Thanks for the fix, Bob. Based on your response I'm not sure you'll welcome this additional bug report, but what the hell, you can always ignore it if you like, and this way at least you'll know. :) (Note that I can't say for sure that this is a brand new bug; but I can verify that it doesn't exist in AFNI compiled on 03-01-2013, which is the other version I have laying around.)

Here's the scenario: I have two versions of a statistical overlay; in one version, the stats have been masked to an anatomical, so there's no stray extra-brain activations. In the other version, no such masking has occurred. As a result, the range of values in the un-masked overlay is larger than in the masked overlay, as you would expect. For reference:

Masked: -.399 to .67
Unmasked: -1.03 to 2.28

If you open a single AFNI controller, all works as it should: the color map reflects the ranges of the loaded overlay. In this example, since the masked version has a smaller range, the overlay colors show more variety, since the same number of colors are spread out over a smaller interval; the unmasked version is the opposite.

Bug #1 manifests when you open a second controller (B) and set its overlay to the other stats file; the issue is that the color map in B reflects the range in controller A's overlay, instead of controller B's. So if A has the unmasked version, and B the masked, B will show less color variation even though its overlay has a smaller range. This is complicated to write but the idea is pretty simple. I'll link two pics, each of a sagittal view of A,B, one where A=masked (more color variation) one where a=unmasked (less.)

[www.dropbox.com]
[www.dropbox.com]

Bug #2: if you close all the windows, and close controller B, controller A keeps the same colormap; so even if you open the unmasked dataset, it's using the former masked colormap. I was screwing around with this trying to get it to reset, and I selected an anatomical for the overlay; and after doing so, was unable to get any color to display, even after switching to a stats overlay again, closing windows, etc. I had to shut down AFNI entirely.

Bug #3: I'm not entirely sure this is a bug, but as long as I'm reporting... When I open [either] of the aforementioned stats files, and select ten colors, then I get a slew of results, some red (positive) some blue (negative.) Clicking around these red and blue blotches reveals what you'd expect, namely, red ones have positive thresholds and beta values; blue ones have negative. However, when you click the 'Pos?' checkbox, not only do the blue chunks disappear, but huge amounts of the red chunks also disappear. Oh wait, I see why: the color bar drops to 8 colors (from 10) and the bottom color appears to be no color, just a grey crosshatched section representing nothing, I guess -- so the weaker activations seem to disappear. If I manually set it to 11 or higher, there is no grey 'no color' mapping, and all the regions come back. I just verified that this behavior happens in the build from March 2013, so it's been doing this for a while. Is this behavior sensible in a way I don't understand, or is it a bug?

Thanks for all your work, and for being so responsive! I can't imagine another project of this scope where a random user posting on a msg board would get such a quick response for a bug of such magnitude.

Shane