Freezes while merging depth map of project with 1068 images

Flew 180 project with my Phantom 4 pro. I wanted to see what I could throw at WebODM and it seems like I found my limit. Been running for three days on a AMD F-X8350 16GB of ram. Not the best machine but was an extra one to try this on. I was wondering if there was any sort of way I could figure out the exact reason it is freezing or seems frozen on

2018-06-24 07:46:18,726 INFO: Pruning depthmap for image DJI_0419.JPG
2018-06-24 07:46:18,746 INFO: Pruning depthmap for image DJI_0229.JPG
2018-06-24 07:46:18,776 INFO: Pruning depthmap for image DJI_0651.JPG
2018-06-24 07:46:23,634 INFO: Merging depthmaps

I want to SSH into the box but not sure where I could see any more descriptive logs. It is running this all through docker on an Ubuntu 16 machine.

Just grasping at straws to where to hunt down maybe new/different settings to get this large project to process.
(I had it resize i think to 1024 instead of 2048 and had it select a calculation of DSM dem etc.) Everything else should of been default.

My docker stats showed this:

64bf5aa6d060 0.02% 258.9 MiB / 15.69 GiB 1.61% 1.97 GB / 3.01 GB 90.7 GB / 38.6 MB 6
6ca8097ba0f8 0.00% 11.04 MiB / 15.69 GiB 0.07% 1.98 GB / 3.22 GB 1.99 GB / 287 MB 4
4fe001afb42e 0.00% 122.6 MiB / 15.69 GiB 0.76% 2.2 GB / 53.7 MB 88.7 GB / 23.2 GB 15
634b4f84d748 0.00% 49.31 MiB / 15.69 GiB 0.31% 3.63 GB / 4.26 GB 111 GB / 20.1 GB 6
5918262e4810 0.09% 60.3 MiB / 15.69 GiB 0.38% 117 MB / 45 MB 81 GB / 1.49 GB 4

Not sure what else to try or do or can opendronemap/webODM just not handle large images at that number of them (about 1068).

You’re not going to be able to process 1068 images on 16 GB of RAM. Not even close. Even if you make it pass the merging depthmaps step, the process will fail during texturing.

You need more memory, probably at least 64GB, but perhaps more.

1 Like

No way to split up and tie in projects I guess? Even my beefiest machine (my gaming pc) is only 32GB. Seems bizarre to me the memory requirements would be so high.

We have some nascent work in this direction:

It is certainly worth a try with your dataset. Unfortunately it is command line only at this time.

Hi Shane,
I am also trying to do a large number of pictures (3000),
Were you able to complete the process with the split and merge process mentioned by smathermather?

I’ve done 15k with split and merge. It doesn’t yet solve everything but we’ll worth a try.

how many pictures will 64 gigs be good for? 500…1000…2000??