An unknown alignment error has occurred and is not completing normally.
We were in the process of merging laser scan data with image scan data.
The project was first started last week.
Without specifying Control Points, it is not aligned at once, so I am trying several times.
By the day before yesterday, the alignment work was completed normally.
However, suddenly there was an unknown error from yesterday, so the Alignment operation is not being completed normally.
The attached image is a capture of the screen where the problem occurred.
Does the log [0x13010\0x10001\0x4999\0x10001] shown in the Command window explain why the error occurred?
I tried it again after the first alignment error occurred yesterday, but it happened again after 8 hours.
I tried after deleting all the Cache in the Cache folder.
I’m leaving a question because I want to check what causes the Alignment error.
Hi @Nuriarchive,
before the new alignment, have you tried also the application reset?
What is your RAM capability? Using 80 000 features in not necessary, also for such settings you will need to have around 160 GB RAM.
As you mentioned that the first alignment yesterday was successful (completed), what have you changed in your workflow?
Can you try to use the pre-defined alignment settings?
Also, maybe this kind of placing of control points is not necessary, as they are too close together:
-
application reset
I didn’t reset the program.
Is this the same way as the image below that runs with Shift + Double Click?
-
My Computer RAM
My computer RAM is 256GB
-
Using 80,000 features
What does 80,000 features mean?
It’s not a figure I’ve arbitrarily specified. -
As you mentioned that the first alignment yesterday was successful (completed), what have you changed in your workflow?
While importing laser scan data first (1031), 8,850 additional photos are imported and aligned.
The reason for continuous attempts is that when laser scan data and image scan data are merged into one, some parts are incorrectly combined and Contol Points (68) are added one by one. -
try to use pre-defined alignment settings.
I didn’t quite understand what alignment settings you were talking about.
Are you saying that you can download and apply the Alignment preset from somewhere? -
Too close Control point
Let’s try to align while reducing the number of control points.
Hi,
- Yes, that is the application reset
- OK, this seems to be OK then
- It seems like in the alignment settings you have set Max features per image as 80 000. Pre-defined value is 40 000. You can reset those settings using the reset button
You had to set it, as it is mentioned here:
(I hope that means that 80 000 features were found on the image) - Are the control points placed on both datasets correctly (minimally on three images/LSPs in both components)?
- see point 3. Also, it is possible to export and reimport those settings (using the up and down arrows
Hello.
As you told me, I tried again after resetting the program, and the result was good.
I think I changed the number of Max Features from the default value of 40,000 to 80,000.
Before that, the results were slightly different from when the alignment work was completed, but they were satisfactory enough.
Reality capture has a lot of detail options (Align, Recon, Tex)
Can you explain how the results change when each setting is changed or tell me in detail in the manual about tips to produce better results?
Ex) Tips such as “Change the ‘Image Overlap’ option from the default Medium to High when image overlap is low.”
Hi,
for most datasets the pre-defined settings are the ideal one.
The tips could vary much, as each dataset could have its specifics. But the most important for RealityCapture is well captured dataset. For more complex explanation of the parameters please check our online Help: RealityCapture Help
Thank you for your answer.
So what does it mean to change the Max Features Fer image from 40,000 to 80,000?
What does Max Features Fer Image mean?
It is a number of the features which will be found on each image. These features are later used in the alignment process.
So, if you double the number, the process will use more RAM to find those features and it will last longer to find them.
Memory consumption during the alignment phase depends on the number of images (not size) and the number of detected features per image. For the default setting of 40 000 features per image (Alignment setting), you can expect the following boundaries:
- 2,000 images - 16GB RAM
- 4,000 images - 32GB RAM
- 8,000 images - 64GB RAM
- 16,000 images - 128GB RAM
By decreasing the number of detected features to half you can approximately decrease the memory consumption by half as well. The approximate formula is: RAM = features x images x 200 bytes.