Subscribe to this thread
Home - General / All posts - Transform Filter (population variance) unexpected behavior
157 post(s)
#12-Apr-23 14:28

When running the Transform Filter (population variance) the output includes additional interpolated pixels along the right and bottom of the image.

Is this expected behavior?

The elevation layers I'm working with are 1m USA National Elevation Data set DEMs.

I'm generally setting the filter radius to values higher than 8 pixels.

I can share a link to the download site for the data sets I'm using if needed.


10,447 post(s)
#20-May-23 12:51

Sorry for a belated reply.

No, that's not normal. We'd be grateful for a link to example data as well as step by step instructions on how to reproduce the issue. You can either post them here or send them to tech support.

157 post(s)
#20-May-23 16:01


Here is the webpage for the data set I was working on.

I downloaded and imported the DEM geotiff. I did not modify the imported projection information.

Using Manifold System (16-Mar-2023)

I performed a variety of tile transforms on the data set, including slope (grade), slope (aspect), and filter, (variance pop).

I used the same shape (circle), and radius (31) for each of the transforms. For the results I selected output to a new table for each transform.

I chose 31 for the radius, as this distance is half of the elevation variance across the data set.

For all of the transforms, I found that the (approximately) 62 pixels along the right and lower edges of the output images were unusually interpolated.

Any thoughts would be appreciated.


10,447 post(s)
#21-May-23 08:10

We'll check what's going on. Thanks a lot for the detailed notes and a link to the example data!

157 post(s)
#21-May-23 20:35


I would also note that the DEM geotiff imported at 10012 x 10012 pixels.

The output image for each of the tile transforms was 10048 x 10048 pixels.

Again an unexpected result.



10,447 post(s)
#23-Jul-23 15:45

I am really sorry for coming back so late once again.

We cannot reproduce the issue.

We take the file you mention, it imports as 10,012 x 10,012. (Good.) We then run the Filter transform with a fairly big radius (see the attached screen, in that screen we used a radius of 10, but we also tried 31), saving the result into a new image + table. The result of the new image is 10,012 x 10,012, as it should be. If we tell the transform to modify the original image, the result is the same, 10,012 x 10,012, as it should be.

If you are still experiencing the issue, please let us know how to reproduce it - perhaps better done via tech support.


Manifold User Community Use Agreement Copyright (C) 2007-2021 Manifold Software Limited. All rights reserved.