search

cdcseacave / openMVS

open Multi-View Stereo reconstruction library
http://cdcseacave.github.io
GNU Affero General Public License v3.0
3.35k stars 909 forks source link

Scalable pipeline Segmentation Fault #1100

Open xiaoxiae opened 9 months ago

xiaoxiae commented 9 months ago

Describe the bug Running the steps outlined in the MvsScalablePipeline.py yields a segmentation fault.

To Reproduce Running the following commands (after InterfaceCOLMAP):

  1. DensifyPointCloud scene.mvs --resolution-level 1 --iters 4 --cuda-device -1 --fusion-mode 1
  2. DensifyPointCloud scene.mvs --resolution-level 1 --iters 4 --cuda-device -1 --sub-scene-area 20000000
  3. DensifyPointCloud scene_0000.mvs --resolution-level 1 --iters 4 --cuda-device -1 --dense-config-file Densify.ini -v 3
    • the Densify.ini contents are Optimize = 0

Step (3) crashes with the following output:

20:48:49 [App     ] OpenMVS x64 v2.2.0
20:48:49 [App     ] Build date: Jan 25 2024, 10:11:25
20:48:49 [App     ] CPU: AMD Ryzen 7 1700 Eight-Core Processor           (16 cores)
20:48:49 [App     ] RAM: 62.74GB Physical Memory 32.00GB Virtual Memory
20:48:49 [App     ] OS: Linux 5.15.146-1-MANJARO (x86_64)
20:48:49 [App     ] Disk: 215.30GB (960.17GB) space
20:48:49 [App     ] SSE & AVX compatible CPU & OS detected
20:48:49 [App     ] Command line: DensifyPointCloud scene_0000.mvs --resolution-level 0 --iters 4 --cuda-device -1 --dense-config-file Densify.ini -v 3
20:48:49 [App     ] MapSMtoCores for SM 8.6 is undefined; default to use 64 cores/SM
20:48:49 [App     ] CUDA device 0 initialized: NVIDIA GeForce RTX 3060 Ti (compute capability 8.6; memory 7.78GB)
20:48:49 [App     ] Scene loaded (5ms):
    19 images (19 calibrated) with a total of 416.89 MPixels (21.94 MPixels/image)
    6482 points, 0 vertices, 0 faces
20:48:49 [App     ] error: too few images to be a tower: '19'
20:48:49 [App     ] Point-cloud composed of 6482 points with:
 - points info:
    1887 points inside ROI (29.11%)
     inside ROI track length: 2 min / 3.28087 mean (1.39024 std) / 9 max
    outside ROI track length: 2 min / 2.48335 mean (0.734873 std) / 7 max
 - visibility info (17602 views - 2.72 views/point):
            0 points with 1- views (0.00%)
         3567 points with 2  views (55.03%)
         1888 points with 3  views (29.13%)
         1027 points with 4+ views (15.84%)
    2 min / 2.71552 mean (1.03766 std) / 9 max
20:48:50 [App     ] K18 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K10 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K9 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K15 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K8 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K6 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K13 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K12 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K11 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K0 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K14 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K4 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K16 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K2 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K17 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:50 [App     ] K7 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:51 [App     ] K1 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:51 [App     ] K3 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:51 [App     ] K5 = 
 1655.7214     0.0000  1279.7132 
    0.0000  1655.7214   816.6619 
    0.0000     0.0000     1.0000 

