Another new image, from the scanback, from the apartment block. I love the shape of the building and its light, its quite seductive and gentle, and the mixed daylight/flourescent lighting gives interesting colour. Apart from that and the typical compositional 'escape!?' conceit (up the stairs and, boom!, you're thrown out the picture top right, only to be dragged back in through one of the closed doors dammit), I've yet to discover what these images are for or about. Maybe its just that.

Throwing meaning aside, technically, this one was a bit of a disaster. I took three scans but the dicomed crashed at the very end of the third - it often sort of crashes giving a 'serial communication error' box, but retains the image anyway, but on this occasion, confirming to tradition, its lost the data. The image is still the composite, or averaging, of two images, the idea being that you get better information in combining the two images. Within any photograph there is information and there is noise. Noise is principally increased when you increase the ISO, and inversely with the tonal value - there's more noise in the shadows. (This is a much bigger problem with digital cameras than film, but present nonetheless.) However, noise is transient - noise will vary from image to image, and sometimes usefully, if you're taking pictures of the same thing. So, with an utterley static image, if you take the image twice and combine the two images you get an increase of 1.4 of the information to noise ratio and with four images you get double the information.

Unfortunately, I think that I must need at least 4 images to get the image that I would like. I'm not talking about the information to noise ratio particularly, that is just a bonus for me, but the banding problems that are happening because I'm shooting in flourescent light. Flourescent lights vary in their brightness and in their colour temperature - and they are particularly bad in the blues. Consequently, I'm getting massive colour variation in horizontal strips across the image, which get better - or more realistically less worse - with each successive scan that you add into the mix.

Finally, here's a concise and clear tutorial on noise reduction through averaging in photoshop.