Support Website Contact Support Blog

crash, step 2, densification

Why does it crash during point densification, 3.5 hours in, group 1.
project is 800 25mb jpg images.
400 gb free space on C drive.
Problem is when it crashes the log goes with it so there is no record of what happened.

Any ideas? My last project was 1,100 same size images with no problem.


Hi @jamie,

Well if you have reached the second step the *.log file should have been generated.
It should be located there:

Could you also share with us the quality report?

My guess is that you where running out of RAM or CPU since these components are the most used during step 2.

In order to troubleshoot, we need the *.log file otherwise we will not be able to understand what is going on.


Thank you for responding Marco.
I’ve attached the report and log below. Thanks also, now I know where to find the log now even when p4d closes. Since I’ve tried processing this file several times the log was too large to upload whole. I truncated the log, deleted several days prior to yesterday and a lot of the repetitious image loading lines. I left a blank line where I deleted the repetitious lines in step 1 and another where step 2 starts.

I originally ran step 1 with the image GPS exif data. Then cleared the image exif data (my GPS is 1 hz and traveling 11 meters per second so not accurate), added 3 GCPs and a processing area of about half the project then reprocessed. That appeared fine so continued to step 2 and 3. It appears to me to fail after densification.

I so appreciate your help.

report.xml (18.9 KB)

log to marco 2 13 20.txt (160.8 KB)


What do you mean by "

Do you mean that you ran step 1 and then you go in the image and modify the EXIF?
This is absolutely not recommended.
I would recommend you to follow this article:

I could not spot any obvious error in your log.
But if you clean the EXIF, it could be the source of the error.


Thanks Marco, I’ll try it without clearing the exif data. In the past I have found projects georeferenced only to GCP are more accurate in absolute terms rather than including my gps reference. I think the gps, which shows 1 to 3 images taken from the same lat long, competes with the GCPs for the final position. Is this feasible?



Well, this statement surprises me. It could be true for some projects though.
However, during the first step the calibration is based on the images’ geotags meaning that our algorithm will use the image geolocation to reconstruct the model.
Processing images without geotags and then georeference it using GCPs is feasible.
But you should not clean the EXIF after step 1 since they have been used during step 1.
Good luck,