Having models that have been proven to be printable is one of the biggest premises that make MW unique. Creators who don’t use Bambu Studio can still print out their model and take a photo. So not having a print profile is no excuse to not test print their model to make sure it’s printable.
Arguing that renders should be allowed but then saying non-3d printed object should not be allowed is not consistent. Renders are non-3d printed objects. It gives out an impression that some creators who use renders are asking MW to crack down on all other types of non-3d objects so they can enjoy a monopoly, or at least carve out an exception for them just because they use renders. In other words, rules are good when applied to others, but bad when applied to us.
If the intent is to deceive downloaders it is wrong. No one wants to waste time downloading models that don’t print properly, or models that don’t look like the pictures. Gaming the community to get more points/money/stuff/fame without providing real value is unethical.
That’s it. That is the whole issue.
If this was a community devoted to woodworking, would an AI generated set of plans, or an AI generated image of a completed project be ok with other woodworkers? Of course not. Why would I get out my saw and start cutting wood if I wasn’t sure the piece of furniture had ever been made? If I didn’t get to see how it actually looked? If I was intentionally tricked, I’d be even more angry.
3D printed objects are physical objects in the same way a table, chair or bookshelf are. They are physical objects that require work to envision, design, test and then publish. And, it is OK that talent, skill, practice and investment of time are required on the part of the designer.
If a person doesn’t know how to take good pictures, learn. Start with crappy ones. They will get better.
If a person doesn’t know how to model, learn. If you want to use AI to make models themselves, no one can stop you. But, you ought to be able to print them and have them work. Then, take a picture of them.
It really isn’t complicated, and it need not be personal. It is simply about a marketplace that offers the ‘plans’ for real objects. So, the objects ought to be real, and Makerworld ought to enforce that.
That’s because due to your hard stance you can’t see my point, while I fully understand yours.
After all - whole McMaster-Carr catalogue is mostly renders and nobody would argue those are deceptive.
Anyway, I’d rather side with Malc’s suggestion to mark renders as such than to throw good use of renders just because bad actors can and do use them deceptively.
I read the discussion, that’s why I noticed the inconsistence. If creators who argued for renders also argued for the use of other types of non-3D printed images/objects, they would have had a stand on this subject.
For example, If they had said that AI generated images should be allowed as cover as long as they are labeled as such, and don’t throw out good use of AI just because bad actors use them, their arguments would have been based on a stand. But so far all I see are shifting arguments, not a consistent point of view.
i’ll give my take here, rendering is not in the same ballpark as an AI image · AI is an algorithmically created 2d image trying to represent a 3D object, a rendering is achieved after you have created the 3d object, and it’s a matter of giving it light / color and background in a digital way, however it is an addition to an existing design, it’s not the same thing at all
that being said, i’m not a fan of renderings either personally, BUT, if they are accompanied with a photo of the printed object then it’s a different story and i don’t see why they should not be allowed. I see that SabreDesign was already mentioned on this thread, and his renders are quite good, but the printed design comes out with the same characteristics that are shown on the render and they are almost identical · only the layer lines aren’t clear, but that’s fine, as we should all know and embrace that FDM will create layer lines
There’s no inconsistencies in my stance that cover image should correspond with the model and resulting print.
Your example with AI would in light of that be incorrect because AI image cannot faithfully represent real object.
On the other hand renders of the 3D model perfectly correspond to the model.
Also - non-3D-printed objects on the cover, like say porcelain object that then is modelled by hand, do not correspond to the model because they are inspiration for the model and most likely the scale is all wrong (in one instance I’ve seen that a porcelain object was HUGE while 3d print inspired by it was just big enough to fit in the hand). The one instance where I think a photo of object would make sense would be when the object was scanned with high-quality 3d scanner and the resulting print is 1:1 replica (but then it should be marked as such).
They are all non-3D printed images/objects. If one type is allowed, others should also be allowed. AI generated images is technically a render, rendered by an AI engine, instead of a CAD engine. BTW, pretty soon, these two will be one and the same. CAD will include some much AI that the distinction will be meaningless.
