Image to 3D model, free, with STL for 3D printing
Make a 3D model from a photo. Upload one object and our AI works out the full shape, including the sides it cannot see, and you download it as STL for 3D printing, GLB for the web and games, or OBJ. No account, no watermark.
By Sam Ridder · Last updated: September 2026
No watermark, no account. STL, GLB and OBJ from one photo: 5 free models a day, the colours on the side you photographed.
What you get
A solid 3D shape of the object in your photo, which you can rotate right here on the page before you download it. The STL and OBJ files are sized so the longest side is 100 mm and the model stands on its base, ready to scale in your slicer.
The GLB and OBJ carry the colours of your photo, painted onto the shape. The back, which your camera never saw, gets colours estimated from the front, so it is less exact. The STL is the shape only: which is exactly what a 3D printer uses.
- Colours from your photo on the GLB and OBJ; STL is shape only, for printing
- STL for 3D printing, OBJ for 3D apps, GLB for the web, games and AR
- Repaired after it is made: loose fragments removed, holes closed where possible
- Tallest side 100 mm, standing on its base
- Free: 5 models a day, no account, no watermark
Photos that work best
One clear object, fully in frame, photographed from a slight angle so some depth is visible: a figurine, a mug, a shoe, a toy, a vase, a piece of furniture. We cut the object out of the background first, so a plain background helps but is not required.
People, animals and very thin or see-through things (glass, wire, hair) are the hardest. Groups of objects come out as one lump.
From a single photo the AI has to guess the back of the object. For simple and symmetric things that guess is usually good; for anything with a detailed back, expect it to be plausible rather than exact.
Printing it
After the AI builds the shape we repair it: small loose fragments are removed, holes are closed where possible and the faces are made to point the same way. The page tells you whether the result is a fully closed (watertight) mesh. If it is not, your slicer can usually repair it on import.
Check wall thickness and overhangs in your slicer before printing, as you would for any downloaded model: thin parts from the photo can come out thinner than your printer can make.
How long it takes, and the limits
About one and a half to two minutes per model on our own graphics card, plus any queue. Building a 3D shape is heavy work, which is why most sites that offer it ask for an account or credits.
Free: 5 models a day. A ShinobiTools pass gives 30 a day and works on our other tools too. Photos up to 20 MB. Your photo and the model are deleted after 45 minutes.
How to turn a photo into a 3D model
- Drop a photo of one object in the box above.
- Wait while our AI cuts out the object and builds the 3D shape: about two minutes.
- Rotate the model on the page to check it.
- Download STL for printing, GLB or OBJ.
Only need a printable file? See image to STL. Looking for an AI 3D model generator? It is this same tool. No photo? Text to 3D model starts from a short description.
Related: Text to STL free.
By Sam Ridder, I build and run PicReviver on my own hardware, on my own. Who I am. I tested this on a series of real objects, a porcelain figure, a teapot, a garden gnome, a helmet, before putting it online.
Real examples: photo in, 3D model out
Picture to 3D in practice: these are three public-domain photos of single objects, each one run through the tool on this page on 18 October 2026, exactly as a visitor would. Left is the photo we uploaded, right is the model viewed from the front. The three runs took between 37 and 67 seconds; with a queue, plan on the usual one and a half to two minutes.
The front matches the photo closely, including the colours. The back was never photographed, so it is the AI's estimate, and fine relief such as the pattern on the mug comes out softer than in the photo.
Sources
- STL (file format): STL stores surface geometry only, without colour
- glTF: Runtime 3D Asset Delivery: GLB is the binary form of glTF 3D files
- Wavefront .obj file: OBJ is a 3D geometry definition file format
Working with old family photos as well as objects? The same site can colorize old black and white photos for free: upload the scan, and the colour comes back automatically, with no account and no watermark.
Can I turn a picture into a 3D model without signing up?
Yes. Upload one photo of one object and you get STL, GLB and OBJ about two minutes later, with no account and no watermark. Five models a day are free.
Can I make an STL file from a photo for 3D printing?
Yes. The STL is the shape only, sized so the longest side is 100 mm and repaired after it is made. Check the wall thickness in your slicer before you print.
Can I use the 3D model in Blender or a game?
Yes. GLB works for the web, games and AR, and OBJ opens in Blender and other 3D apps; both carry the colours of your photo.
Common questions
Is this image to 3D converter free?
Yes. Five models a day are free, with no account and no watermark. A ShinobiTools pass gives 30.
Can I make an image to 3D model free of charge?
Yes: five models a day, with no account and no watermark. A ShinobiTools pass gives 30 a day. The free models are the same quality as the paid ones.
How does photo to 3D model conversion work?
The AI cuts the object out of the background, works out its full shape, including a guess for the side the camera never saw, and builds the model from that. You get STL, OBJ and GLB from one photo, about two minutes later.
Can I 3D print the model?
Yes. Download the STL. It is sized to 100 mm on its longest side and repaired after it is made. Check wall thickness in your slicer before printing, as with any downloaded model.
Does the 3D model have colours?
Yes, the GLB and OBJ carry the colours of your photo. The back is estimated from the front, so it is less exact than the side you photographed. The STL has no colour, because 3D printing only uses the shape.
How does it know what the back looks like?
It doesn't: from one photo the AI makes an educated guess. Simple and symmetric objects come out well; detailed backs are plausible rather than exact.
What kind of photo works best?
One clear object, fully in frame, seen from a slight angle. People, animals, glass, wire and hair are the hardest.
How long does it take?
About one and a half to two minutes per model on our GPU, plus any queue.
Which file formats do I get?
STL for 3D printing, OBJ for 3D software and GLB for the web, games and AR: all three from one upload.
What happens to my photo?
Your photo and the model are deleted after 45 minutes. See privacy.