Originally posted by GilesGuthrie
This is all dependent upon the ratio of original image size : final image size being 1:1. If you are changing the size, you need to change the resolution accordingly. Also, you're almost, but not quite there on the general philosophy.
Your scan resolution is dependent upon your output device resolution, not your input device resolution. This is where you get your figures from, but they're not as black and white as you make out. For example, if you're printing to a 2400dpi printer, you absolutely don't want to scan at 150dpi.
So, your scan resolution should be:
Output device resolution x size alteration.
Also, if you are genuinely scanning for high-quality photo archives, you should scan at the maximum resolution possible, since you don't know the final destination of the image, or indeed how much of your image will be used. If the user wishes to extract a small detail from your image, it will be no use if you have scanned at a lower resolution, because once they have picked their part and enlarged it, your resolution will have fallen dramatically.
I scan photo negatives at 2900dpi, which is equivalent to 12MP on a 35mm frame. This way I can print A4-size at over 600dpi without an issue. Each compressed TIFF file is 35MB.
I'd be interested to hear your views on image file formats - their strengths, weaknesses, and suitable uses.