Wednesday 1 October 2008

colour negative - wondering about getting better images

I once mainly used colour slide as my prefered film but I now commonly use colour negative as my film because I found that even after 12 years of development in scanners when scanning slides I don't get into the shadows very well. In contrast (say, was that a pun?) negative can cope with much greater holding onto highlights than slide can, and so you can tune your exposure to bias exposure to the shadows and expect the highlights to hold. See this example elsewhere on my blog.

Of course Colour Negative works rather differently and instead of getting a "ooohhh" when you see that on the light box you get something like this:


not startling is it ... yet from this surprisingly terrible image, you can pull out the scene below:


Now interestingly, I took a digital of this scene for working on a page I'm writing on exposure of film with a digital camera here, you might find that interesting reading too.


Immediately the colours are more punchy in the digital, but then again we've blown out the clouds to blinkies and there isn't much in the shadows over on the left there (unlike the film).

now, lets look at that area in more detail:


surprisingly good colour (even in the reeds in the shadows). I bet I can't do that with Slides either. Now going back to that horrible looking negative I had up there, lets look at what's in the Red Green and Blue channels (remember we're still in Negative here, so black is white in the finished product).




Despite the RGB looking like a broad histogram, there is heaps in the Red (gosh just look at the bloody thing above) , little in the Green and next to nothing in the Blue.

No wonder the sky is washed out and noisy. So, I'm now wondering if this can be fixed with a scanner which punches more into the dense areas? If so by how much? Perhaps its jut not there at all?? Perhaps the blue channel in the resulting scans of negative are always going to be weak?





I've just had a conversation with a fellow who drives a drum scanner. He says that his device gets into these areas more and has much cleaner results than what he's seen from consumer scanners (gosh why should that be surprising?)

I think that this shows when we look at what happens to the blue channel when we make the level adjustments required to set its black and white points (streching that small data range to cover 0 to 255 effectively).



Notice the numbers and positions of the sliders. In the on the left its as it came from the raw linear scan, and on the right its stretched steps 2 to 35 to cover 0 to 255. It is clear that heaps of noise results. I don't know if this is quantisation noise or CCD noise. But I expect its a bit of both.

The reds in contrast performed much better (well it started out better to begin with)


There was data from 44 to 139 which meant a more modest strech and not to forget that the lower end of the data was further away from the dense areas of the negative and within a more linear and reliable portion of the scanners ability.


So, now I think I understand why people think negs are harder to scan than slides. I wonder if I can get my hands on a linear scan of this neg done with a Nikon?

3 comments:

Dan Hutt said...

Obakesan (your Japanese name?)
This discussion on scanning negatives was very informative and makes me realize that I have to learn a lot more before I start scanning a couple thousand negatives I have. I only want to do it once!
I have a Canon Lide90 flatbed scanner with 4800 dpi. I'm not sure this hardware is good enough to start with. I'll keep reading...
- Dan

obakesan said...

Dan

scanning a couple of thousand is quite a task! I know it seems daunting but apart from the tricky ones the software does a good job. If you don't mind the suggestion I strongly recommend that (if doing mainly 35mm) that you get something like a Nikon LS-IV ED as a minimum (LS-4000, LS-V or LS-5000 is all the better). I sincerely doubt that the optical performance of the LIDE is anywhere near as good as 4800.

Once you've scanned and got images to within ball park I'm sure that you can save space and not worry much if you convert to 8 bit / channel. Definately scan and initially adjust in 16 bit (if you're adjusting.)

Best of luck :-)

Jan said...

Hello Chris,
I have now tried various things for scanning negatives (using a Coolscan LS-40 with Nikon Scan 4.0.2), and the instructions you give on http://www2.gol.com/users/cjeastwd/coolscan.html lead to the best results for me.
One thing I'm left wondering about is: do you set the Gamma to 1 in Nikon Scan, or do you leave it at the default 2.2? And since you're using "Calibrated RGB": which colour space? BruceRGB? I have found "Scanner RGB" quite useful too.
Thanks, and greetings from Newtown (NSW),
Jan