I don't see them plateauing. I see that they are in their infancy. So far AI people were just persistently doing the dumbest possible thing that turned out to work, with very limited understanding, insight and innovation. At some point they will buy all the gpus and all the GWh they can and will be forced to actually figure out how to really improve what they are doing. Then the real breakthroughs will start showing up. There are probably improvements of 3-4 orders of magnitude right behind the finish line of the low hanging fruit picking contest.

AI will just be cheaper procedural environments

Well it kinda did change things up a bit. Me being able to receive payments across borders without significant delay or crazy fees is a decent perk, you can hate crypto culture and grifters trying to make a quick buck but it's applications are very real.

I'm sure he is just trying to sell snake oil.

There are probably already a bunch of Blender Add-Ons or extensions that build with AI that are in the approval queue and just being ignored. https://extensions.blender.org/approval-queue/

Why bolt magic rocket boosters onto a horse?

That's like saying we'll add internet browsing and YouTube to Lotus 1-2-3 for DOS.

It's weird, legacy software for a dying modality.

Where we're going, we don't need geometry node editing. The diffusion models understand the physics of optics better than our hand-written math. And they can express it across a wide variety of artistic styles that aren't even photoreal.

The future of 3D probably looks like the video game "Tiny Glade" mixed with Midjourney and Google Genie. And since it'll become so much more accessible, I think we'll probably wind up blending the act of creation with the act of consuming. Nothing like Blender.

All these pro-AI framings hinge on the fact that they can't tell apart AI data from human arts. That's like saying that because they don't know what improper bounds check is the code must be secure. It's just all broken logic.

>What's the use case of generating a model if all modelling and game engines are gone?

Because using LL3M-style technique will probably be cheaper and better (fidelity and consistency and art direction wise) than generating the entire video/game with video generation model.

> you'll come to realize

No. The Roblox of this space is going to make billions of dollars.

There's going to be so much money to make here.

Just don't slap a release year on this future and I'll be compelled to agree

No, the meshes involved are in the same ballpark as children's drawings for 2D art.

I'm sure the most difficult part here is just understanding blender UI. Clearly more difficult than picking up a pencil. But, a tutorial video should suffice.

For the chair example you pick a face on the default cube and the use the extrude tool on the left. Now you have a base.

Add 4 more cubes, and do the same. Now you have legs.

Then boolean them.

For the hat? Use a sphere, go to the sculpt tab and go ham.

There are way better ways to do this, of course. But really, there is not such a high degree of skill involved here, nor that being just a little more patient (one more day of trying) is that much to ask.

I'm similar, honestly. I've spent countless hours trying to become a good drawer and a good 3D modeler, but I lack the ability to see something clearly in my mind's eye, and it feels like it's always held me back.

The thing is, I've actually worked as a 3D artist for a number of years. Some people even tell me I'm good. I suppose if that's true at all, it's because I've learned to use the computer to do the visualizing for me.

For some other artists, their process seems to be that they first picture a 'target' image in their mind, and then take steps towards that target until the target is reached. That seems impossible to me -- supernatural stuff. I almost don't believe they can really do it.

My process is closer to first finding some reference images, then taking a step in a random direction and asking whether I'm closer or further away from those references. I'm not necessarily trying to copy the references exactly, I'm just trying to match their level of quality. Then I take another random step, and check again. If you repeat this process enough times, you'll edge closer and closer to something that looks good. You'll also develop a vague sense of 'taste', around which random movements tend to produce more favourable results, and which random movements tend to produce more ugly results. It's a painful process, but it's doable.

I guess what I'm trying to say is that the ability to visualize isn't a prerequisite for 3D modeling.

I can agree with this. If someone has some kind of disability, like aphantasia (I don't know if it really applies here, as you can look at a reference image) then perhaps the tool is useful. The thing is, none of the examples presented in this particular AI tool are stuff that require hard 3D-related skills e.g. knowing human anatomy.

I wish I could see you struggling to model a trash can and see if maybe you didn't have too high requirements for the quality of said trash can. After all it's just taking a cylinder, insetting the top face and extruding it down, and the top you can model in the exact same way. The rest is detail that the AI tool in question is terrible at. https://i.imgur.com/xeFrgpP.gif

I could probably write a book to answer this question :D Blender has gone "everything nodes" route, and in particular it created the Geometry Nodes system. I'm very good in geonodes and pretty much specialize in it, and yet I think the system is severly flawed: the nodes in compositing and shading work very well, but in geonodes they are too low-level, and you don't get the easy learning curve of usual node systems (learning geonodes is very hard), while you get all the annoyances of using a very shoddy programming language and having to manage node positioning and fighting with noodles...

...And here's where AI comes into play, If AI could be contained into steps: - Input node: describe where the starting data comes from and AI automatically loads a file from hard drive or Internet or generates a primitive - Select node: describe a pattern by which to select elements of the geometry - Modify Geometry node: perhaps should be split into multiple nodes as there's so many ways to modify the geometry. - Sample/connect data: create an attribute and describe a relation of it to something else to create an underlying algorithm populating this attribute. - Save node: do you want to output the data through the usual pipeline, or maybe export to a file, or save to a simulation cache?

This way AI could do low-level stuff that I think it excels at, because this low-level stuff is so repeatable AI can be well trained on it. While the high-level decision making would be in control of an artist.

These models could still be useful when rendered. But when animated or in a game probably less so. Maybe as a prototype to get funding and hire an artist.