic mediums that would have been out of their reach before, but that as a society we grow to revere and respect man made content far more.
That said, I walk the line on whether AI is a good or bad net positive for the future of humanity, and half the time I am in the doomsday camp. However, even on that side, I very much feel that pandoras box has been opened. These tools can't be uninvented, and even if half the world outlawed it, the other half wouldn't and they would then begin producing the content that the first half consumed. In other words, there's nothing we can do to stop whatever is coming from coming now, so we may as well learn to make the most of the tools we have available to us whilst we can.
@@3DRevolutionWhat a wonderfully response. Not only because I agree with it, but I really appreciate that you have thought about this and took the time to articulate it.
Thanks for tuning in to the conversation and bringing the value-add (no pun intended)
This is degrading, quoting mark twain does not absolve your complicity of the numerous data sets that have been made by scraping artists work without artists consent.
ai i s still crap!!
@@IronMan-yg4qw .. depending on what you compare it too 😊.
@@oljobo im just comparing all the ai modeling software here!!!
IDK, I'm going to try that Rodin one. Even if it just gets me to the base mesh stage faster it'd be worth using IMO. But seriously give it 6 months and it'll be completely different.
Meta 3d Gen 🗿
We found another digital artist realizing they will be out of a job soon. Lol