20:48:51 [App     ] Preparing images for dense reconstruction completed: 19 images (1s408ms)
20:48:51 [App     ] Reference image  14 sees 11 views:  12(219pts,2.05scl)  11(172pts,2.09scl)   9(167pts,2.07scl)  15(157pts,1.05scl)  10(61pts,2.08scl)  17(42pts,1.94scl)   2(56pts,2.47scl)  16(40pts,1.01scl)   7(59pts,2.79scl)   0(24pts,3.59scl)   1(11pts,3.38scl) (383 shared points)
20:48:51 [App     ] Reference image   6 sees 5 views:   5(419pts,0.98scl)   7(240pts,0.97scl)   3(136pts,1.28scl)   4(115pts,1.25scl)  13(6pts,0.79scl) (571 shared points)
20:48:51 [App     ] Reference image   6 sees 5 views:   5(419pts,0.98scl)   7(240pts,0.97scl)   3(136pts,1.28scl)   4(115pts,1.25scl)  13(6pts,0.79scl) (571 shared points)
20:48:51 [App     ] Reference image   6 sees 5 views:   5(419pts,0.98scl)   7(240pts,0.97scl)   3(136pts,1.28scl)   4(115pts,1.25scl)  13(6pts,0.79scl) (571 shared points)
20:48:51 [App     ] Reference image   7 sees 16 views:   6(240pts,1.04scl)   5(132pts,0.94scl)   3(88pts,1.11scl)   4(84pts,1.19scl)  14(59pts,0.36scl)  13(20pts,0.83scl)  12(24pts,0.78scl)  11(16pts,0.94scl)   2(11pts,2.43scl)  10(6pts,1.33scl)  17(6pts,1.10scl)   9(7pts,0.83scl)  15(26pts,0.39scl)   0(5pts,2.10scl)   1(3pts,2.31scl)  16(13pts,0.39scl) (467 shared points)
20:48:51 [App     ] Reference image  18 sees 10 views:  17(540pts,0.98scl)   8(149pts,1.21scl)  11(460pts,1.09scl)   1(49pts,1.71scl)  12(59pts,1.10scl)   2(30pts,1.23scl)   0(21pts,1.98scl)  10(30pts,1.14scl)  13(22pts,0.87scl)   9(5pts,1.14scl) (697 shared points)
20:48:51 [App     ] Reference image  10 sees 16 views:   2(594pts,1.26scl)   9(491pts,0.99scl)   0(177pts,1.67scl)  11(393pts,0.97scl)   1(135pts,1.71scl)  17(196pts,0.87scl)  15(153pts,0.49scl)  16(132pts,0.53scl)  12(136pts,0.94scl)  13(82pts,0.60scl)  14(61pts,0.48scl)  18(30pts,0.89scl)   3(23pts,1.35scl)   4(20pts,1.34scl)   8(4pts,1.08scl)   7(6pts,0.87scl) (1000 shared points)
20:48:51 [App     ] Reference image  13 sees 13 views:  11(158pts,1.60scl)   9(98pts,1.63scl)  17(132pts,1.34scl)  10(82pts,1.68scl)   0(51pts,2.76scl)   2(64pts,2.13scl)   1(40pts,2.85scl)  12(22pts,1.38scl)  18(22pts,1.15scl)   7(20pts,1.22scl)   8(10pts,1.32scl)   6(6pts,1.29scl)   5(4pts,1.24scl) (265 shared points)
20:48:51 [App     ] Reference image   4 sees 8 views:   3(1017pts,0.97scl)   6(115pts,0.86scl)   5(104pts,0.89scl)   7(84pts,0.98scl)   2(31pts,1.41scl)  10(19pts,0.84scl)   9(13pts,0.63scl)  16(10pts,0.39scl) (1061 shared points)
20:48:51 [App     ] Reference image   3 sees 8 views:   4(1016pts,1.04scl)   6(136pts,0.85scl)   5(119pts,0.86scl)   7(88pts,1.08scl)   2(35pts,1.45scl)  10(23pts,0.77scl)   9(19pts,0.67scl)  16(14pts,0.39scl) (1090 shared points)
20:48:51 [App     ] Reference image  15 sees 11 views:   9(288pts,2.07scl)  10(153pts,2.04scl)  16(431pts,1.03scl)  11(139pts,2.03scl)  12(126pts,1.95scl)  14(157pts,0.96scl)   2(111pts,2.50scl)   0(28pts,3.43scl)   1(15pts,3.54scl)  17(11pts,1.91scl)   7(26pts,2.59scl) (686 shared points)
