I am unsure of your question. The plotter is a Q130 with ARMS 3.0. Is this what you mean? We are using Flexi8Pro as our print/cut software.
Method 1: You add a contour object to the image via either Effects->Contour or Arrange->Contour. Flexi adds registration marks to the print and creates a pair of jobs, one to print and one to cut. You send the cut job and the software asks you to position the blade over the first mark, set the origin, and then press 'Continue'. The plotter reads the marks and off you go.
Method 2 and 3: You add your own registration marks to your image along with a non-printing contour object. This contour object is not known as such to Flexi, only to you. Flexi does not think that this is a contour cut job. You cleverly position your registrations marks as the corners of a rectangle of known, by you, size. You print the image with your very own registration marks. and load the print into the plotter.
Method 2s.You invoke the auto sensing feature of the plotter. It finds your marks.
Method 3s. You invoke the manual sensing feature of the plotter. It tells you to position the blade at each of the marks in turn ans press Enter, where in it reads each mark.
Method 2 and 3 finale. The plotter pauses after reading the last mark and displays what it detected for X and Y distances, in turn, and prompts you to enter the actual distances. Then you send the cut job, which is your contour and either the registration marks or the bounding rectangle.
Since you seem confused that there's more than one way to do registration, I assume that you're using method #1. How many marks? What style of mark?
Method #1 is OK for smaller prints, say less than 36" or so long. Any more than that and it seems to get easily confused and fails to sense all the marks. Moreover, the longer the print, the closer to perfectly square you have to load the print into the plotter. The plotter reads the first mark and then advances to the general area where it expects to find the next mark. If that mark isn't reasonably close to right where it thinks it should be, it fails. Thus the caveat about excessively skewed media. The real problem with method #1 it that when it fails, it's very difficult if not impossible, to recover. If the Great White Fathers there at Scanvec Amiable would see fit to publish the dimensions of the registration mark's bounding rectangle, you could recover via method 2 or 3. But they don't, so, usually, you lose.
If the sensing goes OK but the actual cutting is off, especially if it starts out OK but develops an error towards the end of the cut. It's a plotter repeatability issue and you probably have some sort of media slippage happening. For whatever reasons.
