Hi Yuta

Well, the error estimates shown in rows 2 and 3 in that figure cannot be reproduced by DTIPrep (It is not difficult to do so, we just don't have the manpower to add this, as there is limited interest for that particular option).

With respect to the top row of that figure, that simply displays the surviving motion corrected gradients, that should be possible in DTIPrep. For that 
- load the QCed DWI data as a new dataset into DTIPrep
- you may need to go to protocol and generate a default protocol, or you could load the protocol used to generated the QCed DWI
- go to the 3D display and you should be all set, i.e. you should see the sphere with directions etc 
- set opacity to 1.0 (you may need to move it around to have an effect, there's a bug on that slider) to only see the ends of the gradient vectors