20:48:51 [App     ] Reference image   1 sees 14 views:   8(478pts,0.62scl)   0(529pts,0.95scl)   2(123pts,0.68scl)  11(187pts,0.55scl)  10(135pts,0.59scl)  12(94pts,0.54scl)  17(115pts,0.48scl)  18(50pts,0.60scl)   9(77pts,0.57scl)  13(40pts,0.35scl)  15(14pts,0.28scl)  14(11pts,0.30scl)  16(3pts,0.30scl)   7(3pts,0.48scl) (902 shared points)
20:48:51 [App     ] Reference image   8 sees 10 views:   1(477pts,1.62scl)   0(342pts,1.56scl)  11(253pts,0.86scl)   2(183pts,0.80scl)  17(219pts,0.77scl)  18(149pts,0.84scl)  12(135pts,0.90scl)   9(5pts,0.87scl)  10(4pts,0.93scl)  13(10pts,0.80scl) (1040 shared points)
20:48:51 [App     ] Reference image   0 sees 14 views:   1(528pts,1.06scl)   8(342pts,0.65scl)   2(175pts,0.68scl)  11(226pts,0.56scl)  12(138pts,0.55scl)  10(177pts,0.60scl)  17(128pts,0.50scl)   9(119pts,0.58scl)  13(50pts,0.37scl)  18(21pts,0.52scl)  15(27pts,0.29scl)  16(14pts,0.29scl)  14(24pts,0.28scl)   7(5pts,0.53scl) (852 shared points)
20:48:51 [App     ] Reference image   2 sees 16 views:  10(594pts,0.80scl)  11(363pts,0.85scl)   9(364pts,0.80scl)   8(180pts,1.27scl)   0(179pts,1.57scl)   1(124pts,1.54scl)  17(183pts,0.77scl)  12(151pts,0.90scl)  15(110pts,0.40scl)  16(86pts,0.42scl)  18(30pts,0.83scl)  14(56pts,0.41scl)   3(36pts,0.74scl)  13(64pts,0.48scl)   4(31pts,0.75scl)   7(11pts,0.47scl) (1017 shared points)
20:48:51 [App     ] Reference image  11 sees 14 views:  17(1345pts,0.88scl)   2(365pts,1.20scl)  12(982pts,0.99scl)   8(253pts,1.17scl)   0(228pts,1.81scl)   1(185pts,1.84scl)   9(481pts,1.00scl)  10(393pts,1.04scl)  18(459pts,0.92scl)  14(172pts,0.48scl)  13(155pts,0.63scl)  15(138pts,0.50scl)  16(39pts,0.52scl)   7(16pts,1.12scl) (2253 shared points)
20:48:51 [App     ] Reference image  12 sees 14 views:  11(977pts,1.02scl)  17(663pts,0.88scl)   0(138pts,1.83scl)   2(150pts,1.15scl)   8(135pts,1.12scl)   9(328pts,1.03scl)   1(95pts,1.85scl)  14(219pts,0.49scl)  15(124pts,0.52scl)  10(136pts,1.07scl)  18(57pts,0.91scl)  16(23pts,0.52scl)  13(22pts,0.72scl)   7(24pts,1.30scl) (1325 shared points)
20:48:51 [App     ] Reference image  17 sees 13 views:  11(1351pts,1.13scl)  12(668pts,1.14scl)   2(181pts,1.33scl)   8(219pts,1.30scl)  18(541pts,1.02scl)   9(216pts,1.10scl)  10(195pts,1.16scl)   0(130pts,2.03scl)   1(114pts,2.08scl)  13(132pts,0.75scl)  14(42pts,0.52scl)  15(11pts,0.52scl)   7(6pts,0.92scl) (1740 shared points)
20:48:51 [App     ] Reference image   9 sees 16 views:   2(362pts,1.26scl)  10(488pts,1.01scl)  11(480pts,1.01scl)  15(287pts,0.48scl)  12(326pts,0.98scl)   0(119pts,1.73scl)  17(213pts,0.91scl)  16(236pts,0.51scl)  14(165pts,0.49scl)   1(77pts,1.75scl)  13(96pts,0.62scl)   8(5pts,1.16scl)   3(19pts,1.52scl)   4(14pts,1.62scl)   7(7pts,1.24scl)  18(5pts,0.88scl) (1122 shared points)
20:48:51 [App     ] Selecting images for dense reconstruction completed: 19 images (328ms)
Geometric-consistent estimated depth-maps 15 (78.95%, 32s, ETA 8s)... Segmentation fault (core dumped)

