reality capture crash after more than ten hours of rendreding a point cloud to normal detail. TI can rendrer a preview but when i ask him to rendet1882 images out of 2200 photographs it work during all those hour to stop at the end pf the processss
Hello Robert,
Did the computer crashed and restarted or did you just get an error message? If so what error message? Check if you are using the latest graphics card drivers. Try both studio and game ready drivers. Do you have enough space for the cache? Did the crash only occurred on this project or did it happen on different projects with approximately the same number of images in the past? After this happens can you please try to search for RealityCapture.log before opening the application again and send it to us?
it is not the computer who crashy it is the software. By the way is it normal that i have the 1.0 version Beta?
It says that it was anormal and that reality capture team will check the error messagte and try to fix the problem
baronetr, the label “Beta version” does not mean that the software is unstable or that the algorithms are incorrect. The sole reason for this label is that we are still developing and adding new features to the feature package. The software will not be classified as Beta anymore once the list for the package is complete.
understood, if I rent the 750euros version for one month could i puit this version on microsoft server in order to reconstruct model with more than 20,000 images.
is it possible that because i install the software on a removable hard drive or external hard drive I have this problem
can i reinstall the software on the local hard drive
these are the notes i received when the application crash
Hello and sorry for the late response. Thank you for the screenshots.
Yes you can reinstall on local hard drive, the license is remembered.
Definitely check your GPU drivers, this issue was fixed by the drivers in the past. Then clear your cache. Maybe even reset the application. That is done by holding Shift key when double clicking on RC icon. In the screenshots I can see that you are using a reconstruction region, that is good. Try to change it slightly, make it bigger by 10 to 15% and also try to rotate it a little bit by 10 to 15 degrees. Then reconstruct it again. If there are still some problems you may try to reconstruct the scene in blocks:
https://support.capturingreality.com/hc/en-us/articles/360016441831-Huge-scene-reconstruction
What do you exactly mean by Windows server in your previous question? The operating system or something else? Data sets around 100 000 photos were reconstructed in RC. For aligning 20 000 images on a single machine with the default settings you would need 160GB of RAM. But following the link above you can reconstruct huge scenes on a computer with much less RAM.
I mean is it a good idea if reality capture permits it to put the software on a farm computer server and had many CPU to work the project.
Also why is it better to have the reconstruct box rotated?
and is it better to have Reality capture on the hard drive of the computer
“I mean is it a good idea if reality capture permits it to put the software on a farm computer server and had many CPU to work the project.”
Now I understand, it is always better to test with the demo first and check if it will run.
“Also why is it better to have the reconstruct box rotated?”
It helped in some cases of this error message.
“and is it better to have Reality capture on the hard drive of the computer”
Yes, definitely.
Many thanks last i will rent the 750 euros version for one month do yout hink i have enough memory 32gig ram to run maybe 30,000- 40,000 images i have to reconstruct a small model village of 117 houses and farm all together.
Bu the way i wish to try to participate in the mine reconstructed contest that reality capture offer, when you say georeferencied photographs if i do it with drone and camera on the ground is it okay et does the mine has to be active?
the software has crash for the third time, i did re install Reality Capture , i rotate the reconstruction box hoping for the best
Is it normal that it takes about 22 fours to reconstruct a cloup point of 2050 image in jpeg
Kindly
Robert
Hello Robert,
RC did not crash the last time during reconstruction? Did changing the reconstruction region helped?
The reconstruction process uses both GPU and CPU, the first part (Depth map calculation) of the reconstruction is GPU heavy. This is affected by CUDA cores count and frequency. You never mentioned your PC specs so it is hard to say if it took long or not. You can check this RC benchmark by Puget systems and compare your times with others:
https://www.pugetsystems.com/labs/articles/RealityCapture-Benchmark-1458/
Jakub
Allo Jakub
Many thanks for following me through the process, you were right it is better to have the reconstruction box on oblique it help. It crash while i was texturing and found many advice on the texturing process that help me.
Our Pc is a I7 core
as you will see in fact the CPU is working most of the time at 100% and the GPU is almos not working 1%, so I really think that while alligning and even reconstructiong the CPU are very utilise. the memory is used at about 23 %
Bu the way I can’t find a way to do a video even if i look on some tutorials how do we start the grabber put some movement and stop it?
Again many thanks I think i will be successful in managing the anticipated 30,000 images thisi is the portion i was tryin to reconstruct
You are welcome Robert, I am glad it helped. The GPU is used in the beginning for depth maps calculation. When you visualize CUDA you will see that it is at 100%.
To start and to stop the Grabber press these there 3 keys together: CTRL + SHIFT + ~.
~ sign is above the TAB key.
The Grabber settings are located in the Workflow tab in Application Settings.
ok thanks but where can I find the precise procedure to make a video?
look at my graphic the GPU is at 3%
it crash again on a fine reconstruction