Hitem3D is focusing on 3D printing for its image-to-3D AI system.
Image-to-3D systems have been around for about a year, with massive improvements in quality taking place. The idea is to present a 2D image of a subject to the AI model, which then “guesses” the hidden sides and generates what it believes to be a 3D model of the entire subject.
In practice, early image-to-3D systems really didn’t understand the backside of subjects, and they were often just rounded shapes to “fill in the gap”. These were effectively unusable for most practical purposes.
However, over time the AI models gained more smarts through extensive training. Nowadays most systems usually can achieve decent results on all sides of the 3D model.
Does this provide you with a properly printable 3D model? Well, sometimes. And sometimes not.
Very often 3D model generators are producing output intended for visual systems, like games, 3D environments, virtual reality, etc. These are areas where the concept of “printability” really doesn’t matter. There can be discontiguous segments, holes in the model, 2D parts and more that would work in, say, a VR environment, but would absolutely not be 3D printable.
Most of the image-to-3D AI services do not recognize printability aspects, as they were originally intended on generating virtual assets. That’s a huge market, so don’t blame them for missing the printability features.
Now I see a press release from Hitem3D, a Chinese startup that offers a powerful image-to-3D service. I last wrote about them in January, when they released version 2.0 of their AI generator, which provided significantly more detail.
In the release, Hitem3D describes a new focus on printability. They write:
“Hitem3D’s evolution illustrates this shift. By strengthening high-resolution geometry generation and prioritizing structural coherence over purely aesthetic output, the platform positions AI-generated models as viable starting assets for physical production rather than experimental prototypes. This approach aligns with how professional users evaluate tools in practice—by how confidently a model can be prepared for printing without extensive correction.”
In other words, they are going for printability.
A look at their website confirms this. The opening tagline is “AI that makes your ideas printable”. As far as I know, no other image-to-3D service has this focus. They add:
“Upload an image, Hitem3D instantly generates a production-ready 3D model for 3D printing with effortless precision.”
And, unlike many of the competitors, they offer download in STL format. Many image-to-3D services only allow downloads in virtual asset formats, like GLB, FBX, or others, and therefore you have to convert the file into a printable format.
This is good news for 3D printer operators who have been swamped with countless image-to-3D services that mostly don’t fully address their needs. Here we see at least one service that recognizes the need for printability and has reorganized their service around that goal.
Via Hitem3D and PRNewswire