Trying to run the command for other scenes:

Looking at what command (2) produces:

20:24:04 [App     ] OpenMVS x64 v2.2.0
20:24:04 [App     ] Build date: Jan 25 2024, 10:11:25
20:24:04 [App     ] CPU: AMD Ryzen 7 1700 Eight-Core Processor           (16 cores)
20:24:04 [App     ] RAM: 62.74GB Physical Memory 32.00GB Virtual Memory
20:24:04 [App     ] OS: Linux 5.15.146-1-MANJARO (x86_64)
20:24:05 [App     ] Disk: 216.71GB (960.17GB) space
20:24:05 [App     ] SSE & AVX compatible CPU & OS detected
20:24:05 [App     ] Command line: DensifyPointCloud scene.mvs --resolution-level 1 --iters 4 --cuda-device -1 --sub-scene-area 20000000
20:24:05 [App     ] MapSMtoCores for SM 8.6 is undefined; default to use 64 cores/SM
20:24:05 [App     ] CUDA device 0 initialized: NVIDIA GeForce RTX 3060 Ti (compute capability 8.6; memory 7.78GB)
20:24:05 [App     ] Camera model loaded: platform 0; camera  0; f 1.201x1.201; poses 2
20:24:05 [App     ] Camera model loaded: platform 1; camera  0; f 1.144x1.144; poses 1
20:24:05 [App     ] Camera model loaded: platform 2; camera  0; f 1.402x1.402; poses 1
20:24:05 [App     ] Camera model loaded: platform 3; camera  0; f 0.680x0.680; poses 1
20:24:05 [App     ] Camera model loaded: platform 4; camera  0; f 1.305x1.305; poses 2
20:24:05 [App     ] Camera model loaded: platform 5; camera  0; f 1.407x1.407; poses 3
20:24:05 [App     ] Camera model loaded: platform 6; camera  0; f 1.014x1.014; poses 2
20:24:05 [App     ] Camera model loaded: platform 7; camera  0; f 1.976x1.976; poses 1
20:24:05 [App     ] Camera model loaded: platform 8; camera  0; f 0.873x0.873; poses 1
20:24:05 [App     ] Camera model loaded: platform 9; camera  0; f 1.949x1.949; poses 7
20:24:05 [App     ] Camera model loaded: platform 10; camera  0; f 0.647x0.647; poses 837
20:24:05 [App     ] Camera model loaded: platform 11; camera  0; f 1.486x1.486; poses 18
20:24:05 [App     ] Camera model loaded: platform 12; camera  0; f 0.728x0.728; poses 1
20:24:05 [App     ] Camera model loaded: platform 13; camera  0; f 2.121x2.121; poses 39
20:24:05 [App     ] Camera model loaded: platform 14; camera  0; f 1.703x1.703; poses 2
20:24:05 [App     ] Camera model loaded: platform 15; camera  0; f 1.071x1.071; poses 2
20:24:05 [App     ] Camera model loaded: platform 16; camera  0; f 1.642x1.642; poses 4
20:24:05 [App     ] Camera model loaded: platform 17; camera  0; f 0.821x0.821; poses 6
20:24:05 [App     ] Camera model loaded: platform 18; camera  0; f 0.990x0.990; poses 1
20:24:05 [App     ] Camera model loaded: platform 19; camera  0; f 0.896x0.896; poses 4
20:24:05 [App     ] Camera model loaded: platform 20; camera  0; f 1.769x1.769; poses 2
20:24:05 [App     ] Camera model loaded: platform 21; camera  0; f 0.938x0.938; poses 1
20:24:06 [App     ] Camera model loaded: platform 22; camera  0; f 0.780x0.780; poses 8
20:24:06 [App     ] Scene loaded from interface format (849ms):
    946 images (946 calibrated) with a total of 20426.95 MPixels (21.59 MPixels/image)
    311337 points, 0 vertices, 0 faces
20:24:06 [App     ] Found a camera not pointing towards the scene center; the scene will be considered unbounded (no ROI)
20:24:06 [App     ] error: does not seem to be a tower: X(44.74), Y(2183.39), Z(10851.75)
20:24:07 [App     ] error: opening file 'depth0966.dmap' for reading depth-data
20:24:16 [App     ] error: opening file 'depth0896.dmap' for reading depth-data
20:24:20 [App     ] error: opening file 'depth0875.dmap' for reading depth-data
20:24:20 [App     ] error: opening file 'depth0873.dmap' for reading depth-data
20:24:22 [App     ] error: opening file 'depth0843.dmap' for reading depth-data
20:24:22 [App     ] error: opening file 'depth0842.dmap' for reading depth-data
20:24:24 [App     ] error: opening file 'depth0834.dmap' for reading depth-data
20:24:24 [App     ] error: opening file 'depth0833.dmap' for reading depth-data
20:24:40 [App     ] error: opening file 'depth0712.dmap' for reading depth-data
20:24:40 [App     ] error: opening file 'depth0711.dmap' for reading depth-data
20:24:40 [App     ] error: opening file 'depth0706.dmap' for reading depth-data
20:25:03 [App     ] error: opening file 'depth0221.dmap' for reading depth-data
20:25:03 [App     ] error: opening file 'depth0216.dmap' for reading depth-data
20:25:05 [App     ] error: opening file 'depth0193.dmap' for reading depth-data
20:25:05 [App     ] error: opening file 'depth0192.dmap' for reading depth-data
20:25:06 [App     ] error: opening file 'depth0180.dmap' for reading depth-data
20:25:11 [App     ] error: opening file 'depth0046.dmap' for reading depth-data
20:25:12 [App     ] error: opening file 'depth0035.dmap' for reading depth-data
20:25:12 [App     ] error: opening file 'depth0034.dmap' for reading depth-data
20:25:21 [App     ] error: opening file 'depth0113.dmap' for reading depth-data
20:25:32 [App     ] error: opening file 'depth0365.dmap' for reading depth-data
20:25:32 [App     ] error: opening file 'depth0369.dmap' for reading depth-data
20:25:32 [App     ] error: opening file 'depth0370.dmap' for reading depth-data
20:25:38 [App     ] error: opening file 'depth0398.dmap' for reading depth-data
20:25:38 [App     ] error: opening file 'depth0399.dmap' for reading depth-data
20:25:38 [App     ] error: opening file 'depth0400.dmap' for reading depth-data