So are renders, which is a 2d image trying to represnent a 3D object. A render is not a physical 3D object one can touch.
The argument should be based on a stand, not based on what’s good for someone we know, or for us. This is the source of shifting arguments.
They should be allowed, but not as the cover photo.
It’s really simple:
If a model has been printed (which is the only way to prove that it can be printed), take a photo of that printed model to use as the cover. Use renders, AI generated images, or photo of a porcelain toy that the model may have been based or scanned, to make your point if needed.
If a model has not been printed, or for 3D model artists who have no interest in doing 3D printing, it’s not the end of the world. There are other ways to properly go about it:
Publish it on platforms that don’t emphasize on model printability. Once the model has been printed by users, ask for their permission to use their photos to use as cover and publish on MW.
Ask a member of the community to partner with the artist to test print and take a photo to use as the cover.
A render is based on the design. An AI generated image is compiled from already existing models found in the internet and stitched together in a way to make it look good.
Renders show functionality, AI generated images don’t. Renders show dimension accuracy and sometimes angles that otherwise a camera wouldn’t be able to get. AI generated images show the beauty of software “imagination” only but most of the time not possible to be used.
I agree in some parts but in 99% of what you say, i have to disagree. The point is that i think you don’t understand how renders and ai generation work.
A render is a preview of the already finished model, with all the vertices and polygons in the mesh. It’s just a coating, if well applied, of how it will look like.
Ai generation is a compilation of already present models available in the internet. I can also tell you that the AI must be trained with those models to be able to stitch parts of it together, depending on your prompt.
Example: “create a modern moon object that can hold a tealight”
The AI will then search and compile from all models that had moon and tealight in the name/description and will try to create an image with all the characteristics of already existing models. It will pick images from that ai guy and his sad attempt of a moon, will look for images of tealights holders and some other models and try and make something from it. It’s also called copyright infringement because it uses, most of the time, models that are protected by a license.
AI cant generate anything that don’t yet exist while a render cannot be created without a designed model. Ai needs existing designs printed while a render needs a design. Those are the biggest differences.
Now back to other topics
Edit: Not to defend any of those practices, but sometimes a render serves a big purpose. I’ll leave this example where i could not show on pictures all the details and yet, since it’s a render, the accuracy and dimensions are exactly the same. Also have a picture to show it it’s functional part.
Let me get this out first: I am not using AI to generate models, becuase it’s not yet good enough for it. So I am not making the argument for my own interest. In fact, what MW decided on is very close to one of my suggestions on how to handle AI generated models/images
I am not arguing that renders shouldn’t be allowed. I am saying it can’t be used as the cover image.
Without going too much into what AI is and isn’t, I would say that I know something about AI. I use it daily on quite a number of things. From what I know, the arugment to allow renders but not allow AI generated models and images is not based on a principle, but rather based on personal preference or interest.
you make it sound like you don’t really know a difference between AI and a rendering, it’s completely different processes, they aren’t remotely similar
edit/ and let’s use AI as it’s good for this kind of stuff
1. Source and Creation Process
AI-Generated Image:
Created using machine learning models trained on vast datasets.
Output is often a 2D image generated from a text prompt or parameters, with no underlying 3D structure.
The process is typically less labor-intensive and requires minimal technical expertise compared to 3D modeling.
3D Render (e.g., from Fusion 360):
Built from a detailed 3D model designed by the user.
Requires technical knowledge of CAD software and involves precise control over dimensions, materials, and other physical properties.
The output is a rendered visualization of the model, which can include photorealistic lighting, textures, and environmental effects.
2. Level of Customization
AI-Generated Images:
Customization is guided by the user’s prompts but limited to what the AI model can interpret and produce.
It can lack precision in areas like specific measurements, fine details, or consistency across iterations.
3D Render:
Highly customizable with exact control over dimensions, proportions, and appearance.
Precision is essential for engineering, manufacturing, or other practical applications.
3. Output and Real-World Applications
AI-Generated Images:
Used for conceptual art, mood boards, or generating visuals for non-technical purposes.
Not inherently linked to any physical or functional model.
