niessner / BundleFusion

[Siggraph 2017] BundleFusion: Real-time Globally Consistent 3D Reconstruction using Online Surface Re-integration
http://graphics.stanford.edu/projects/bundlefusion/
Other
1.5k stars 355 forks source link

FriedLiver stuck #20

Open liangxu007 opened 6 years ago

liangxu007 commented 6 years ago

Hi, everyone. I have some problems when using Kinect 2.0 for online Reconstruction. In zParametersDefault.txt I changed s_sensorIdx to 2, and it compiled successfully. My platform is Windows 10、Cuda 9.1、VS2013、Kinect SDK 2.0、Microsoft DirectX SDK (June 2010). Everytime I set the program running, it stuck, without any response. Thanks very much for any suggestions!

liangxu007 commented 6 years ago

error

xiaomaoo commented 6 years ago

Try to change the parameter "s_maxNumImages " from 1200 to 400, in the zParametersBundlingDefault.txt

liangxu007 commented 6 years ago

@xiaomaoo I tried but it stuck again. Even change to 100, it stuck again.

Keudeugi commented 6 years ago

Hope this help for you (or the guys who are struggling with build stage using Kinect One), but I'm trying to adapt OpenNI2 patched with libfreenect2 to make Kinect One work in this project. Currently, I can read only the first frame of stream, but I think It can be solved after I figure out some data processing methods...

I'm currently working on the platform with Windows 10, CUDA 9.0, VS 2015.

bundlefusion

liangxu007 commented 6 years ago

@Keudeugi Hi, have you figured out the data processing methods now?

quanqhow commented 6 years ago

I also tested my Kinect V2 (for XBox One) under the FriedLiver.exe, but it still stuck at the moment. So I have tested both kinds of Kinect and the FriedLiver stuck at the two conditions.

psyduck04 commented 6 years ago

@Keudeugi Hi, what's your directX version ?

Keudeugi commented 6 years ago

@94zhaofang Since my O/S is Windows 10, as well as I didn't installed any additional drivers for DirectX, I think it doesn't matter for Direct X version. One thing I remember is this project is using somehow lower version than default Windows 10's version, which is 12. (In short, a version around ~ X11 is used for this project)

Keudeugi commented 6 years ago

@liangxu007 Sorry for late reply, but I gave modifying up since it spent too much time (Because the reason that I don't have enough knowledge in this field still). Please mention me if there's a people who make it work with Kinect One.

hejinying commented 5 years ago

@Keudeugi hi,when i use this project with vs2015, i met the problem here: error C2280: 'ml::D3D11Canvas2D::ElementBillboard::ElementBillboard(const ml::D3D11Canvas2D::ElementBillboard &)': attempting to reference a deleted function do you know how to solve this problem ??

tkbala commented 5 years ago

I have the same issue as @hejinying . Any ideas?

FishHe commented 4 years ago

Same question. The program is stuck at the very beginning.

kuangqi93 commented 4 years ago

Same question. The program is stuck at the very beginning.

Have you solved that?

FishHe commented 4 years ago

I have tested on Win10 +CUDA 10.2 + VS2013 + GTX 2080 Super. The most important fix is: #56

chethanab16 commented 4 years ago

@liangxu007 Did you solve the problem of the window stuck when the program starts.

chethanab16 commented 4 years ago

I have tested on Win10 +CUDA 10.2 + VS2013 + GTX 2080 Super. The most important fix is: #56

@FishHe I have tried this! but it did not fix the problem.

BruceWANGDi commented 3 years ago

Hi, guys. There is some problem with my project when I using KinectOneV2(My environment is Win10, VS2013, CUDA8.0, and RTX2080Ti), I don't know whether my GPU can't support CUDA8.0. When I running my .exe, it stuck, then crash without any response. The FriedLiver window just gets dark. Does anyone have some solutions to help me to solve this problem? Thank you so much.

mshmoon commented 3 years ago

I think the reason for this problem is that the thread is locked while waiting for SIFT feature extraction, which makes it unable to continue execution. However, I don't know what causes the feature extraction failure.

Please see the DepthSensing.cpp line 1016 and line 1017:

1016: //wait until the bundling thread is done with: sift extraction, sift matching, and key point filtering 1017: ConditionManager::lockBundlerProcessedInput(ConditionManager::Recon);

mshmoon commented 3 years ago

After further debugging, I found that the program can execute the code to extract SIFT features, but when the program runs to ProgramCu.cu line 1424,it can be stuck! `int ProgramCU::CheckErrorCUDA(const char* location) {

if (defined(_DEBUG) || defined(DEBUG))

    cudaDeviceSynchronize();
cudaError_t e = cudaGetLastError();
if (e)
{
    if (location) fprintf(stderr, "%s:\t", location);
    fprintf(stderr, "%s\n", cudaGetErrorString(e));
    assert(0);
    return 1;
}
else
{
    return 0;
}

else`

Jason-YJ commented 3 years ago

@mshmoon Hi, do you have further progress on solving the problem? Thanks