Copying color negatives with camera

I won' t claim expert status in this area but I have carried out a lot of experimentation in the area and scanned a lot of colour neg material, I have also built a custom made device for carrying out the scans, including sectional scans which can be stitched together to liberate vey high levels of resolutions.  Anyhow perhaps I can suggest a few pointers that may help, without getting into the technicalities of this process, and the fact is there is a lack of easily obtained info on the subject.

1) Scanning colour neg is the most difficult of all scanning tasks, mainly due to the mask and limited contrast of the original neg which make the Raw conversion much harder to do well.

2 ) There are huge variations in final file quality due to the capability of the Raw convertor you choose, some are far better than others in dealing with the limited tonal range of the original file, some for example cause banding, weird contrast shifts, poor and unfixable colour shifts in highlights and shadows. Photoshop is actually pretty poor at this job, the best results I have achieved were with Iridient Raw Developer or Aperture on the Mac, I have tried a whole array of tools, but not the GIMP so I can't comment on that.

3) By a long long way the best results will be obtained by pre-filtering the lightsource in such a way as to get a capture that matches the native colour characteristics of the cameras sensor and accounts for the colour temperature of the lightsource and the films mask characteristics.  This is pretty tricky to do, involving quite a bit of testing but the differences in the final result are like chalk and cheese and the resulting files are far easier to edit.

4) Despite the accepted wisdom that film only needs to be scanned at say 8 megapixels becuase it is only holding that much data the truth is you really need far higher levels of resolution to extract the information in the neg without  alaising the grain.  As an axample I have regularly scanned fuji 400 in sections for a final stitched output of about 100 megapixels, at this resolution you actually do resolve the grain of the film without alaising, the alasing causes the grain to look much larger than it really is and messes up many of the subtleties of the neg.  Once fully edited the files are then downsampled to something more sensible like 16 megapixels and then print beutifully and look much more detailed than any other scanning method.  Frankly even the best commercial scanners have insufficient resolution to really do the job properly.  Lower speed films like Fuji Reala need around 200 mp or more at capture. I realize most people will think I am talking theough my hat but I promise as you go higher to these extremes the granularity of the film capture shrinks and becomes more tightly focused making for a subtly better final result.  Ultimtely with scanning there are two approaches to resolution, either scan low enough so grain is smoothed out altogether ( which will greatly soften textural details etc but is fine for web and small prints) or scan high enough to sharply render all the grain in the film, anything in between is just messy.

4) Pretty much everything in terms of colour and tone and inversion can be done via a curves inversion and tweak in the Raw Pro Processor, if you have nailed the pre- filtering the inverted image will be very close to correct in all aspects straight out of the inversion.

5) When exposing go for a histogram that is roughly centered, exposing to the right does not help with neg scanning and underexposure will cause radical increases in image noise in the highlights of the inverted image, it is totally different to how we would work with normal digital capture. Small deviations from correct exposure can and will mess up the colour in the highlights or shadows and make editing much harder, you must use manual exposure methods for the capture and by far a filtered flash is the best light source but I have had great results with high powered and filtered LEDs.

6) If you go down the high res sectional scanning route you files will need virtually no sharpening and indeed doing so will mess them up, the only sharpening you do should be very very subtle and ideally of the high pass variety, sharpening is generally best done on the final oversampled file.

There is way more I could say, but hopefully this gives those interested some food for thought and potential pathways to follow.

cheers

Brad Nichol