20:25:38 [App     ] error: opening file 'depth0402.dmap' for reading depth-data
20:25:56 [App     ] Point-cloud saved: 8 points (0ms)
20:26:03 [App     ] Scene split (2e+07 max-area): 50 chunks (1m57s227ms)
20:26:04 [App     ] Scene saved (3ms):
    19 images (19 calibrated)
    6482 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (2ms):
    13 images (13 calibrated)
    3716 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (1ms):
    7 images (7 calibrated)
    2289 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (0ms):
    0 images (0 calibrated)
    0 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (8ms):
    20 images (20 calibrated)
    6352 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (14ms):
    27 images (27 calibrated)
    10067 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (15ms):
    35 images (35 calibrated)
    10579 points, 0 vertices, 0 faces
20:26:04 [App     ] Scene saved (10ms):
    29 images (29 calibrated)
    8943 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (26ms):
    53 images (53 calibrated)
    21698 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (3ms):
    16 images (16 calibrated)
    5855 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (13ms):
    22 images (22 calibrated)
    10683 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (15ms):
    26 images (26 calibrated)
    11753 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (12ms):
    32 images (32 calibrated)
    10342 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (7ms):
    16 images (16 calibrated)
    5933 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (10ms):
    21 images (21 calibrated)
    7328 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (21ms):
    45 images (45 calibrated)
    16404 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (16ms):
    38 images (38 calibrated)
    14221 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (6ms):
    11 images (11 calibrated)
    5917 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (7ms):
    18 images (18 calibrated)
    5763 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (8ms):
    24 images (24 calibrated)
    6549 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (6ms):
    18 images (18 calibrated)
    4920 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (12ms):
    32 images (32 calibrated)
    10797 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (4ms):
    28 images (28 calibrated)
    6334 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (2ms):
    12 images (12 calibrated)
    2794 points, 0 vertices, 0 faces
20:26:05 [App     ] Scene saved (2ms):
    12 images (12 calibrated)
    3380 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (6ms):
    28 images (28 calibrated)
    9537 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (8ms):
    36 images (36 calibrated)
    12608 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (4ms):
    21 images (21 calibrated)
    8342 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (14ms):
    37 images (37 calibrated)
    12316 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (28ms):
    51 images (51 calibrated)
    20730 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (5ms):
    34 images (34 calibrated)
    9340 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (14ms):
    41 images (41 calibrated)
    19058 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (12ms):
    28 images (28 calibrated)
    10443 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (12ms):
    28 images (28 calibrated)
    10904 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (5ms):
    13 images (13 calibrated)
    4623 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (1ms):
    9 images (9 calibrated)
    1137 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (24ms):
    38 images (38 calibrated)
    18420 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (6ms):
    10 images (10 calibrated)
    4881 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (11ms):
    29 images (29 calibrated)
    8772 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (8ms):
    17 images (17 calibrated)
    7483 points, 0 vertices, 0 faces
20:26:06 [App     ] Scene saved (28ms):
    47 images (47 calibrated)
    21315 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (3ms):
    12 images (12 calibrated)
    4510 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (23ms):
    52 images (52 calibrated)
    20515 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (15ms):
    28 images (28 calibrated)
    14725 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (6ms):
    21 images (21 calibrated)
    5524 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (1ms):
    15 images (15 calibrated)
    2664 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (2ms):
    10 images (10 calibrated)
    3331 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (1ms):
    7 images (7 calibrated)
    894 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (9ms):
    39 images (39 calibrated)
    7006 points, 0 vertices, 0 faces
20:26:07 [App     ] Scene saved (10ms):
    53 images (53 calibrated)
    18547 points, 0 vertices, 0 faces
20:26:07 [App     ] MEMORYINFO: {
20:26:07 [App     ]     VmPeak:  6554220 kB
20:26:08 [App     ]     VmSize:  5320520 kB
20:26:08 [App     ] } ENDINFO

