Just posted: source for an updated AFNI. Binaries for Linux are also up now, and the others are compiling at this moment.

Changes since the last update (a whole week ago) include:

- 3dfim+ now accepts a polort value greater than 2, and uses Legendre polynomials for the baseline (rather than simple powers)

- fixed a bug when AFNI displayed a non-AFNI dataset as the Overlay -- the sub-brick statistics weren't loaded at the proper time, so the color pbar range wasn't set appropriately until the next time you Switched to that dataset

- the 3dDWItoDT program (by Daniel Glen) now uses a nonlinear least squares regression to compute the diffusion tensor from the diffusion weighted images; this method is guaranteed to produce a positive definite matrix, and so the fractional anisotropy calculation will not fail

- fixed a bug in the AFNI/SUMA 'talk' functions that slowed the conversation waaay down (sorry, Ziad)

- a few minor changes that should let AFNI support datasets > 2 GB on some machines

bob cox