Pyramids versus Overviews

Question

I am trying to understand the performance differences between building overviews and building pyramids.  I have 600 MB tiled orthophotos and I want to know if I should skip building the pyramids and use overviews entirely.  Otherwise, I am guessing I would need a few levels worth of overviews on top of the pyramids.  If I choose to apply some color-balancing and other image processing, would this make overviews even more beneficial.  The mosaic datasets only need to be setup one time.  The source data won't change.

Thanks,

Nathan

deleted-user-yIALGElyA_K9 · Answer

Hey, Joe? What did you end up finding to be the best practice? I have a similar scenario to yours and am trying to decide between pyramids and overviews.

Thanks!

Jacob

Indiana DNR Water

JoeBryant2 · Answer

I'd like to re-iterate Nathan's question: What is meant by > than 3,000 "columns"?

Our orthophotography mosaic dataset references 79 different TIFFs; each TIFF is on average around 8k x 8k pixels. Do you mean greater than 3,000 pixels per image?

I'm trying to determine if I should check "Force Overview Tiles (optional)" when using the Define Overviews tool, so that overviews are used instead of pyramids, since the overviews are supposed to be more efficient.

NathanHeick · Answer

What about the case where the source rasters are 12,000 by 18,000 and tiled by the vendor?  We might want to do some image processing using a referenced mosaic dataset, but we want the source mosaic dataset to remain untouched.  When do pyramids outperform overviews?