Build a transparency/negative adaptor
I got this idea from the scanning experiments on Don Maxwell's Web site. The tent-like device described there was only partially successful for me, producing dim slides and unusable negatives. Negs require more intense illumination. I didn't have anything suitable to use as a backlight so I put my mind to improving the tent idea.
After some false trails I came up with a version that pumps a huge amount of light through the transparency. This is a genuine result I obtained from a 35mm negative:
It has been scaled down to fit the page and could benefit from minor tweaking of contrast, brightness etc. Not all negatives will respond so well, and I suspect that some scanners will not work effectively with a device like this.
Dedicated transparency scanners use back-lights. If your scanner is built conventionally, that is to say with a bright light tube and CCD (charge coupled device) sensors, as mine is, then this is worth experimenting with. If it is one of the newer breed of low-cost scanners that use LEDs as a light source and CIS (contact image sensors) then I can't say how well this will work. Feedback on this point will be welcome!
To business: you need two 150mm (six inch) mirror tiles, a few strips of wood and some glue. Make a tent from the mirrors held at 90 degrees to each other by some end pieces. Make a cradle to sit the mirrors on. The result looks something like this:
Before starting work you need to make sure your scanner's bed is wide enough to accommodate the cradle sitting directly on the glass. The apex (ridge) of the tent points top to bottom of the scanner bed. The transparency or negative goes parallel with the apex and offset to one side.
The finishing touch is a pair of short legs at the end of the cradle nearest the scan bar's resting position. The length of the legs is determined by the way your scanner is made. On mine they are 2–3mm tall.
The purpose of the legs is to tilt the mirror assembly. Without them, a light beam from the tube would hit a mirror, be reflected to the other mirror, and from there be reflected back down to the lamp but a bit further along from where it started. The light therefore misses the sensors.
The right degree of tilt deflects the beam slightly, directly into the sensor array. To find the right angle, perform experimental scans with a paper pad at the 'north' end of the cradle. Dark scans indicate the height is way off. Serious colour fringing means you are getting close or the cradle is not parallel with the sides of the scanner. I can't vouch for the geometry of other scanners so you may even need the legs at the 'south' end... suck it and see.
Small adjustments should enable you to get a result like this:
Both scanner glass and negative must be very clean and dust-free, as dirt will loom large. Scan at the highest resolution possible. Crop into the required area. Rotate if necessary. Save the file and it is ready for processing.
The next page tells you how, but before moving on I should say that this is just a bit of fun. It is not intended for professional use, though you can get good results that look better than scanned prints. But because the scanner light is being reflected from the underside of the negative as well as passing through it, output will always be inferior to a good back-lit device where the scanner tube can be turned off.