Dear Yang,
it is true that the CMP is thought to follow diffusion data processing from the Dicom files to the final result.
Anyway the Dicom step can be by-passed by copying your Nifti files in the NIFTI folder (T1.nii.gz; DSI.nii.gz or DTI.nii.gz or QBALL.nii.gz; eventually T2.nii.gz; eventually fMRI.nii.gz).
The DICOM converter step also saves the first volume of the diffusion dataset (NIFT/DSI_first.nii.gz or DTI_first.nii.gz or QBALL_first.nii.gz), and the first volume of the diffusion dataset resampled to 1x1x1 mm voxel size (Diffusion_b0_resampled.nii.gz). For this reason if you want to skip the DICOM converter step you should also provide these files.
Finally, before running each step the CMP is checking for the existence of at least one Dicom file in the RAWDATA/T1 and RAWDATA/
diffusion folders. You can copy just one whatever Dicom file in each of these folders and skip the DICOM converter step.
Of course this process is tricky..
I'm going to release these days a new version of the CMP with easiest support for the Nifti.
Alessandra