Fail another building orthosias image

after fail so many times. I thought maybe my mutibands problem.
I download a sample data from pix4d.
Dataset link is here.
dowload link:
Number of images 36
Image size 4592x3056
Image geolocation coordinate system WGS84

After dowloading, I processed in webODM. Failed again Surprised me.

If webODM fails often, how do we know whether it is data problem or program problem?

Interesting that you’re having problems with this dataset. I’ve processed it many times in the past.

I’m checking it again right now with my settings, then I’ll re-process it with your settings.

That WebODM instance you’re using, is that running on your own hardware or is that WebODM Lightning?

1 Like

My own webODM.

My hardware:
host: Intel [email protected], 16GB Ram, win10, SSD256GB
Virtual PC: 1CPU, 10981MB

You set your mesh octree depth to 18. Why? You should probably never set that above 13. That would absolutely cause it to crash. We probably need some guardrails for this.

But, summary: if you are testing to see if ODM works on a typical dataset, I recommend not using exotic settings. Use some defaults, some pre-baked settings. Then slowly vary from defaults. And I wouldn’t mess with settings you don’t know the effects of and conclude ODM doesn’t work. :smiley: That’s going to waste a lot of your time and not be much fun for anyone.

I choose 3D preset. not anything changed.

Something strange happened. Choose 3D preset, go in and move that value back down to 12 (which should be the default) and rerun. It should complete.

1 Like

I always use default first. As you see, all my muti-bands datasets failed. Then I try one of my DJI 3D dataset, it works in webODM anytime.
Then I try other dataset from others including OdM samples and pix4D. The result is frustrated.
At the same time, I fly some small datasets special for webODM to check whether my previous dataset s are to big for webODM under my laptop. At the sametime, I processed them in pix4D field all successes and finished in under 1 hour.

That is why the different make me thinking maybe there are something wrong with my vc setting or webODM setting. I try to find out some mistakes I properly made but I am not aware.

Yup: we’ll help you out with any learning curve, and your dataset that failed in multispectral was a new one for us and should be fixed in WebODM in a few weeks now that it’s been fixed in OpenSfM.

It’s strange that the default for 3D was set so high: that should never happen on its own. But, once you reduce that value, this dataset should run to completion.


Thank you. after finish this all default setting running, I will try 3D presetting again.

I really appreciate your time and your help.

The reason that I try to use webODM is the opportunity of the further process using FIELDimageR.
I found FIELDimageR can count trees and other customer setting. This is Pix4D can’t do.


I just found this log

18:32:53 [App ] CPU: Intel(R) Core™ i7-8550U CPU @ 1.80GHz (1 cores)
18:32:53 [App ] RAM: 10.47GB Physical Memory 3.51GB Virtual Memory
18:32:53 [App ] OS: Linux 4.19.130-boot2docker (x86_64)
18:32:53 [App ] warning: no AVX compatible CPU or OS detected

But you can see my vc setting is 10GB.

Is that webODM is another VC in the VC?

My physical RAM is 16GB. VC is 10GB, the webODM showing 3.51GB?

Pix4D field is not doing any 3D reconstruction, unlike your settings for WebODM.


Yes. It doesn’t do 3D. I dont have other pix4D license to test.

Maybe my problems is VC setting? :astonished:

Just update to 4 processors.
I hope it can faster.

1 Like

It’s not going to be exactly 4x as fast, but it’ll be significantly faster :slight_smile:

1 Like

how about the RAM? Why does it show only 3.51GB?

Since ODM doesn’t support my data yet, can it do RGB only now?
Sorry for so many questions.

Run under default. Failed again.

File “/code/opendm/”, line 338, in run
File “/code/opendm/”, line 338, in run
[Previous line repeated 3 more times]
File “/code/opendm/”, line 319, in run
self.process(self.args, outputs)
File “/code/stages/”, line 24, in process
File “/code/opendm/”, line 162, in screened_poisson_reconstruction’{bin} --in {infile} ’
File “/code/opendm/”, line 79, in run
raise Exception(“Child returned {}”.format(retcode))
Exception: Child returned 137

Since ODM doesn’t support my data yet, can it do RGB only now?

Other RGB data: yes. Other multispectral data: Yes. Your RGB or multispectral data from that sensor: no.

Run under default. Failed again.

That is a different dataset. Can you run the original dataset with default settings? Otherwise, we are comparing apples and oranges and chasing what may be several different issues.

Regarding VM setup, I don’t run on Windows, but @saijin_naib can point you to how to optimize your VM settings. You surely won’t be able to do much if you only have 3GB of RAM allocated.

1 Like