I can see this being really popular for creating sketches and animatics as well as for brainstorming possible ideas. At the rate this is improving it is hard to imagine where we will be in a couple of years.
It will scale up pretty fast. In a few years this will be very disadvantegeous for architects who are on a salary. If you are owner of your own business then it is good for you.
I think architects will be just fine. The constellation of factors in which an architect operates goes far beyond "form making", there's so much real-world stuff that needs to be considered and reacted to, that I don't see AI coming even close to it in the foreseeable future. Gamedev background asset creation though... that field is getting a shake-up.
IMO, I think this is good because an 3d artist can generate the base mesh with textures and materials, then modify it to their liking. I mean programmers do this all the time. We get some base code, or someone else's code, and modify it to the needs.
The only issue is that the automatically generated UV maps are pretty nasty to adjust - a ton of islands that don't have much apparent logic to their distribution
@@samthesomniator but then I would be doing the UV unwrapping, while the AI does the creative initial concept design. Might as well end it all there then
@@DesignGoBrr It's ok, the fun, creative part will be done by AI and the robot work will be left to us biorobots for a while..... and then it will do that later.
@@DesignGoBrr perhaps, although I guess the fact that you don't get actual geometry from a Gaussian Splat makes it a bit different. The beauty of this AI is that it actually creates the PBR materials, so reflections and metalness etc. can be a bit more accurate than the baked in results you normally get from photogrammetry. If this new tech also means you need only a handful of images as opposed to 1000s for a high quality 3D scan then that would be a game changer! Thanks for an interesting video as always 🙌
Really depends on the skill-level. I'd say a few hours would be a good bet. The only issue is that - after a few hours of working on the model, the professional will be able to offer minor adjustments/changes in stylistic direction to "nail down" the look. Generating slight adjustments, that align with the given art-direction of the client with AI would be much more frustrating.
I'd say 10-20 minutes max to speed sculpt this quality! I think it's a cool tool to get some base shapes established but far from finished production quality design. Thanks for sharing @designGoBrr
AI is going after all professions now ha? 😅I guess there is no choice but to know these tools and use them to our advantage. Thank you for sharing this
It's hilarious how AI can conjure 3D meshes from a flat picture, yet rendering engines are still stuck in the Stone Age with quad topology, fumbling around like they're allergic to anything non-quad for realistic refraction! For Rhino, I just wish AI could retopologize my 3D mesh into real Quads-nothing more! 👽 Meanwhile, after three years, the retopology modifier is still unreliable, with the same slow and buggy mess from Rhino to 3ds Max-all are the same and full of bugs!
Thank you interesting process and product, love the print, going to give it a try now.
I can see this being really popular for creating sketches and animatics as well as for brainstorming possible ideas. At the rate this is improving it is hard to imagine where we will be in a couple of years.
It will scale up pretty fast. In a few years this will be very disadvantegeous for architects who are on a salary. If you are owner of your own business then it is good for you.
I think architects will be just fine. The constellation of factors in which an architect operates goes far beyond "form making", there's so much real-world stuff that needs to be considered and reacted to, that I don't see AI coming even close to it in the foreseeable future. Gamedev background asset creation though... that field is getting a shake-up.
IMO, I think this is good because an 3d artist can generate the base mesh with textures and materials, then modify it to their liking.
I mean programmers do this all the time. We get some base code, or someone else's code, and modify it to the needs.
The only issue is that the automatically generated UV maps are pretty nasty to adjust - a ton of islands that don't have much apparent logic to their distribution
@@DesignGoBrr Yes. But no Problem with projecting them to a better UV unwrapped Map. 🤷🏻♂️
@@samthesomniator but then I would be doing the UV unwrapping, while the AI does the creative initial concept design. Might as well end it all there then
@@DesignGoBrr It's ok, the fun, creative part will be done by AI and the robot work will be left to us biorobots for a while..... and then it will do that later.
Will be interesting to see if this merges with or even replaces photogrammetry tools in the future!
Hmm, I wonder if gaussian splatting has any relation to this.
@@DesignGoBrr perhaps, although I guess the fact that you don't get actual geometry from a Gaussian Splat makes it a bit different.
The beauty of this AI is that it actually creates the PBR materials, so reflections and metalness etc. can be a bit more accurate than the baked in results you normally get from photogrammetry.
If this new tech also means you need only a handful of images as opposed to 1000s for a high quality 3D scan then that would be a game changer!
Thanks for an interesting video as always 🙌
thanks youre always awesome Ged O7
O7
*Chaos Cosmos sweating profusely*
100% working on ChaosAI
A few more years and everyone will be back dancing in ballet...
I'm more of a hand-crafted furniture guy myself
As someone not familiar with 3d software, how long would this process have taken a professional to finish with the same quality 3d object?
Really depends on the skill-level. I'd say a few hours would be a good bet. The only issue is that - after a few hours of working on the model, the professional will be able to offer minor adjustments/changes in stylistic direction to "nail down" the look. Generating slight adjustments, that align with the given art-direction of the client with AI would be much more frustrating.
I'd say 10-20 minutes max to speed sculpt this quality! I think it's a cool tool to get some base shapes established but far from finished production quality design.
Thanks for sharing @designGoBrr
Thanks for showing what is available. But, the model doesnt look much like the image.
I disagree - if you grayscale the image, the model is actually pretty close to it.
AI is going after all professions now ha? 😅I guess there is no choice but to know these tools and use them to our advantage. Thank you for sharing this
Always has been. Yea, knowing what's out there and how to use it is the best way to keep yourself in the front of this new wave :)
This chair is based on the Alessandro Mendini Proust 💺
Cheaper though
It's hilarious how AI can conjure 3D meshes from a flat picture, yet rendering engines are still stuck in the Stone Age with quad topology, fumbling around like they're allergic to anything non-quad for realistic refraction!
For Rhino, I just wish AI could retopologize my 3D mesh into real Quads-nothing more! 👽 Meanwhile, after three years, the retopology modifier is still unreliable, with the same slow and buggy mess from Rhino to 3ds Max-all are the same and full of bugs!
I've seen some progress made there!
@@DesignGoBrr Can you make a video on this or give some pointers? I'm learning 3D modelling using AI and I really appreciate this video.
instant meshes or other retopolgy softwares can be used for this right?
link dont work
Should be fine now
Lumalabs is free and better than that. I think they have a long way to go before they can start charging for it.
What? Lumalabs is a video generating tool, while this is a 3D model generating tool.