The colour chips have a repeating 2x2 filter grid pattern, either as Red, Green, Green, Blue (SXV-H9C) or as Cyan, Yellow, Green, Magenta (MX7C / MX5C). At first sight, this seems to reduce the effective resolution to that of a 2x2 binned imager, but it is not as extreme as this. For example, imagine a colourless scene (just greyscale) and assume that the filters all have an equal transmission in their passbands (manufacturers try to approach this state of affairs). The CCD ouput signal resolution will then be unaffected by the presence of the filters, as there is no contribution from them to image 'modulation'. All of the variations from pixel to pixel are due to scene detail and the resolution is the same as for a mono CCD. In an astronomical scene, the colour saturation is typically quite low and so it approximates the situation described above. This means that an algorithm which can take into account the filter modulation, will be able to extract an almost full-resolution 'brightness' image from the data and, as brightness is much more important than colour to the human eye, you can reconstruct a very sharp image. The colour information is less well resolved, as it is only 1 or 2 in every 4 pixels which responds strongly to any one colour in the image, but the colour detail is much less important to the 'sharpness' of the image. By combining the low resolution colour data with the high resolution brightness data, we can construct a sharp colour image that is only slightly less resolved than one from a mono CCD.