Some comment is justified on the manner in which the mirrors are handled during a showing of the picture. The philosophy should always be to make the matchline most unobtrusive. This generally means that picture match should always be sacrificed for best blend. Since picture match usually applies to some small relatively insignificant detail, it is not good practice to introduce a large top to bottom disturbingly obvious band into the picture with an abrupt movement of the whole panel. This is a very small return for a high price. Of course there are times when the detail is the subject of attention and then of course it is worth sacrificing the blend, but the instances are rare. Note that the presence of a good blend, the picture must be searched to find a picture error while the audience attention is usually centered elsewhere, whereas no effort is required to discover a bad blend.

    The printer attempts to overprint a neutral wedge over the positive picture varying linearly in transmittance (not density) so that the pictures may be combined optically on the screen. This is a much different problem than combining exposures from the three strips onto a single frame of a single film strip before processing, a relatively simple procedure. While this latter problem is perfectly possible from a single setup regardless of color or exposure level, the method used in Cinemiracle is theoretically possible for the range of print densities in which all portions of the matchline fall on the linear portion of the H & D curves for all three color components. Of course special masks can be made to match shoulder and toe conditions for grays, but this is rather impractical. Since sky presents the worst problem, the masks are made to give the best blend at a gray sky density in the vicinity of 0.4.