• Win a Free Custom Engraved Brass Coin!!!
    As a way to introduce our brass coins to the community, we will raffle off a free coin during the month of August. Follow link ABOVE for instructions for entering.

AI for creating 3D models

E

EspenT

Joined
Oct 23, 2024
Messages
74
Points
88

Location
Selbu, Norway
Are there any good AI engines out there that can create models for 3D printing?

It would be fun to try to create some sailors or other things to add some dynamics to the models.

I don’t have the skill, interest or time to make them from scratch.
 
That would be interesting. 3D drawing has been around long enough to be old school for the folks that are skilled in it (not me:)) so maybe some other easier methods are coming out.
Allan
 
Did some quick research and there are several programs/sites that look promising and not expensive. Does anyone have experience with the following:
  • Tools and Platforms
    • Meshy AI:
      Offers a user-friendly interface with a multi-view option to upload multiple images of an object and generate a 3D model.

    • Rodin:
      An AI generator that supports multi-view input for creating 3D models from several images of the same object, providing more accurate mesh details and textures.
    • TripoSR:
      An advanced open-source AI model that can generate detailed 3D models from 2D images. It often has a Gradio interface for easier use and allows for direct installation and operation on a computer.
      THANKS,
      Allan


 
I tested Meshy for an hour today. It is very easy to get going and produces good looking models.

Some problems I have faced, but could just as well be down to user errors:

Creating a model of f. ex a captain is easy. You can either describe it with words or have it use a reference photo. The issue comes when you want the rest of the crew. It is difficult to get consistent and reproducible results. They don’t get an overall similar look.

It also gets tricky if it looks very good, but you want to change a specific area of the model.

Since the development in AI is the only thing that is faster than light, this could already be improved from earlier today

I will play around with this and see if I can get better results.
 
From the prompt "English naval officer from age of sails"

IMG_2729.jpeg
IMG_2730.jpeg
IMG_2733.jpeg

I do not have a paid version yet, so I cannot export the file and try to print it.

normally you would just get a grey model, but I asked it to add texture just to see the result.

On a 1/64 scale model these guys will be about 30 mm in height so I am not worried about details on the model. They won’t be noticeable anyway.

Edit: I forgot to write that I also added a reference photo to this version.
 
Interesting! I wonder if the resulting mesh file will print. Because that’s often a problem. But I read that there are there are online tools to check and repair the files.
 
My 3D modeling instructor @ Universal Studios predicted AI-created modeling two decades ago. Most of the students in my class, (over twenty years younger than I) were too young to understand the implications of what he was saying.

Making a stand-alone small shape, (like a posed character) is one thing; AI modeling can do it with the correct inputs. As for larger parts that require precise fitment, the capability isn't going to happen soon, so CAD design is safe for the time being. Currently, it can take years for a team to design/test/revise/test and finalize a complex design and it isn't cheap. R&D is why a single aircraft can cost in the tens and hundreds of millions of dollars to produce. The materials are a small part of the equation. Cost savings will drive this trend, and if your competitor gets the capability first, guess who will be the favored vendor when bids are generated?

It simply isn't financially sustainable to build weapons in quantity like stealth aircraft, naval ships and submarines that can be destroyed by far cheaper weapons. The war in Ukraine is proof enough. Three years of fighting and a far superior force cannot establish control of the air over the battlefield, which has led to massive losses. Current US doctrine is to establish air supremacy before ground and naval surface forces advance. That doctrine is in doubt until we can neutralize small drones efficiently. But I digress. . .

Eventually, human-led AI will replace human CAD jockeys, who will specify requirements based on the tasking and let the AI generate multiple designs for evaluation. Today's young, mechanical engineers will likely see this capability become an indispensable tool during their careers in large corporations by the time they become Program Managers.

Don't give up learning 3D modeling just yet!
 
I can only imagine this for now, but I keep thinking of parts such as port hinges, or pintles and gudgeons and similar parts being set up as base designs then a pre programmed set of scantlings for a given era and ship size in a library of sorts. Pick and choose the era and rate of ship, click on it and either a 3D STL drawing or some such thing can be downloaded and sent via email to any 3D printshop to have parts made. It is done now very easily with 3D drawings but a library that anyone can use seems possible. For cannon, it is cheap ($0.75) to print a perfect one now and there are many dozens of drawings available for free that have been discussed here at SoS but a more complete library with caliber, length, pattern/era, nation seems achievable even now. Even if it is not there yet, it is nice to dream as they are accurate, so much better in appearance overall and less expensive parts than having to use aftermarket sources that provide minimal choices at ridiculous prices. I prefer wood for parts that were wood, but many folks are good with parts that were really metal but printed in resin for the model. Sorry to carry on but the future is here and really fascinating.
Allan
 
AllanKP69 said:
I can only imagine this for now, but I keep thinking of parts such as port hinges, or pintles and gudgeons and similar parts being set up as base designs then a pre programmed set of scantlings for a given era and ship size in a library of sorts. Pick and choose the era and rate of ship, click on it and either a 3D STL drawing or some such thing can be downloaded and sent via email to any 3D printshop to have parts made. It is done now very easily with 3D drawings but a library that anyone can use seems possible. For cannon, it is cheap ($0.75) to print a perfect one now and there are many dozens of drawings available for free that have been discussed here at SoS but a more complete library with caliber, length, pattern/era, nation seems achievable even now. Even if it is not there yet, it is nice to dream as they are accurate, so much better in appearance overall and less expensive parts than having to use aftermarket sources that provide minimal choices at ridiculous prices. I prefer wood for parts that were wood, but many folks are good with parts that were really metal but printed in resin for the model. Sorry to carry on but the future is here and really fascinating.
Allan
Click to expand...

i agree with your imagination of a library of 3D printable parts. with AI anyone can create 3D parts and send them to be printed. Cannons, rigging blocks, figureheads and all sorts of fittings can be done. I have been exploring the idea of stern kits with 3D STL files that can be downloaded and printed by the builder or a print service. I believe we will see big advances in this hobby within a few years. Rigging blocks can be printed in perfect details for 5 cents. cannons for .50 cents. right now it is the high cost of 3d artists creating the original 3D print file but as AI gets better and better they will become obsolete. Like everything else as a new idea is introduced into the market the price is quite high but in time as more and more businesses get involved the prices begin to drop.
I think we are reaching the limits to the cost of kits and the solution is kit builders will begin to create their own kits from a variety of sources
Think of it this way any business creating a complete kit will require a shop to mill all the wooden planking, a laser to cut parts, a 3d resin printing set up, plus the people to created the print files, shop workers, someone to run the laser cutter.
But if you specialize in 3D printing parts for kits that is all you have to invest in, same with wood milling if that is what you do then there is no big investment in printing or casting fittings just mill work. AS a library of files increases all the printer has to focus on is printing.
so i agree the future of the hobby is in the creation of a 3D print library and small specialize operations that do not require large investments.
expensive kits might be the death of the hobby
 
Last edited:
You must log in or register to reply here.
Back
Top