-
Notifications
You must be signed in to change notification settings - Fork 19
License
zewt/maya-implicit-skinning
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
Note ---- This code is unmaintained, but released in case it's useful to anyone. Building -------- To build in Windows, just open msvc/implicit_framework.vcxproj. If you're not in Maya 2015 or CUDA 6.5, you may need to update the include and library paths. Building in Linux with CMake hasn't been tested in a while and will probably need to be updated. This requires CUDA with an SM 3 device or better. Maya quick start: - Open the script console. - Load the plugin in the MEL tab: loadPlugin "/path/to/implicit_framework.mll" - In the Python tab, add the top directory containing "implicit" to the Python path: import sys sys.path.append('C:/Users/73000150/Desktop/maya_is') - Open a scene and select a mesh. The mesh must already be skinned. - In the Python script tab, import and run the main script: import implicit implicit.main() This will create a number of nodes, which will all be placed in a group named after your skinCluster: - One ImplicitShape will be created for each skin joint that has painted weights on the shape. These are parent constrained to the joints, but are placed in a separate group to avoid cluttering your bone hierarchy. These nodes can be seen and viewed in the viewport. Normally you'll want to hide them. - One ImplicitBlend shape will be created, connecting all of the ImplicitShape nodes together. This is also shown in the viewport as a merging of all of the individual ImplicitShapes. This node isn't parented to anything and never moves, since it lives in world space. You'll usually want to hide this, too. - An ImplicitDeformer is added to the deformer chain of the original object, with the ImplicitBlend node as its input. This performs the actual deformation. WARNING 1: this is a CUDA project only compatible for Nvidia graphic cards. The project only guarantee support of GPU card down to compute capability sm_20 (FERMI architecture). But it might work for architecture below sm_20. WARNING 2: This is a research application. It means worse than beta version... It means clumsy GUI with no undo feature and lots of buttons. It means it can crash anytime. You have been warned. Have fun warrior :) ===================== COMPILING UNDER LINUX ===================== Be sure to have gcc installed (package build-essential has most dependencies you need). Be aware that cuda only works with old versions of gcc. At this time only gcc-4.6 is supported with CUDA 5.0. so install gcc-4.6 and g++-4.6. To use other versions you'll need to modify the CMakeList.txt and replace occurences of gcc-4.xx with the correct compiler version You also need cmake and make. Install CudaToolKit from : https://developer.nvidia.com/cuda-downloads Don't forget to setup environnement variables in your ~/.bashrc or equivalent: export LIBRARY_PATH=/usr/lib/nvidia-current:$LIBRARY_PATH export LD_LIBRARY_PATH=/usr/local/cuda/lib export PATH=$PATH:/usr/local/cuda/bin export CUDA_LIBRARY_PATH=/usr/local/cuda/lib It is always a good thing to ensure everything is working properly by installing and compiling the Cuda SDK (same link above). You just need to run make at /CUDA_SDK_PATH/C/common and then launch a random app in /CUDA_SDK_PATH/C/common/bin Now that you have checked everything is working install these dependencies : Library dependencies : - glew - qt framework 4.7.xx Building the project is pretty straightforward, just run : $ cd build $ cmake -DCUDA_COMPUTE_CAPABILITY=sm_xx ./.. $ make replace sm_xx by the compute capability of your graphic card (i.e sm_10 sm_11 sm_(..) sm_21 etc.) you can check the compute capability against your GPU name with wikipedia https://en.wikipedia.org/wiki/CUDA#Supported_GPUs You can also build the project in debugging mode by doing : $ cd build $ cmake -DCUDA_COMPUTE_CAPABILITY=sm_xx -DCMAKE_BUILD_TYPE=Debug ./.. $ make ****************** *Linking problems: ****************** Sometimes your system won't find the libraries /usr/lib/nvidia-currrent/libcuda.so.xxx which will result in errors at linking for the functions of family cuXXXXX() The only fix I have found so far is to make symbolic links of these directly within /usr/lib : ln -s /usr/lib/nvidia-current/libcuda.so /usr/lib/libcuda.so ln -s /usr/lib/nvidia-current/libcuda.so.xx libcuda.so.xx ... ======================= COMPILING UNDER WINDOWS ======================= Setup under windows is much more complicated, you'd better be ready ! Thirst you need to install visual studio 2010. Add it to your PATH (e.g "C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin"). Prefer the 32 bit version to the 64 bit because the later is really more of a hassle to get working. Install the Cuda toolkit from https://developer.nvidia.com/object/cuda_3_2_downloads.html choose the 32 bit version also you need to have your video card drivers up to date. Install the Cuda SDK and check if some of the apps are working properly. Be sure CUDA_SDK_ROOT_DIR environnement variable is set up correctly (e.g CUDA_SDK_ROOT_DIR = C:\ProgramData\NVIDIA Corporation\NVIDIA GPU Computing SDK 3.2\) Copy the glew32.lib and GL headers from the SDK into the correct directory of visual Install QT binnaries for visual studio 2010 (and not mingw nor visual 2008) don't forget to run the configure.exe located in the root folder of the Qt binnaries. Also add Qt's /bin directory to your PATH You also need cmake for windows (https://www.cmake.org/cmake/resources/software.html) Don't forget to add it to your PATH ----------------------------- Build under Visual Studio IDE ----------------------------- Simply run : $ cd build $ call "C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\vcvars32.bat" $ cmake -DCUDA_COMPUTE_CAPABILITY=sm_x ./.. - Replace the path to your visual studio path. vcvars32.bat is needed to setup environment variables before calling visual studio executables through cmake. - replace sm_x by the compute capability of your graphic card (i.e sm_10 sm_11 sm_(..) sm_21 etc.) - Open ALL_BUILD.vcproj and stroke F7 ---------------------- Build under Qt Creator ---------------------- Add QtCreator to your path in order to jom.exe to be reachable $ cd build $ call "C:\Program Files (x86)\Microsoft Visual Studio 10.0\VC\bin\vcvars32.bat" cmake -DCMAKE_BUILD_TYPE=Debug -DCMAKE_RC_COMPILER="F:/Program Files (x86)/Microsoft_Visual_Studio_10/VC/bin" -DCUDA_COMPUTE_CAPABILITY=sm_x -G"NMake Makefiles JOM" ./.. - Replace the path to your visual studio path. vcvars32.bat is needed to setup environment variables before calling visual studio executables through cmake. - replace sm_x by the compute capability of your graphic card (i.e sm_10 sm_11 sm_(..) sm_21 etc.) - when specifying path to visual with '-DCMAKE_RC_COMPILER' don't use backslashes but slashes. - Open Qtcreator
About
No description, website, or topics provided.
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published