suggests that certain depth maps are malformed, which could perhaps be a reason for the issue?

xiaoxiae commented 9 months ago

The depthmaps in question are actually all missing, not malformed.

cdcseacave commented 9 months ago

pls check the log for the first step to see why some depth-maps were not estimated; I'd appreciate if you can help debugging this

xiaoxiae commented 9 months ago

This is the log of the first command:

DensifyPointCloud-2402011542308B498E.log

The issues seem to persist, I don't see why they weren't generated.

I tried re-running the pipeline overnight (with --resolution-level 2) and encountered the same issue. I will do so now again with -v 3 and see if it reveals anything.

xiaoxiae commented 9 months ago

Rerunning the pipeline with

  1. DensifyPointCloud scene.mvs --resolution-level 2 --iters 4 -v 3 --cuda-device -1 --fusion-mode 1
  2. DensifyPointCloud scene.mvs --resolution-level 2 --iters 4 -v 3 --cuda-device -1 --sub-scene-area 20000000
  3. DensifyPointCloud scene_0000.mvs --resolution-level 2 --iters 4 -v 3 --cuda-device -1 --dense-config-file Densify.ini

segfaults on the 3rd command (this time with CUDA error: invalid argument (code 1)) and yields the following log files:

  1. DensifyPointCloud-2402021007118B4721.log
  2. DensifyPointCloud-2402021340068B475B.log
  3. DensifyPointCloud-2402021345248B475F.log

The last line before the crash is

13:48:21 [App     ] Reference image  22 paired with 4 views: 873(1.74scl) 875(1.90scl) 879(1.00scl) 874(1.93scl) (609 shared points)

which likely crashes because depth maps 873 and 875 are missing.

Looking over the logs, I don't see anything overtly suspicious for why the depth maps are not created.

I'm now going to run badblocks on the disk storing the depthmaps to ensure this is not a hardware issue. Besides that, I'm out of ideas and am happy to try out whatever you suggest!

xiaoxiae commented 9 months ago

Digging further, it seems that the InterfaceCOLMAP (run before 1, 2 and 3) only exports a subset of images (946 out of the 972 total ones):

10:07:07 [App     ] OpenMVS x64 v2.3.0
10:07:08 [App     ] Build date: Feb  2 2024, 00:09:03
10:07:08 [App     ] CPU: AMD Ryzen 7 1700 Eight-Core Processor           (16 cores)
10:07:08 [App     ] RAM: 62.74GB Physical Memory 32.00GB Virtual Memory
10:07:08 [App     ] OS: Linux 5.15.146-1-MANJARO (x86_64), which is likely because some images weren't matched. Weirdly enough, 
10:07:08 [App     ] Disk: 199.47GB (960.17GB) space
10:07:09 [App     ] SSE & AVX compatible CPU & OS detected
10:07:09 [App     ] Command line: InterfaceCOLMAP -i /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs -o /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/scene.mvs --image-folder /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/images
10:07:09 [App     ] Reading cameras: /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/sparse/cameras.bin
10:07:09 [App     ] Reading images: /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/sparse/images.bin
10:07:09 [App     ] Reading points: /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/sparse/points3D.bin
10:07:11 [App     ] Reading patch-match configuration: /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/stereo/patch-match.cfg
10:07:11 [App     ] Reading depth-maps/normal-maps: /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/stereo/depth_maps/ and /app/dst/new-wall-pipeline-smichoff/03-LargeDenseCheckpoint/project/mvs/stereo/normal_maps/
10:07:11 [App     ] Exported data: 946 images, 311337 points (2s467ms)
10:07:11 [App     ] MEMORYINFO: {
10:07:11 [App     ]     VmPeak:   377528 kB
10:07:11 [App     ]     VmSize:   345276 kB
10:07:11 [App     ] } ENDINFO

