In the last stage of generating a preliminary model the 2 or 3 class averages from the previous step are used to build a 3D model. This step is performed as part of startcsym or starticos.
In the previous step, euler angles were determined for the 2 or 3 class averages generated earlier. These images are then combined to form a preliminary 3d model. This model, being based on only 2 or 3 images, will look really really bad. This is fine. The refinement loop will generally converge with even a really bad starting model. If you look at this model, don't be surprised that it looks bad. Just take a leap of faith and try the refinement procedure.
Last Modified: