AI image to 3D model
The converter below builds geometry from brightness. This page does something different: an AI model reads the picture, works out the shape of the object in it, and builds a mesh you can print or open in Blender. Upload one photo, choose STL or GLB, and rotate the result before you download it.
Advanced options
Only useful for GLB in a PBR renderer such as Blender or Unity. It makes generation noticeably slower and changes nothing in an STL print — leave it off unless you need it.
Choose one clear subject shot — a single object against a plain background gives the cleanest mesh.
Your 3D result appears here — drag to rotate.
This mode is different from our other tools: your image is sent to a 3D generation server to be modelled, then the file comes straight back to you. Nothing is kept after that, and no account is required.
What an AI image to 3D model actually does
An image is flat. To get a printable object out of one, something has to invent the third dimension. Two approaches do that, and it is worth knowing which one you are using.
The browser converter (photo to STL) reads brightness as height. Dark pixels become thick walls, light pixels become thin ones. That is exactly what you want for a lithophane, a logo plaque or a relief map, and it is why that tool loads instantly and never asks you to sign up.
The AI mode on this page is different: a generative model looks at your photo, guesses the object behind it, and produces closed geometry with a front, a back and sides. A photo of a mug comes back as a mug. A photo of a friend comes back as a bust. That guess is the whole value, and it is also where results vary — the model cannot see what is not in the picture.
Picking a photo that comes back well
Three things matter more than anything you can set in a menu:
- One subject. Two overlapping objects confuse the shape estimate. Crop first, upload second.
- Clear silhouette. Bright subject on a plain background beats a busy scene every time. This is why product photos convert so much better than holiday snaps.
- Even light. Hard shadows are read as geometry, so a strongly side-lit face tends to come back with a dent.
If the first result disappoints, the usual fix is a tighter crop, not a better model.
From generation to print
STL out of an AI model is rarely printable as-is: it is dense, it can have thin limbs, and it is often too small or too large in the slicer. The normal workflow is to open the STL in your slicer, scale it to the size you actually want, then repair or remesh if the walls are under about 0.8 mm. Bambu Studio, PrusaSlicer and Cura all report thin walls before you print, so check the preview rather than trusting the file.
For anything graphic — a photo, a logo, a line drawing — the browser converter will usually beat the AI model, because there is nothing to guess. Use AI when the subject is a thing, and use the height-map tools when it is a picture.
When the AI mode is the wrong tool
- Text and logos: height-map mode keeps every letter crisp; AI mode rounds them off.
- Photos of people for lithophanes: use photo to STL, it is the intended use.
- Anything you need in seconds: the browser tools are instant, AI runs a queue.
Questions
How is this different from the free converter on the homepage?
The homepage converter maps pixels to height — perfect for photos, logos and lithophanes, and it runs entirely in your browser. The AI mode works out a three-dimensional shape instead, so a picture of a shoe or a figurine comes back as an object you can spin around, not a flat relief. It needs a server, so it needs a connection.
Which photo works best?
One clearly lit subject, filling most of the frame, against a plain background. Products, toys, statues and pets do well. Wide landscapes and blurry snaps come back as rough blobs — that is a limit of every image-to-3D model, not just this one.
STL or GLB?
Choose STL if the file is going into a slicer for 3D printing: it carries geometry only. Choose GLB if you want the texture carried along, for Blender, Windows 3D Viewer, Unity or a web viewer. STL also gets the interactive preview on this page.
How long does it take and how often can I use it?
Generation usually takes a minute or two. The service runs one job at a time, so a short queue can form when the site is busy. Free generations are limited per visitor per day to keep the bill payable — the counter resets every day.
Can I sell the models I generate?
Check the terms of the model provider behind this tool before selling anything. We pass the generated file straight to you and claim no rights in it, but the underlying AI service has its own terms.
Other converter modes
Photo to STL
Lithophane mode: brightness becomes wall thickness, so the print only reveals the photo when backlit.
Open photo mode →PNG to 3D model
Plaque mode: a transparent PNG becomes a raised cut-out on a solid base.
Open plaque mode →Image to OBJ
Relief mode with OBJ export, for software that prefers OBJ over STL.
Open relief mode →