Image2model with Tripo3D and Blender

Sometimes you want a drawn or AI-generated character to become real—even if it’s just a small plastic figurine. Today it’s easier than it seems: you can turn a 2D image into a three-dimensional model in a few minutes.

In this post I share my process: from a little dragon picture to a finished figurine.

Source image

I generated it in ChatGPT with the following prompt:

an image of something interesting, attractive, unusual, and cute on a white background, which can be easily transformed through text2model AI into a 3D model for easy 3D printing

That is, literally the first thing that came to mind for a demo. The options are quite diverse, but all interesting in their own way. Just look at this snail with phantom eyes:

You can use any picture where the object is isolated and clearly visible. A white background is not required, but if you’re creating a new image, it can make the next steps easier. Sketches, artworks, and AI generations will do. I took an image of a small cartoony dragon (because my girlfriend liked it the most):

Generating a 3D model with Tripo3D

Upload the image to Tripo3D—an online service that turns a 2D picture into a volumetric model. It also amusingly shows an intermediate model as a point cloud in 3D.

A couple of minutes later we get the finished model:

You can rotate it, inspect it, and download it if necessary.

Editing in Blender

The downloaded model often contains artifacts: extra elements “hanging” in the air, dents, non‑manifold surfaces. They are convenient to fix in Blender.

I used basic tools: selection, deleting meshes, sculpting for smoothing. If you don’t feel like fixing things manually, you can try regenerating the model in Tripo3D. In general, such defects aren’t very noticeable or critical for 3D printing, but they bothered me.

Preparing for printing and the print itself

Export the model to STL, open it in a slicer, add supports, and start printing. If there are small details, it’s important to tune the layer height and speed.

My support settings also got messed up because I printed with variable layer thickness to reduce banding, so I had to suffer a bit.

Finishing touches

And so it wouldn’t look completely blank, I painted the eyes:

Takeaway

In literally an hour you can turn almost any idea into a physical object. The process is simple, requires a minimal set of tools, and just a little practice.

The barrier to entry in almost any field is now lower than ever thanks to the prevalence of AI. And it will drop even further—which is great.

Thingiverse model, if it caught your eye: https://www.thingiverse.com/thing:7131500

Also, check out a YouTube Short about the making: https://www.youtube.com/shorts/NcQOwbzL2jk