New toolkits released

February 22nd, 2011 by Henri Leave a reply »


ETH-V3D Structure-and-Motion software was created by Christopher Zach. The original source code with dataset is available at Christopher Zach Open-Source page (GPL license). I have created a windows port of V3DSfMToolkit with scripting wich is available as both binary ( and source (github).

I’ve tested this toolkit with the dataset given by Christopher Zach (see above screenshot) the reconstruction looks good but I only managed to get partial reconstruction from my own dataset.


Quasi-Euclidean Epipolar Rectification: MissStereo created by Pascal Monasse, Neus Sabater, Zhongwei Tang. The original source code is available at the IPOL related page under GPL license. You can download my windows port as both binary ( and source (github).

I’m interested by this method to estimate the fundamental matrix without prior focal length knowledge.

PhotoSynthToolkit with XSI support

With the help of Julien Carmagnac (3D Graphist and XSI advanced user), I’ve duplicated the 3DS Max texture projection rendering solution for Softimage XSI:

As usual, the new version of PhotoSynthToolkit including this new feature is available on his dedicated page.

SFMToolkit with sequence matching

SFMToolkit was packaged with BundlerMatcher, my own matching solution using SiftGPU: this is a trivial quadratic exhaustive matching implementation. This implementation is well-suited for unstructured (unordered) pictures but if you are using a sequence of images as input (movie) you can reduce the complexity of the matching to linear. You can do that by only comparing frame N with frame N+1, N+2 [...], N+p. I’ve also fixed a bug that may occur on some system (Windows 7 x64 with UAC activated): thanks to Josh Harle for the bug reporting!

The new version of SFMToolkit is available on his dedicated page: this new feature is hidden, you need to edit “1 – Bundler.wsf” and set SEQUENCE_MATCHING_ENABLED to true (replace false by true).


I’m going to the 3D-Arch’2011 Workshop at Trento: 3D Virtual Reconstruction and Visualization of Complex Architectures. I hope to see amazing things about 3D reconstruction ;-)

Jean-Philippe Pons (CSTB, Sophia-Antipolis, France): High-resolution large-scale multi-view stereo



  1. Do the images have to be a special png format for missStereo?

    I have tried various images from a 3D camera, no results for any.

  2. admin says:

    @Olafur: I don’t know, I only have tried with PNG-24.

  3. Pascal Monasse says:

    @Olafur: if you could post one of your problematic images, I would gladly take a look. Any 8-bits PNG image should be fine (RGB or greytone), but 16-bits are not.

  4. spamsac says:

    Interesting stuff! Thanks for sharing.


  5. I found out what the problem was… I always meant to send this to you Henri…
    So I work on MacOS then I take images onto a PC if I take the whole folder MacOs will leave hidden files in the folder… Therefore not only will MissStereo not work but none of the toolkits… So for the fellow MacOs users, make sure you only move images not the whole folder across, same with any hidden files make sure there aren’t any. It might be worth if possible that all the tool-kits will simply ignore hidden files.

  6. Henri, have you ever contacted Shubao Liu and David Cooper to see whether they would be willing to give you their code for their inverse ray tracing technique: ?

  7. Ivan RM says:

    A little OT, I have ported SiftGPU ( to OSX 10.6 64 bits. Only OpenGL is working, but it must be as fast as CUDA.

    So Bundler, CMVS, PMVS and BundlerMatcher are ported too. (360º reconstruction is working and tested)

  8. @Ivan RM Got any more information on this? Would love to try and run a toolkit on the Mac…

  9. Pierre says:

    @Ivan RM : If you want I could report your changes on my CMake version of Bundler, PMVS, CMVS source.

  10. Ivan RM says:

    @Olafur Try this (binary + libs)
    It’s bundler with SiftGPU + BundlerMatcher(without sequence matching), plus cmvs and pmvs2

    @Pierre Only problem with bundler is the binary dependency over David Lowe’s sift (win/linux). And some other dependencies are hard to get on mac (fortran).

  11. Cesar Lopez says:

    I was trying to use cloud computing that offers windows hpc with nvidia gpu access in beta.

    So i was trying to run bundlermatcher with that and got “Shader is not supported by your hardware”
    “Framebuffer object not supported”
    Error: can not initialize opengl context for SiftGPU”

    I tried other CUDA application and seemed to run fine.

    Am i missing something?

  12. @Ican RM
    I am getting this error for BundlerMatcer

    [Found 20 good images]
    [- Matching keypoints (this can take a while) -]
    ./ line 62: BundlerMatcher: command not found
    [- Running Bundler -]

  13. Ivan RM says:

    @Olafur I’m repackaging the binaries and dynlibs, because they depend on my custom macports setup.

    Your current problem is because I put bundler and bundler/bin in my $PATH env variable. But you will get problems with libs later, I will post a fix later.

  14. Ivan RM says:

    Tested with a virtualized clean install 10.6 64bits

  15. Great, I will have a look at this tomorrow. Thanks a lot for the port…

    I will report back on the forums, as I saw you posted there.

  16. Kjellis85 says:

    Tried the V3DSfMToolkit, but it keeps crashing after writing the feature files. Tried with different settings. Not a mem issue. Any ideas?

  17. admin says:

    @Kjellis85: What is your Windows separator , or . (in regional settings) ? Can you try to switch to the . separator ? (This is just my first guess).

  18. Kjellis85 says:

    I tried, but to no avail.

  19. Matt says:

    I tried installing the Bundler for Mac 10.5, but when I ran bundler it said it wouldn’t work because of ‘libGLEW.1.5.1.dylib’ couldn’t be found. I put the attached file, ‘libGLEW.1.5.8.dylib’ into the specified path, usr/bin, but it still didn’t work.

    I am just using the more updated version of libGREW that is included with the above install. Shouldn’t this be good enough?

    Here is the error

    [Found 11 good images]
    [- Matching keypoints (this can take a while) -]
    dyld: Library not loaded: /usr/lib/libGLEW.1.5.1.dylib

  20. Basilio says:

    Thanks a lot for sharing this, both to Henri, and to Ivan for porting it to osx!
    Dumb question: only 3 images are selected from the ET set, are there options that could be changed or is it only able to match those? (I ran the osx version provided by Ivan without any option changes). The software seems to work pretty fine with higher resolution images.

    @Matt, you can rename the libGLEW.1.5.8.dylib to *.5.1.dylib and put it in your usr/bin path, or you can use
    “install_name_tool -change /usr/lib/libGLEW.1.5.1.dylib @executable_path/libGLEW1.5.8.dylib speed” (from the /bin directory)

  21. Tom says:

    Hi, I´m trying to run bundler from Mac as well, but get the same error as Matt.

    The solution from Basilio did not work for me.

    Tried copying the file, but got the same error.
    Tried executing the install_name_tool but get the following error
    “install_name_tool: can’t open file: speed (No such file or directory)”

    Am I missing something here?
    Would appreciate any suggestions.

  22. Simone82 says:

    Hi guys, I tried to test this toolkit with my set, but I can’t continue after verify_ED_edges.exe, that crash constantly on Windows 7 64bit. Can you help me to resolve the problem? :(

  23. Henri says:

    @Simone82: is the sample provided working? If not, this could be a problem with your setup, otherwise you should contact the author of the source directly ;-)

  24. Simone82 says:

    Right Henri, I not remember to write that the toolkit not working with examples too. Probably, if not resolve, I contact the author. Thank you.

Leave a Reply