Meshroom (AliceVision)

Meshroom is free, open-source photogrammetry software. It turns a set of ordinary photographs into a textured 3D mesh. It is developed by the AliceVision project and runs its computer-vision pipeline through a node-based interface, so you drop photos in, press Start and get a 3D model out. It is aimed at 3D artists, researchers, educators and makers. It is not a dental program, but dental researchers have tested it for digitising physical casts.

If you are looking for a “Meshroom download”, the official builds are for Windows and Linux. The current release line is Meshroom 2025.1, whose main release (2025.1.0, 19 August 2025) needs an NVIDIA GPU with CUDA for the full-quality reconstruction. Without such a GPU you can still build a lower-quality “draft” mesh on the CPU.

Meshroom at a glance

What it does

• Runs a full photogrammetry pipeline: feature extraction, image matching, structure-from-motion (camera positions), depth maps, meshing, filtering and texturing.
• Reads common photo formats (JPEG, TIFF, PNG, EXR and several camera RAW formats) and video files.
• Exports meshes as OBJ with texture maps (JPEG, PNG, TIFF or EXR), and point clouds and cameras as PLY or Alembic.
• Uses a node graph you can edit, for example to decimate or smooth the mesh, change texture size or rerun a single step.
• Offers a draft meshing mode that works on the CPU when no NVIDIA GPU is available.
• From 2025.1 it has a plugin architecture. AliceVision is one plugin (photogrammetry, camera tracking, HDR panoramas, LiDAR meshing), and extensions add, for example, semantic segmentation.
• Can run from the command line for batch processing.

How dental teams use it

Meshroom has no dental features. Its dental relevance is photogrammetry research and education. An in-vitro study published in The Journal of Prosthetic Dentistry (Buzayan et al., online 2023) compared Meshroom with commercial photogrammetry programs and conventional impressions for making digital casts of a nasal defect for maxillofacial prostheses. Meshroom produced usable casts, but they were less accurate than those from the two best commercial programs. That is a fair summary of where it stands: free and capable, but not a replacement for a lab scanner.

Digitising a stone cast or teaching specimen

1. Put the object on a turntable in even, diffuse light. Matte, textured surfaces reconstruct best; a plain white gypsum cast may need a light pattern (e.g. a thin spray) to give the software features to match.
2. Take several dozen sharp, strongly overlapping photos all around the object from two or three heights, keeping focus and zoom fixed.
3. Drag the photos into Meshroom, save the project, and press Start.
4. When it finishes, open the textured OBJ from the Texturing node’s output folder.
5. Scale the model: photogrammetry has no real-world size by default, so measure a known distance on the cast and scale the mesh in a mesh editor.
6. Clean and export to STL in a mesh tool if you need it for 3D printing or comparison.

Teaching 3D reconstruction

Because the node graph shows every step, Meshroom works well for showing students how camera positions, depth maps and meshing lead to a surface model. Researchers can then compare that model with a scanner model.

Install and first run

Windows: the release is a zip archive. Unpack it to a folder with a short path and no special characters, then run Meshroom.exe. No installer is needed.

Linux: unpack the tar.gz and start the Meshroom launcher from the folder.

macOS: there is no official Mac binary. The manual lists macOS as “some work required”, meaning a source build, and dense reconstruction needs CUDA, which current Macs do not provide.

Requirements: the manual’s minimum is Windows x64 or Linux, a recent Intel or AMD CPU and 8 GB RAM. It recommends a Core i7/Ryzen 7-class CPU, 32 GB RAM and a recent NVIDIA GPU. Release 2025.1.0 is built for CUDA 12 and needs compute capability 5.0 or higher. Allow plenty of disk space: each project stores large intermediate files.

Gotchas: blurry, poorly overlapping or shiny photos give holes or failed reconstructions. If the GPU is not supported, the depth-map step fails; switch to draft meshing.

Limitations

• For full quality it needs an NVIDIA CUDA GPU, and there is no official macOS build.
• Models have no metric scale until you add one, and accuracy depends heavily on photo technique.
• Output is OBJ/PLY. STL export and clean-up need another tool.
• It is not suitable for intraoral capture and has no clinical validation.
• Processing can take a long time on large photo sets.

Meshroom vs alternatives

Among photogrammetry tools, COLMAP is another free, open-source option, and Agisoft Metashape and RealityCapture/RealityScan are commercial ones. In the study mentioned above, Metashape and 3DF Zephyr were more accurate than Meshroom. After reconstruction, OpenFlipper can repair holes, decimate and smooth the mesh, and ParaView can compare or visualise 3D data. If your source is a CBCT scan rather than photos, you need segmentation software instead, not photogrammetry.

FAQ

Is Meshroom free?

Yes. Meshroom is free and open source under the Mozilla Public License 2.0.

Is there a Meshroom download for Mac?

No official one. Releases are for Windows and Linux; macOS requires building from source and cannot use CUDA for the dense reconstruction step.

Can I use Meshroom without an NVIDIA GPU or CUDA?

Partly. The manual says you can use draft meshing, which runs on the CPU, but the full-quality depth-map step needs a CUDA-capable NVIDIA card.

What are Meshroom’s requirements?

At minimum Windows x64 or Linux, a recent CPU and 8 GB RAM. For good performance, 32 GB RAM and a recent NVIDIA GPU; 2025.1.0 needs CUDA 12 and compute capability 5.0 or higher.

What is the difference between Meshroom and AliceVision?

AliceVision is the photogrammetry and computer-vision framework; Meshroom is the application and node-based interface built on it.

Meshroom vs COLMAP or RealityCapture?

COLMAP is also free and open source. RealityCapture and Metashape are commercial. Meshroom’s strengths are its visual node graph and a complete photo-to-textured-mesh pipeline in one free package.

How do I use Meshroom?

Import a set of overlapping photos, save the project, click Start, and open the textured OBJ produced by the Texturing node when the pipeline finishes.

Last checked against official sources: 29 September 2026 (developer website: alicevision.org). Versions and licence terms change — confirm on the developer's site before deploying in a clinic.