Never has been and never will be.
And ofcourse i never-ever use blender or 3d max as a CAD. The same way i will never use my chair as a mallet or dust pan as a kettle.
And im saying this as a person who was in charge of design a cnc metal milling machine in Solidworks + all dust protections and enclosure. So im pretty much understand what is CAD are really about.
Well, since what you describe is quite blurry and speculative it’s hard to agree or not…
Resources like quixel megascan completely removed the need to build our own nature assets like it was the case years ago. But this didn’t changed DCCs it’s more a different way to work in 3D for users.
In the meantime, when you deal with very specific art direction and style you end up doing everything manually anyway.
Maybe at some point tools will allow to take a 2D design and do a 3D model out of it, but it’s a long way especially when you know the specifics of a production. Let’s discuss that when we get there ? My bet is that we’ll see quite interesting AI tools but they won’t completely cancel what we have now.
Another thing I’d like to mention is that Blender is always behind when it’s about integrating new tech, UDIM, USD, Alembic, where always been implemented in commercial DCC first, and much later in blender.
While it probably delayed blender adoption for some people, this never killed blender on the other hand. Ultimately, if more companies relies on blender for their pipeline, they might also jump in and help integration of these new tech. And on a smaller scale this applies to users as well.
It would be a very dangerous move for BF to start investing in speculative tools when first we don’t know how they should look like, and secondly there are a lot of regular things to improve and add in the meantime.
2 Likes
That statement will be true for similar software with same goals.
As a Free Open Source Software, it has a better chance to appear as an alternative to corporations providing closed AI solutions.
It can also be an hub to them, providing ability to create base work guiding AI, in a more efficient way than just text.
AI is delivering a result according to its training. So, an AI without bias can not exist.
There will always be a need for a user to free himself from AI bias.
So, the need for tools providing manual modeling or manual animation will endure.
Blender will continue, as long as, there will be users, to finance its maintenance and development.
It does not really matter if userbase will be made of more professionals than hobbyists or the opposite.
Blender will correspond to what manual control ; users will want to keep.
3 Likes
CAD is a very wide term. We use ArchiCAD for interior design in the studio I work at. Blender is not that far from the CAD functionality needed by interior designers in my opinion. Quite far, but not that far. I often make 2d drawings of furniture with more complex forms or stairs or roofs that would be difficult or more time consuming to model or draw in ArchiCAD for the interior designers. The drawings lack any curves 
which is ridiculously bad for technical drawings, but since they are often viewed printed out on paper and manufacturers remake technical drawings for their manufacturing processes anyway, that works fine. I believe, most CAD drawings produced by interior designers could be done in Blender. That would be way worse than what they make with ArchiCAD, but would still work for the purpose just fine. Except there is no easy way to save them to PDF or print, so one would need to do a lot more work and also use some vector drawing software like Inkscape to make them presentable. So… It’s probably not the best idea, but still… possible.
1 Like
I think that due to the fact Blender is open source and anyone can use the code (with foss restrictions), the reverse could happen.
People using AI will be free to use Blenders code to help them generate 3d content. Blender could end up in the center of AI created content.
I already mentioned that AI content generation tools are developed at an accelerated rate. I can’t give you precise blueprints for what might happen in coming months. My point was that the 3D modeling/rendering/content creating as we know today might not exists in couple years whether I personally approve it or not. BF will be at crossroads at some point to what to do about it.
Being FOSS is not necessarily an antidote for what is happening in the content creation world. So far All the big tech companies have been using immorally (and unlawfully in many cases) obtained data to train their AI. There are already scraped 3D repositories for 3D data training hence my point that generating 3d models will be just a prompt away for instance.
We used Blender for synthetic data training projects some years ago, so this is an area I know well. Blender like tools will still be needed to create content for AI training, but that won’t be forever.
1 Like
I do not see foss as an antidote for AI creations, more as tools that can be used for AI creation, possibly a catalyzer.
The future is unsure and will (in my opinion) go further than we anticipate at the moment, I agree that many underestimate the impact AI will have on our lives. Yes it is scary.
It is totally possible that AI will be capable of creating its own tools, based obviously on the existing ones we have today.
1 Like
Well, I understand, but I find hard to build upon what you say, especially in relation to the topic.
AI will certainly change our lives just as much as the introduction of the personal computer back then, or in a similar way to how CG changed traditional medias.
How should blender adapt to that ? We can’t tell yet since we can’t predict the future, and anyway, AI tools like cycle’s denoisers are already there in blender.
If a real breakthrough append, I’m pretty sure it will come to blender, just like a lot of things.
When I started CG, sculpt/Z-brush didn’t existed, and now it’s the preferred way for character creation, and eventually it was added to blender since there was a real need from the comunity.
And anyway, if
Then I might probably find another job, but in the meantime I’ll still enjoy doing 3D just like I do now and maybe even more if cool AI tools come out.
3 Likes
The subject of AI in yet another thread?
It looks like we found the successor to Godwin’s Law. The longer a thread goes on, the chance of the subject of AI coming up approaches 1. It is as if we do not have anything else to discuss regarding things like how to deal with Blender’s increasing bug count (in which I think the BF is aware, as that might be a driving point behind the decision to go back to non-overlapping release cycles).
