ContextCapture Processing is CPU and GPU intensive. To prevent memory overflow in case of large datasets, it is recommended to apply tiling to the reconstruction.
In ContextCapture, when creating a new reconstruction, a warning appears if the block contains a large quantity of data.
In that case we recommend using tiling
The Tiling setup mode options are accessible from the “Spatial Framework” tab of your reconstruction.
There are 3 types of tiling:
There are two aspects to be considered regarding the tile size:
The threshold of when it is no longer beneficial to increase the tile size depends on the global performance of your machine.
For typical processing using a computer with a good CPU, good GPU and 64 Gb of RAM, the optimal maximum RAM target (depending on your tile size selection) is 25 Gb.
These types of tiling are defined by their size (in meters).
When selecting the size for one of these tiling modes, an “Expected maximum RAM usage” is displayed. Adjust the tile size so that it fits your hardware capabilities and/or the optimal tiling size.
Adaptive tiling will be defined based on the maximum RAM you want to allocate to your ContextCapture Processing.
The tiling will be automatically generated so that no tile should require more RAM than the defined limit.
This type of tiling results in an optimized size and distribution of the tiles. It is recommended for projects for which the resolution can dramatically change (with the addition of close-ups for example), or for project in full 3D such as buildings, towers, bridges, etc.
If the tiling is not correctly set, or the RAM estimation is too big for your computer, the following error message may appear: “Failed: insufficient memory. You may solve this issue by subdividing the reconstruction into smaller tiles. Please review your reconstruction settings.”
If you are getting this error, we recommend adding tiling or reducing the tile size as mentioned.