Update: new 0.4.0 binaries for Windows

Some folks encountered a DLL-loading related issue, which curiously isn’t reproducible on my Windows 7 test machine. The most likely cause of the problem has been identified and fixed.

If you download Mitsuba again, it should hopefully work.

Update 2: turns out that the problem was a missing Intel runtime library. The current binary distribution now includes this.

24 comments

  1. Unfortunately, it doesn’t.

    • what does the error message look like now? I should have a backslash in it, no?

      • The error message still has a forward slash after plugins.

        In your change, it says, too:
        + fs::path shortName = fs::path(“plugins/”) / name;

        However since the path resolve seems to work in general, the problem might be even more obscure. Are you sure you are handling the error from the DLL loading routine correctly? Maybe the “DLL not found” message can be caused some other way than just failed path resolves.

  2. Unfortunately, the issue persists when trying to load e.g. bitmap.dll or envmap.dll.
    However, the Cornell Box scene (that doesn’t use textures) does work and it seems to be loading the DLLs required for that from the plugins path.

  3. It looks exactly the same.
    ERROR load [plugin.cpp:72] Error while loading plugin C:\…\Desktop\Mitsuba 0.4.0\plugins/envmap.dll

  4. By using Dependency Walker I found out that bitmap.dll depends on svml_dispmd.dll that couldn’t be found.
    I copied it over from Maya, it works! 😀
    The fun thing about the “DLL not found” error is that it has no way to tell you *which* DLL in the loading chain actually wasn’t found.

  5. How true! Excellent job, mars. :)

  6. Yeah, dll problem sorted out :)

    Unfortunately, I’m still having crashes when combining the bidirectional integrator, a dielectric bsdf, and geometry with per-vertex normals (or that’s what it looks like, I haven’t been able to reproduce it with per-face normals). I’ll try to submit a report to the tracker.

    Also, there’s a little problem not exactly related to the renderer itself: it looks like the crash message window is in the same process as the renderer, so when Windows kills the process because it’s not responding, the error window is also closed, and then you can’t click “OK” to submit the crash info. It would be better to create an independent process for it.

    • Hi DED,

      could you upload your scene to the bugtracker? That would be super-helpful!

      The crash reporter on Windows is not ideal, though it seems to be doing its job most of the time (I’ve gotten a whole bunch automated emails and already fixed some of the issues).

      Wenzel

      • Yeah, the crash reporter works fine for me as well, most of the time you have enough time to send the report. I’m a nitpicker, you know 😉

    • just saw that you already did — thanks :)

  7. Great find, that dll

  8. Thank you for all of your work Jakob, I’m just wondering, is everything within mitsuba material base, only manually inputted. Will there be a GUI for the material editing, kernel, camera, film?

  9. Great work!
    It works fine.

    I like SSS boxes.
    :)

  10. yee
    seldom i see such a great piece of code, not only well typed but well thought and clean – excellent job on that baby!

    one thing – studio – ability to load mesh with tex and play within Mitsuba with matts & lights… this is im longing for…

  11. and the second – there was a blender mitsuba – any chances for update on this?

  12. I just wanted to let you know, that I updated Blender exporter to new version.
    http://bartoszstyperek.wordpress.com/2012/10/07/mitsuba-exporter-1-0/
    When sky and sss fix will be out most parameters should work ok.

Leave your reply to mars