4 Likes
LOL !
Yes indeed, you are right, let’s get back on track, but let us remind that this topic started with questions about going back to 2.79 UI, at least with AI subject we are headed at the future rather than the past… should we say that the conversation is moving in the right direction then ?
Interesting , do you have any link where this is discussed ? In the meantime I’m not that surprised since a lot of changes are introduced in these last release, maybe it’s time to focus on stabilizing after 4.0, IIRC the tracker curfew worked well back then, maybe it’s time for the second opus ?
Agreed, I’d prefer AI speculation to stay in the AI speculation thread.
4 Likes
Agree. CADwork definitely is.
Mechanical and Architectural CAD approaches has different workflow requirements.
CAD modeling features are one of the most requested at the moment.
We use Blender for manufacturing purposes a lot, and also made different extentions to enhance CAD-like precise modeling workflows.
Sorry, I am not familiar with what you use Blender for.
I am actually an AutoLISP programmer, who wrote a thousand of LISP routins, including AutoCAD-to-Blender exporter. For example, we use Blender for making and editing general plans (sometimes, up to 50 hectares). Such a modeling (conversion of a messy raw drawing into topologically manifold surface) in Blender usually takes a day:
From the interoperability point, Command Design Pattern resulted in Commands Aliases (PGP) system.
The difference of a commands/aliases and hotkeys approaces is that commands could be invoked as an active tool (noun behaviour) or modals (verb behaviour) on demand.
- Commands are indifferent to the selection order, so noun/verb ways originally was seamlessly integrated.
- Hotkeys system brought a strong differentiation - you have to chose what hotkey will do, call an active tool (noun) or invoke an operation (verb).
Originally hotkeys-based software design went different ways, 3dsmax with its design derivatives (like Sketchup) has gone noun way, and Blender has gone verb way
In Blender noun way in form of an Active tools was implemented quite recently so we has got this differentiation:
Different ways has different goals and limitations, for example Pure Active tools concept is easier to learn but has significant expansion limitations. That’s why, for example, Sketchup, which was built around pure active tools noun way, has been looking so crowded lately.
3 Likes
What 3d repositories? Are they downloading models from Pirate Bay?.. I was thinking about this - how long will it be until AI takes my job and I will only need to write what I want and a 3d model will appear… Well… It will be a while, since there is no good data in meaningful amounts for training. Image or natural language generators are trained on unimaginably huge data sets. Like the whole internet of text and images. Those are unimaginable amounts of data. That’s why we see results. Existing 3d models online lack in quantity and in quality by many orders of magnitude compared to images and text. Anyway, Blender’s development hasn’t been influenced much by recent AI breakthroughs, so that’s probably getting a bit off topic…
2 Likes
The big player Nvidia is currently working on 3D generative AI tech, time will tell how it will affect Blender and the 3D industry.
Funny fact, both Google and Nvidia used my CC0 models for AI training and even making presentations)
2 Likes
You are not grasping the speed of the development in this area. Because you do not see a lot of free models around does not mean someone else does not have millions of them ready for data training. Text to 3D and image to 3D is already are already developed, and it is in an infant state now which means that it will be ready in couple years+
I brought up the AI into Blender conversation, not because I want to bog down the conversation with it but because AI 3D tools will be reality soon, and Blender could end up a tool that only hardcore people use it, most will adapt to easier way of creating content.
This thing seems to have close to a million models if their numbers are correct. It is from Allen Institute of AI.
Why do you think Epic bought Sketchfab? If you do not take me serious maybe you would take Tim Sweeney of Epic serious, what he said yesterday.
https://twitter.com/TimSweeneyEpic/status/1687231131576340480
Disclaimer, I am neither pro AI nor anti AI, just stating what I am seeing based on my own technical experience.
Sure.
It is expected that creative donut-level modeling areas will be replaced in the first row, and deep technical modeling like multiref and so one are supposed to fall last.
This will naturally bring back an extremely high entry threshold Blender users originally was prepared for.
2 Likes
That is fine, except that the popularity of the tool what brings the funding for the development.
1 Like
I am feeling kind of excited about that when I see it, but still I am quite sceptical about it. I think the results are kind of OK, but it might be, that there is no data for training for them to become amazing like with image generation. I think AI capabilities in this area might get stuck at some level of quality for a long time and while it’s still amazing, I don’t think that level will be very usable for anything serious or that it would get close to human ability level. Not that I would think the technology is not capable of that - I definitely believe AI can surpass humans in 3d modelling, there is nothing special about it, it’s just that I don’t see where the data for training would come from. Maybe AI will get better at learning in time.
To get back to the topic of Blender development, I don’t think it being open source has any advantages in this area. I am actually a bit worried about development of AI related functionality in Blender in the future. Companies like Adobe and Autodesk are not stupid - they understand that they can profit greatly from the technology and AI training takes a lot of resources that they have. Blender Foundation may struggle to compete with big companies in this area simply because they aren’t as rich.