Mosaic in Arcgis pro

Question

Hi,

I have a huge dataset (Aerial photos) and I have done Mosiac in Arcgia pro according to its help. (Mosaic datase----addraster----build footprint-------build pyramid and overview

when I created of pyramid and calculated statistics it takes long time. Does any one know how I can reduce the time or another method for Mosaic in Arcgis pro?

Thank you very much

JeffreySwain · Answer

The idea of the overviews are to improve the performance once the overviews are created. However, the processing speed is dependent on a lot of factors. Without more information about the spatial extent of the mosaic dataset, the type of input imagery used in the mosaic dataset, or the overview generation. These overviews are designed to improve the performance ultimately and if you have a lot of overviews to generate, your generation settings are turned to ignore the pyramids, or have input highly compressed images.

A few tips that I follow, is try to put the contiguous portions of the input images together and process them as a set. You can also create a mosaic dataset with other image services or other mosaic datasets. So if you have a mosaic dataset with disparate images, this is something to consider.

BashirRokniDeilami · Answer

Hi David,

Thank you for your Help. in term of trade offs for performance against the size of the overview cache generated (e.g. build the overviews/cache only in urban areas, reduce the cache/overview levels etc.).Could you please send me any documents or video,  Also,do have time for quick chat I have some questions.(I am in Toronto(Canada). which time is good for you?

Regards,

Bashir

DavidPike · Answer

A Mosaic Dataset is the best option I can think of, and I recommend it.

The steps I would take if doing it myself:

1.  Create the Mosaic Dataset in the correct projection required.  Then load the rasters into it.  DO NOT check the boxes to calculate statistics or build overviews.  Wait for it to fully complete.

2.  Calculate statistics on the Mosaic Dataset using a very large skip factor (e.g. 1000 x and y), then work your way down values, repeating depending on how long it takes until you find an acceptable amount of time taken against the accuracy of the stats.

3.  If everything displays as expected then you can Define then Build your overviews.  You will need a clear strategy of what you want to achieve by doing this as if the dataset is as large as you say, then you may have to make trade offs for performance against the size of the overview cache generated (e.g. build the overviews/cache only in urban areas, reduce the cache/overview levels etc.). If you're not aware of how to do this then I'd recommend reading up on the overview tools/watch tutorials etc.