Instead of a single cross, two dim crosses appeared on the screen. This was first attributed to spin compensation. The FG-100-V automatically compensates for pipelined processing delay by changing the mapping between the frame buffer and the screen (and the host) so that the pipeline delay is invisible. This is called spin compensation. Pixels which are fed back through the Feedback/Input LUTs for processing experience a delay. The pixels are written back to memory at a coordinate offset from the original positions, since the frame memory is scanned sequentially. This read/modify/write process results in the image in the frame memory being shifted to the right by the amount of the pipeline delay (sixteen pixels in zoom factor 1). The image shifts once per 1/30 second frame time. The problem was determining when the spin compensation occured. It could be in the odd frame time, or in the even frame. The system was first designed to compensate for spin delay, since the pixels of the cross were protected from geing over written by the next frame, in every odd frame time when drawing the red cross, Drawing the cross on even frames was tried. There was some improvement, but the dim crosses still occured in 1/3 of the screen.
This problem was solved by not trying to compensate for the red cross by setting the bit for the cross in the video acqusition mask buffer to be unprotected. Spin compensation happens at each even vertical retrace. If a cross is drawn, it will be at the correct position until the start of the even frame, at which point everything will shift. However, this is not the whole story. The image in the screenbuffer is being built in real time from the previous image and the image from the camera, at the same time that the image is being displayed, and at the same rate. As soon as a pixel has been displayed, if it is unprotected, it is replaced with a new one. This new pixel always contains all zeros (black) in the planes containing the cross image. This means that the cross is erased as soon as it is displayed, which means that there is no problem with spin compensation. By the time spin compensation is done, the cross is gone. So when the cross is drawn at the start of the even frame, spin compensation has just taken place. The video board will display and erase the cross before the next vertical retrace takes place, at which time a new cross will be drawn. The position of the cross is updated on every frame, and there are no residuals.