3D Render:
Often linked to physical products, prototypes, or engineering designs.
The rendered image is a visual representation of a real-world object that can be fabricated or manufactured.
4. Quality and Realism
AI-Generated Images:
Can achieve photorealistic quality, but may sometimes introduce inaccuracies or “artifacts.”
The realism depends on the model’s training and the user’s skill in crafting the prompt.
3D Render:
Realism is determined by the user’s skill in setting up materials, lighting, and rendering parameters.
Typically more controlled and predictable, especially for achieving specific photorealistic effects.
5. Flexibility and Versatility
AI-Generated Images:
Versatile for generating a wide range of artistic or conceptual visuals quickly.
Limited in creating outputs that require technical accuracy or precise structural details.
3D Render:
Versatile within the constraints of the model; it’s purpose-driven and designed for practical use.
Offers less flexibility in conceptual designs compared to AI but excels in technical precision.
Summary
The comparison depends on the intended use:
AI-generated images are great for artistic and conceptual visualization.
3D renders are indispensable for engineering, product design, and manufacturing.
If the goal is to visualize a functional product, 3D modeling and rendering are superior. If the aim is creativity or ideation without physical constraints, AI-generated images can be a faster and more accessible option.
I also didn’t meant to say what you think i am saying or defending. I also use AI, both locally and on the cloud, since the first days it came out. It’s cool for some use cases as programming but not yet there for modeling. And to be honest, it will never be because it infringes on copyright since it uses other models as “inspiration”. Every AI generation, be it in 3d models or not, is illegal. Even using AI code to program is illegal and morally wrong because it uses examples from where? From already available code on the internet. It’s a simple “crawl the internet for X” and that is what the trained model will show.
I have trained AI models on my face and body for months and it still can’t create a reproduction of me, not even 60% close.
I am also not defending it in my own interest. The example i gave you was honest. I could have chosen not to post it and that was it. If i tried and generate an AI image of what i wanted, trust me, it wouldn’t be the same because i would not model it exactly as the “AI” imagined it.
A better way to present one’s opinion in this situation will be to start with “To me, the difference between AI and rendering …”. Starting with a statement questioning or trying to put down other’s knowledge is probably not a good way to participate in a discussion.
sorry if it sounded harsh to you, but we are talking about facts and not opinions, it’s not a “to me” thing… and i’ll leave it at this, you can continue believing whatever you want, that’s ok for me
I’ll just leave this here since it all began exactly with this clear example of what is creating an image from AI and passing it to a functional model that looks to what is promised. A render can only be deceiving if it’s not based on a self made design or if made without any regard for the model while AI generation can be deceiving just by looking at the image itself.
Now i’ll let myself go out of the discussion, for the time being, because i am still waiting on those “moderated” models and moderated feed that @MakerWorld promised to improve
Don’t worry about it. That’s just an advice or reminder, as I have seen too many threads getting too personal fast to the point that they are no longer an exchange of views but rather each trying to put down the other side.
When the day comes and this platform starts to bring in rules like “you are only allowed to have actual photos as covers” this would be against my workflow and I would leave. I am no full-time creator, just a consumer hobbyist. Rendering gives me the possibility to have a nice collection with uniform pictures also while working in my evening leisure time with bad light conditions. I strongly agree on the rule for an actual print image to be attached, makes absolutely sense.
Also, I feel bad for new users of 3D printing, which are getting tricked by nice renderings with faked scenery. That honestly makes me sad.
However, it would make more sense to rely on the model ratings instead of community witch hunters. There is actually mostly unbiased data which can be primarily utilized for such purpose. And it would remain the creators’ freedom to act how they like. And as exposure is already manipulated by the platform, why not using also this data for doing so.
I am only using rendered images for my mostly functional prints. I also avoid doing any fake scenery. But when such a rule is established, I also would be affected as, in my opinion, collateral. And there are many other users like me.
You have also rendered pictures in retail for OEM products. Most of you would never come to the idea and complain about this…
This forum is meanwhile pested by jealous people being mad at each other. And Makerworld becomes more and more a slave of them. That’s my opinion, and I am fine if you don’t agree
