A 3D model built from several photographs usually holds its shape better than a mesh guessed from a single still. The free multi-view to 3D tool on CanvaPro turns photos of the same object into a GLB you can preview and download. You do not need a desktop photogrammetry suite or an account, and the usual file does not pick up a watermark from the page.
Front is required. Back, Left, and Right are optional, and you should add them when you have matching shots. Each photo can be a JPG, PNG, or WEBP of 6 MB or less. The form also exposes three reconstruction controls: octree resolution, from 64 to 512 with a default of 256; number of chunks, from 1,000 to 500,000 with a default of 20,000; and target face count, from 1,000 to 100,000 with a default of 10,000. Complete the security check and click Generate Multi-View 3D. Jobs often take a few minutes. When the queue finishes, preview the mesh and download the GLB.
How this differs from a one-photo mesh
The image-to-3D page on CanvaPro builds a geometry mesh from a single still with Hi3DGen. This page is for several views of the same object, with more control over resolution and face count. Use it when you can photograph the front and at least one other side. Use the single-photo tool when all you have is one picture and you only need a rough blocking shape.
The photos should show the same object, with similar lighting, scale, and pose. A front shot in daylight and a side shot in a dark room will fight each other. Do not mix two different items, and do not change the arrangement of parts between frames.
If you only have a single still, turn the photo into a 3D model is the simpler, one-image path.
How to build the model, step by step
- Photograph or collect matching views. Front is required. Add Back, Left, and Right when you have them, and keep the object in the same place in the frame as much as you can.
- Upload JPG, PNG, or WEBP files of 6 MB or less in the matching slots, and only use photos you have the right to work from.
- Leave the defaults unless you have a reason to change them, or raise the face count for a print or a hero prop, or lower it for a simple game object.
- Complete the security check and click Generate Multi-View 3D once. Leave the tab open for a few minutes, then orbit the preview and download the GLB if the shape is useful.
Upload the front view and any extra angles, keep the defaults unless you have a reason to change them, and run the multi-view 3D tool once per set of photos.
When to change the defaults
For most objects, 256 octree resolution, 20,000 chunks, and 10,000 faces is a sensible starting point. Raise the target face count toward the top of the 1,000–100,000 range when you need a denser mesh for a close-up or a print. Lower it when you want a lighter prop for a real-time scene. Higher octree values and chunk counts can capture more detail and can also take longer. If you are not sure, run the defaults first and only then experiment.
If you only have one photograph and you cannot shoot the other sides, open the image to 3D tool instead of inventing extra views. A honest one-photo mesh is often better than a multi-view job fed with mismatched pictures.
If the mesh looks collapsed or uneven
Check that every upload is really the same object, that the backgrounds are reasonably simple, and that no file exceeds 6 MB. Missing sides will be inferred. A front-only job can still run, but the back will be a guess. Reshoot rather than stretching a blurry frame. Keep the camera distance similar across views so the scale does not jump.
Uploads travel over HTTPS. Files are processed to complete your request and are not sold.
Inspect a GLB that does open with the 3D file viewer before you decide the shoot is unusable.
After you download the GLB
Open it in your viewer or engine and inspect the sides you did not photograph as carefully as the front. Check scale and orientation early. A mesh that is useful for a web preview may still need cleanup, retopology, or new materials before a game or a print. Name the file with the object and the views you supplied, and keep the source photos in the same folder.
If you later capture a cleaner back or side, run a new job that includes those shots instead of trying to sculpt a missing side by hand on a weak first mesh. The second reconstruction is often faster.
Orbit the download locally with the steps to view a GLB or GLTF file in your browser before you convert it any further.
A short checklist
- Upload a front photo, and add back, left, or right views of the same object when you have them. Each file can be a JPG, PNG, or WEBP under 6 MB.
- Start with the defaults: octree 256, 20,000 chunks, and 10,000 faces, unless you need a denser or lighter mesh.
- Click Generate Multi-View 3D once and wait a few minutes for the preview.
- Download the GLB if the shape is useful, and keep the source photos with it.
A small, consistent set of photographs will beat a large, messy set. If you can only light and shoot the object once, take the extra minutes to walk around it and capture the sides while nothing has moved.
When you are ready, start with a clean front view and any matching sides you have, run one reconstruction, and save the GLB only after you have orbited the preview.