Weirdly enough, the difference is 972 - 946 = 26 but the reading error happens for 27 depthmaps:

13:45:01 [App     ] error: opening file 'depth0966.dmap' for reading depth-data
13:45:02 [App     ] error: opening file 'depth0896.dmap' for reading depth-data
13:45:02 [App     ] error: opening file 'depth0875.dmap' for reading depth-data
13:45:02 [App     ] error: opening file 'depth0873.dmap' for reading depth-data
13:45:03 [App     ] error: opening file 'depth0843.dmap' for reading depth-data
13:45:03 [App     ] error: opening file 'depth0842.dmap' for reading depth-data
13:45:03 [App     ] error: opening file 'depth0834.dmap' for reading depth-data
13:45:03 [App     ] error: opening file 'depth0833.dmap' for reading depth-data
13:45:05 [App     ] error: opening file 'depth0712.dmap' for reading depth-data
13:45:05 [App     ] error: opening file 'depth0711.dmap' for reading depth-data
13:45:05 [App     ] error: opening file 'depth0706.dmap' for reading depth-data
13:45:08 [App     ] error: opening file 'depth0221.dmap' for reading depth-data
13:45:08 [App     ] error: opening file 'depth0216.dmap' for reading depth-data
13:45:08 [App     ] error: opening file 'depth0193.dmap' for reading depth-data
13:45:08 [App     ] error: opening file 'depth0192.dmap' for reading depth-data
13:45:08 [App     ] error: opening file 'depth0180.dmap' for reading depth-data
13:45:09 [App     ] error: opening file 'depth0046.dmap' for reading depth-data
13:45:10 [App     ] error: opening file 'depth0035.dmap' for reading depth-data
13:45:10 [App     ] error: opening file 'depth0034.dmap' for reading depth-data
13:45:12 [App     ] error: opening file 'depth0113.dmap' for reading depth-data
13:45:14 [App     ] error: opening file 'depth0365.dmap' for reading depth-data
13:45:14 [App     ] error: opening file 'depth0369.dmap' for reading depth-data
13:45:14 [App     ] error: opening file 'depth0370.dmap' for reading depth-data
13:45:15 [App     ] error: opening file 'depth0398.dmap' for reading depth-data
13:45:15 [App     ] error: opening file 'depth0399.dmap' for reading depth-data
13:45:15 [App     ] error: opening file 'depth0400.dmap' for reading depth-data
13:45:15 [App     ] error: opening file 'depth0402.dmap' for reading depth-data
xiaoxiae commented 9 months ago

badblocks on the disk in question produced no errors so the problem likely isn't in the disk.

xiaoxiae commented 8 months ago

Ran MemTest86 to check for RAM issues (possibly corrupting the DMAPs on export), no fails there.

xiaoxiae commented 8 months ago

Rerunning the code on the machine the issue was present on and also on a different one to see if the problem is with the data.

xiaoxiae commented 8 months ago

The errors occur for the same depth maps when run multiple times.

xiaoxiae commented 8 months ago

I'm at a loss for what to try at this point, let me know what I should do if you have any ideas.

xiaoxiae commented 8 months ago

Found a minimal reproducible example.

Taking this COLMAP project: 1-colmap-input.zip I ran COLMAP's image_undistorter, producing the following: 2-after-undistortion.zip

Using OpenMVS, I ran

  1. InterfaceCOLMAP -i . -o scene.mvs --image-folder images
  2. DensifyPointCloud scene.mvs --resolution-level 2 --iters 4 --cuda-device -1 --fusion-mode 1
  3. DensifyPointCloud scene.mvs --resolution-level 2 --iters 4 --cuda-device -1 --sub-scene-area 20000000
  4. create Densify.ini with Optimize = 0
  5. DensifyPointCloud scene_0000.mvs --resolution-level 2 --iters 4 --cuda-device -1 --dense-config-file Densify.ini

The last step crashes with 

malloc(): unsorted double linked list corrupted
Aborted (core dumped)

Simply running DensifyPointCloud scene.mvs --resolution-level 2 --iters 4 --cuda-device -1 works perfectly fine and produces the correct model.

Here are all produced logs and files: 3-after-crash.zip