Stopped with error: Check failed: !elems.empty()

[markmgit]

Using version 0.9.0 on windows 10 for the first time. It sopped on the MultieviewStereoCL node, so I checked the log and near the end I see the section below, with Check Failed message.

Any suggestions ?

StereoFusion::Options
---------------------
max_image_size: 3200
min_num_pixels: 5
max_num_pixels: 10000
max_traversal_depth: 100
max_reproj_error: 2
max_depth_error: 0.01
max_normal_error: 10
check_num_images: 50
cache_size: 32

Reading workspace...
Reading configuration...
Fusing image [1/2] in 5.970s (161 points)
Fusing image [2/2] in 0.979s (161 points)
Number of fused points: 161
Elapsed time: 0.136 [minutes]
Writing output: C:/Users/x/MeshroomCache/MultiviewStereoCL/a519ee3e9e0925b6f4438ceaf257a1f0b6f8ec07\fused.ply
Triangulating points...
Triangulation has 160 using 161 points.
Initializing ray tracer...
Initializing graph optimization...
Integrating image [1/2] in 0.009s
Integrating image [2/2] in 0.000s
Setting up optimization...
Running graph-cut optimization...
Extracting surface as min-cut...
Creating surface mesh model...
F0925 14:50:36.071591 13976 math.h:233] Check failed: !elems.empty() 
*** Check failure stack trace: ***
    @   00007FFDBC785FCB  public: void __cdecl google::LogMessage::Flush(void) __ptr64
    @   00007FFDBC784B29  public: __cdecl google::LogMessageFatal::~LogMessageFatal(void) __ptr64
    @   00007FF63F9AD401  (unknown)
    @   00007FF63F9CCB2A  (unknown)
    @   00007FF63F9CCFBC  (unknown)
    @   00007FF63F88F849  (unknown)
    @   00007FF63F8817A6  (unknown)
    @   00007FF63F8C41BF  (unknown)
    @   00007FF63FDDEAD4  (unknown)
    @   00007FFE07D17034  BaseThreadInitThunk
    @   00007FFE088826A1  RtlUserThreadStart

[markmgit]

Re-ran it, and it completed without error this time. Strange. But the output was not good so I'm starting again from scratch.

[markmgit]

MultiViewStereoCL Stopped with errors again:

StereoFusion::Options
---------------------
max_image_size: 3200
min_num_pixels: 5
max_num_pixels: 10000
max_traversal_depth: 100
max_reproj_error: 2
max_depth_error: 0.01
max_normal_error: 10
check_num_images: 50
cache_size: 32

Reading workspace...
Reading configuration...
Fusing image [1/2] in 0.395s (0 points)
Fusing image [2/2] in 0.083s (0 points)
WARNING: Could not fuse any points. This is likely caused by incorrect settings - filtering must be enabled for the last call to patch match stereo.
Number of fused points: 0
Elapsed time: 0.010 [minutes]
Writing output: C:/Users/jacob/Pictures/!3d/MeshroomCache/MultiviewStereoCL/19f2772afb98b55710a65e0e166590c1198de9f4\fused.ply
Triangulating points...
Triangulation has 0 using 0 points.
Initializing ray tracer...
F0925 20:45:49.941437 10648 math.h:233] Check failed: !elems.empty() 
*** Check failure stack trace: ***
    @   00007FFDBC785FCB  public: void __cdecl google::LogMessage::Flush(void) __ptr64
    @   00007FFDBC784B29  public: __cdecl google::LogMessageFatal::~LogMessageFatal(void) __ptr64

Could it be my images? I took 70+ images, I know they are not great, but for me it was a first experiments to see if it's worth continuing down this path. The first 2 nodes seem to work OK. But MultiViwStereoCL is crashing. I'd love to solve this if there are any suggestions based on this log?

Is there a Testing Set of images somewhere that I can try to make sure this actually works?

[Oop-o]

Same issue. My images appear to be good quality. This is my second model to be generated with meshroom. The first model completed successfully with 170 images. This time I only used 19 images. I wonder if the failure has something to do with image matching as less images leave less feature overlap for the program to reference...

[ledbrainlet]

Same issue, tried multiple testing datasets (even the one recommended for meshroom, ranging in sizes 3 to 50 pics) but it still fails. Does anyone have any idea how this can be solved. Seeems to be a common issue, ive seen multiple issues opened on this matter.

I have 32 gb of ram, a ryzen 5 5700x and a rx6800 with 16gb vram so i definitly dont think this is a perfomance issue