At 11/24/2014 08:23 AM, Bruce Fischl wrote:
Hi Graham
have you tried freeview? We stopped development on the tktools some time ago.
cheers Bruce
Hi Bruce, thanks for replying.
Yes, I did proceed to try the same task in Freeview (which in general has a nicer UI, to be sure). Some notes:
Test case: For subject bert, display principal curvature k1 (of smoothwm) on smoothwm and on inflated. Using Freeview latest version, in Linux Mint 15.
Freeview Problems/Bugs ----------------------- 1. The Threshold frequency histogram widget with color adjustment handles is a great idea, but has several problems.
a. Horizontal range of the freq hist defaults to entire range of the variable, which, for k1, is one to two orders of magnitude larger than the range of the bulk of the data. Consequence: the bulk of the data appears compressed to a tiny range near zero, in a tiny number of bins, and it's impossible to set the adjustment handles.
This mechanism needs to accept manual upper and lower limits (and options to clip outliers to that value, or exclude them).
b. The "Use percentile" checkbox doesn't seem to do anything that I could discover.
c. The mouse hit detection for selecting and dragging the color adjustment handles often doesn't work. This is some combination of:
-- The target area of the handles is too small.
-- Clicking the handle but missing instead moves some other handle to the clicked location
-- Sometimes the hit detection seems to think you're pressing/clicking on the mirror-image side of the graph. Press at +10, and it's like you pressed at -10.
2. It would be great to be able to clone (or save/load) display settings associated with one surface onto another. This way, time spent configuring the display of a particular variable on, say, smoothwm, would not have to be redone (inevitably imperfectly) to display the same variable on another surface like inflated, which is a frequent need.
mris_convert ------------- 3. crv->text, text->crv. I didn't find a tool that would convert between crv and text format and back, to make it easy to use some other tool to clip k1 to tractable limits.
Maybe mris_convert does this, but if so, the help for that program could use some revision: --------------- Usage: mris_convert [options] <input file> <output file> ... -c <scalar file> input is scalar curv overlay file (must still specify surface) ---------------
a. 'curv' perhaps also includes 'crv'?
b. Why is 'surface' needed? The curv/crv files contain everything they need for conversion. Requiring surface makes it seem like the conversion is going to do something more elaborate than just convert binary <--> text.
c. And does 'surface' mean just an argument like 'orig' or 'pial', or the name of a surface file? Perhaps it just means that mris_convert still looks for an <input file> argument, but doesn't use it?
Anyhow, I didn't trust that this was going to do what I intended.
------- I DID end up getting the display I wanted: I inspected a k1.crv file in a hex editor, discovered it was the same as old "new" .curv format, and wrote some python code to read/write/filter such a file. With this I could produce a clipped version of k1, which did work tractably in Freeview and produced a nice display.
-- Graham