------------------------------------------------------------------------------

Return-Path: <chuck@GS77.SP.cs.cmu.edu>
Received: from GS77.SP.CS.CMU.EDU by ux3.sp.cs.cmu.edu id aa11690;
          26 Jan 99 0:33 EST
Received: from localhost by gs77.sp.cs.cmu.edu id aa12854; 26 Jan 99 0:33 EST
From: Chuck Rosenberg <chuck+@cs.cmu.edu>
To: mitchell+@cs.cmu.edu, danieln+@cs.cmu.edu
cc: chuck+@cs.cmu.edu
Subject: Voxel Plotter
Date: Tue, 26 Jan 1999 00:33:00 -0500
Sender: chuck@gs77.sp.cs.cmu.edu
X-UIDL: 0fa9b9911c40ce4636cc4d306dc45efd

Hi,

I've modifed the program "pff2ppm" to do the voxel plotting.  The program
is in the "code" directory.

An example image "lb-pred_t00.ppm" can be found in the backprop directory.
I'm afraid that it is not particularly impressive because of the small 
number of voxels we are predicting.

An example call is:

../code/pff2ppm -po -0.8 -ps -3.0 -ds 10000 -cm 5 -pf lb-predictability.vox -zm 1 /afs/cs/project/theo-35/fmri/02882_34/detrend.dat lb-pred 0 0 0 6

When this call is made from the "backprop" directory it generates the image 
"lb-pred_t000.ppm" which is a plot of the NRMSE data in lb-predictability.vox.

An explanation of the command line options:  (you can also just type pff2ppm)

  -pf lb-predictability.vox = specifies the plot data file
  -cm 5 = specifies the color map for plotted data, in this case spectrum
          (-cm 4) specifies the red / green map that Tom suggested
  -zm 1 = a flag which indicates that all 7 z slices should be combined 
          into a single montage
  -po -0.8 = the plotted data offset, this specifes that first the
             value -0.8 should be subtracted from the value before plotting
  -ps -3.0 = the plotted data scale, this indicates that after the offset
             is added to the data, the result should be scaled by this factor
             in this case a negative value is used so "good" values (small
             NRMSE) will appear as positive numbers after being offset
  -ds 10000 = antennuation factor for the background brain image, the
              larger this number, the darker it will be

The numbers "0 0 0 6" tell the program to use time slices 0-0 and z slices 0-6
as the background brain image.

If you have any questions, please e-mail.

Thanks,

Chuck

------------------------------------------------------------------------------

Convert data from a PFF format file to a series of PGM images.

Usage: pff2ppm <opt> <in_fname> <out_fname> <t_start> <t_end> <z_start> <z_end>

Arguments
   <in_fname>  = input  data file
   <out_fname> = output image file

   Note that using the filename stdout for the output image
   will cause it to be dumped to stdout.

Options
   -ds <float> = set data scale, default = 4000
   -do <float> = set data offset, default = 0
   -id <bool>  = invert pff data in image, default = 0
   -t <float>  = set threshold value, default = don't threshold
   -b <float>  = set scale factor for thresholded background, default = 0.5
   -zm <bool>  = merge all of the z images into a single, default = 0
   -bf <int>   = background fill value when in merge mode, default = 0
   -pf <str>   = name of plot file, default = none
   -po <float> = plot file data offset, default = 0
                 desired plot data range is -1 -> +1
                 the offset is added to the data before any other operation
   -ps <float> = plot file data scale, default = 1
                 desired plot data range is -1 -> +1
                 the data is multiplied by the scale after the offset is added
   -pa <int>  = auto scale and offset data, default = 0
                 0 = do not auto scale
                +1 = auto scale
                -1 = auto scale and multiply scale factor by -1 to reverse order
   -cm <int>   = color map (default = 5)
                0 - 0 = white, + = red,    - = blue
                1 - 0 = black, + = yellow, - = cyan
                2 - 0 = blue,  + = yellow, - = cyan
                3 - spectrum (varying brightness)
                    0 = blue, +=magenta,red, -=cyan,green
                4 - 0 = black, + = green,    - = red
                5 - spectrum (full brightness)
                    0 = blue, +=magenta,red, -=cyan,green
   -xs <int>   = x start for cropping (default = 0)
   -xe <int>   = x end   for cropping (default = 63)
   -ys <int>   = y start for cropping (default = 0)
   -ye <int>   = y end   for cropping (default = 63)
   -sc <int>   = output image scaling factor (default = 1)

------------------------------------------------------------------------------

Pretty voxel plotting of LB MSE data - full brain version - straight to gif:

rm -f lb-pred_t000.gif
pff2ppm -sc 2 -pa -1 -ds 10000 -cm 5 -pf lb-predictability.vox -zm 1 /afs/cs/project/theo-35/fmri/02882_34/stats/02882_mean.pff stdout 0 0 0 6 | ppmquant 256 | ppmtogif > lb-pred_t000.gif

------------------------------------------------------------------------------

Pretty voxel plotting of LB MSE data - cropped version - staright to gif:

rm -f lb-pred-crop_t000.gif
pff2ppm -sc 4 -xs 35 -xe 55 -ys 10 -ye 30 -pa -1 -ds 10000 -cm 5 -pf lb-predictability.vox -zm 1 /afs/cs/project/theo-35/fmri/02882_34/stats/02882_mean.pff stdout 0 0 0 6 | ppmquant 256 | ppmtogif > lb-pred-crop_t000.gif

------------------------------------------------------------------------------