pretty sure AI would say ‘life is too short for that crap…’ >grins< or they would if they had consciousness, anyway. hehehe
Hi community, I’ve been reflecting a bit lately on when/if/how Blender will use AI to generate 3D models sometime in the future. It would be interesting to hear out your opinions on Blender’s future and AI. Do you think Blender should start focusing on AI? If so, how realistic is that? If not, why?
To get started, here are my reflections on the topic.
It’s undeniable that AI has become one of the most talked about techs recently. With its potential to revolutionize numerous industries, AI is poised to transform the way we work. In the realm of 3D, Blender as an open source powerhouse could be well positioned to leverage this technology.
The recent advancements in generative AI could open up a unique opportunity for Blender to become the first complete 3D app to fully integrate this technology. By doing so, it could set itself apart from its competitors such as Autodesk’s 3ds max and Maya who are yet to embrace this innovation. This move could position Blender as the industry standard for 3D-generative AI, much like how Adobe has been a pioneer in integrating 2D-generative AI in Photoshop powered by Firefly.
Potential Benefits:
-
Blender could become the first complete 3D app to fully integrate generative 3D AI, a move that might propel it ahead of its competitors such as 3ds max and Maya. This could lead to increased popularity and adoption within the industry
-
AI-driven generative features in Blender would empower users to create models and props rapidly, streamlining their workflows and enhancing productivity. This could be particularly useful for creating background elements or fillers to scenes that don’t require precise details (as AI technology today is not perfect).
Potential Challenges:
-
The development and implementation of powerful AI would require significant resources and expertise, which may be beyond the reach of the Blender Foundation. They could consider partnering with organizations like Stability AI (developer of Stable Diffusion) which focuses on open source AI technology to develop and implement the AI features.
-
Integrating generative 3D into Blender would require careful consideration regarding how the AI weights would be run - either on a cloud or locally on users’ computers. The latter might pose technical challenges as average computer specifications might not support such demands. Alternatively, cloud-based solutions could raise concerns about data privacy and security. If running on a cloud service, another issue to consider would be financing: Who would cover the costs of hosting and maintaining the AI models and weights?
-
How would these AI systems be trained? From Blender Market? It might not provide sufficient content for AI training purposes due to copyright restrictions and user concerns about their work being used as training material without permission (which is fully understandable). This raises a challenge: How to obtain adequate training data while respecting the intellectual property rights of creators and maintaining community trust? Perhaps a collaborative approach, involving voluntary user contributions could provide a solution that benefits everyone?
Or perhaps, Blender should just skip the AI hysteria altogether? After all, there’s no need to jump on every bandwagon that comes along.
2 Likes
This is quite timely- welcome to BA, and see the post directly above you as to why your thread was moved here 
1 Like
If the topic of this thread was “owning a cat”, it seems to me that responses that involved not wanting a cat, not liking cats, fear of cats, love of cats - might or might not be directly applicable to the thread’s (intended) topic, but given it’s off-topic nature relative to the forum, i don’t believe we would need two separate threads for cat discussion.
If participants really need two to three different topics to discuss this, so that AI lovers don’t have to interact with people that don’t like ai, perhaps it would be best to move the entire conversation to a website where AI is actually the topic.
2 Likes
- AI models take multiple gigabytes, and Blender distribution is about 300MB. I don’t think this is a good tradeoff.
- It’s already easy to integrate any AI process into Blender through Python if you want. If I remember, there are already multiple addons that do just this.
- It’s quite likely not worth the investment for the devs to do any of this development at the cost of everything else.
- It’s a controversial topic that could affect Blenders reputation very negatively.
Also I don’t understand why moderation allows snide and dismissive off-topic meta-comments like @thorn
1 Like
Well said!
I didn’t interpret it that way, I saw @thorn-neverwake explaining their perspective in a reasonable and rational way. I don’t see anything snide or dismissive about it- in fact, I read it as supportive and helpful, an attempt to explain why this thread is what it is, and honestly, they did a better job than I did
Now, with that established, let’s get back to the AI discussion, thanks
What’s really going to happen is a few very low paid humans with AI will replace thousands who have fully embraced AI. Learning AI tools to stay competitive is a fools errand. There will not be enough prompting jobs to go around, and independent artists for the most part will not be able to afford massive farms of computers like big corporations can. The bottleneck for AI generated content is not human labor, because the humans do almost nothing, the bottleneck is hardware.
3 Likes
Thank you, that moderation was fast. While I was aware of this thread, I hesitated to contribute due to concerns that this thread appeared to be more focused on “Artist vs AI” while my perspective was centered around “How Blender can use and implement AI”.
But thanks for correcting the place of my post.
1 Like
The “Artist vs. AI” is certainly setting the tone immediately in the wrong direction. A much more constructive thread would be “AI general”
I don’t think the future of ML/AI is general big models. I think it’s many smaller, well controlled models, which have the benefit of:
- Better controlled source material without IP violations
- Better results and output given the goals
Also like you said, Adobe already has a model that is claimed to be free of copyright violations, and they offer indemnity protection. Is open source going to be left behind and will the future of art be tightly controlled by big corporations charging high rent for their tools? I think this is an important point that many people should think about.
Even Adobe does not own that many images to train a complicated model from scratch. I think they are skewing the facts and the underlying external weights they might have used. They probably also scraped the internet for “freely available” images from non art sites. Again because something is freely available does not mean you have freedom to use it without attribution.
You have any proof for that?
They actually scraped the containers of every artist who uses Adobe products, as they had that claim in the contract that allowed to do that. IIRC
evil thought< : time to upload a billion cubes to behance, each slightly not-cubic.
1 Like
I thought you seemed like a nice person, then you write that. something only an evil genius would say 
Have you no consciousness 
1 Like
the better question is do i dream of electric sheep. mwahahhahhahaaaaa
Shouldn’t that be Baahahahahahaaaa
1 Like
@joseph I think that’s a good suggestion and would more likely guide people towards this thread. Any chance this thread’s title can be changed?
1 Like
If Blender Foundation ever would develop AI, it’s likely that (with current technology) the AI models would be offered as a separate download for users who desire this feature.
The most likely scenario, in my opinion, is that if AI were to become a feature in Blender, it would be offered as an add-on module, accessible through a script interface, much like you suggested.
This is indeed very possible, especially as computational capabilities continue to advance it is likely that AI software will be more advanced and become increasingly proficient in generating content with minimal training data.
1 Like
Sounds good. Updated!
2 Likes