colmap_output format

[q-eldar]

Hi! Thanks for sharing your work.

I'm trying to integrate PatchmatchNet with colmap, at stage stereo_fusion. My input to colmap is dense folder generated with colmap_output.py from dtu/scan1

.
├── images
│   ├── 00000000.jpg
.
.
│   └── 00000048.jpg
├── sparse
│   ├── cameras.txt
│   ├── images.txt
│   └── points3D.txt
└── stereo
    ├── confidence_maps
    ├── consistency_graphs
    ├── depth_maps
    ├── fusion.cfg
    ├── normal_maps
    └── patch-match.cfg

7 directories, 54 files

Issue: colmap_output.py generated patch-match config and fusin configs are not consistent with colmap.

PathmatchNet patch-match.cfg:

00000000.jpg
00000010.jpg, 00000001.jpg, 00000009.jpg, 00000012.jpg, 00000011.jpg, 00000013.jpg, 00000002.jpg, 00000008.jpg, 00000014.jpg, 00000027.jpg
00000001.jpg
00000009.jpg, 00000010.jpg, 00000002.jpg, 00000000.jpg, 00000008.jpg, 00000013.jpg, 00000014.jpg, 00000012.jpg, 00000007.jpg, 00000015.jpg
00000002.jpg
00000008.jpg, 00000001.jpg, 00000007.jpg, 00000009.jpg, 00000003.jpg, 00000015.jpg, 00000014.jpg, 00000016.jpg, 00000006.jpg, 00000010.jpg
00000003.jpg
00000007.jpg, 00000006.jpg, 00000002.jpg, 00000004.jpg, 00000008.jpg, 00000005.jpg, 00000017.jpg, 00000016.jpg, 00000001.jpg, 00000015.jpg
.
.

colmap patch-match.cfg:

00000000.jpg
__auto__, 20
00000001.jpg
__auto__, 20
00000002.jpg
__auto__, 20
00000003.jpg
__auto__, 20
00000004.jpg
__auto__, 20
.
.