Meshroom: Free photogrammetry

To the point:

Meshroom allows you to run the whole photogrammetric pipeline. Put your images and it will generate a textured mesh automatically.

It’s mostly GUI for AliceVision framework:

AliceVision is a Photogrammetric Computer Vision framework for 3D Reconstruction and Camera Tracking.

Great thing is that they are using Alembic, OpenEXR and such, so no problem with exporting meshes. Also the Meshroom GUI is made in Python so, maybe there is a chance to integrate it with Blender itself. They have plugin for modeling in Maya.

Here are some links to read/watch more about it:

Official website: https://alicevision.github.io/
Examples on Sketchfab: https://sketchfab.com/AliceVision
Overview with presentation in Blender: https://www.youtube.com/watch?v=NdpR6k-6SHs
Article from Scanbox: https://scanbox.xyz/blog/alicevision-opensource-photogrammetry/

12 Likes

Here’s a Blender integrated option. Some alternatives, free and not, are mentioned in the linked comparison:

Cheers for the heads up btw @cgslav.

2 Likes

@LoboTommy Thanks, I saw this topic before. As far as I remember nobody mentioned there Meshroom. For me it’s super new and even more interesting. Remarkable ease of use and node based system in GUI :wink:

I have a large collection of precision (48 views) macro-photography images of insects. I have not set up a photogrammetry pipeline, but have been wanting to try one out. Is anyone with a working pipeline interested in trying out an insect? I would be curious to see the results.

Agreed. That’s why the cheers / thanks.

Hi @wph4, I’d give it a shot if you’d like to upload the photos somewhere. I’m most familiar with VisualSFM. Here’s a tut if you’d like to try it for yourself:

I can process it with Meshroom if you want.

Do you have any windows binaries done that we can download rather than have to build from github? Cheers J

We’ve been postponing the release (with linux and windows binaries) for a few weeks, but it should be out this week.

5 Likes

Here’s a Dropbox link with 48 images of a Dragonfly. I’m curious to see how well it deals with semi-transparent wings.
https://www.dropbox.com/sh/bzznw8riavcbtb6/AACHSZvTu_IotU0uCzBmE4C0a?dl=0

Bill

Yes it would be great! I hope soon it will be possible to download the installation version under Windows. And the link will be on the github? or on the official site?

Yes, the binaries will be hosted on github, and we will add a link on the website.

5 Likes

Thank you!)

wow thanks for sharing. I’m looking forward to trying it out, not sure what i will use it for yet though

You’ll notice us when they will be available? I’d like to try it but I’m not able to compile it

thanks in advance

@fabiencastan, Thank you for this.
Here in Kubuntu 18.04. For AliceVision in addition to packages that are mentioned in documentation, I had to install “libboost-all-dev” package.
Now I can configure, but I get this error with “make”.
http://pasteall.org/1048811

Anyway, it would be great if you could share portable binaries for Linux.

Linux binaries would be super awesome. I’ve tried to build it on Arch but it was harder than I imagined. Building Blender is one thing, building this one is amazingly error prone.

You need OpenImageIO >= 1.8

Also a Windows build should be really appreciated :wink:

1 Like

Thank you!
Now this:
http://pasteall.org/1049506
I have installed official CUDA Toolkit 9.1.
Edit:
I had not seen the message about gcc version (have installed v7). Looking for it now.
Edit 2:
Mmm, I have installed gcc v4.8 and v6 too
Edit 3:
Ok, I used this to force configuring with gcc 6 in AliceVision:
cmake -DCMAKE_C_COMPILER=/usr/bin/gcc-6 -DCMAKE_BUILD_TYPE=Release . ../AliceVision

@cgslav. The problem with Linux builds is that it must be a portable build, including important libraries or Appimage package for example. Otherwise, this usually brings a lot of problems and does not work in many distros.

1 Like

“/usr/local/cuda-9.1/include/crt/host_config.h:121:2: error: #error – unsupported GNU version! gcc versions later than 6 are not supported!”

You need to choose compatible gcc and cuda versions. We cannot do much about that.