Shooting Film - Scanning Film and Other Horrors

2009_008_11_lvl_grey.jpgHad a couple of interesting phone conversations that started with emails from some readers that are considering shooting some film. A couple of experiences that they had in the past turned them off to it so they had a couple of questions. While they were all fine and dandy with the cost argument when wanting to up their quality game the big concern was their intent to scan the film and their experience was that scanned film looked like s**t and totally undermined what quality they may have gotten by moving up in capture size/reduced magnification.

The big big big deal for them was film grain and how crappy it looked even on slow speed film. There is a ton of information out there on scanning film but I thought I would share a couple of things that surprised the folks that I talked to because they had never read/heard/thought about them before. Namely some more info on noise and grain and why it is such a big issue with digital folk. Here are a couple of things to think about that are way different when producing digital output vs. analog output.

Let's for a second talk about noise and grain as if they are exactly the same thing. For the most part grain noise is really not that high of contrast compared to overall picture contrast in most circumstances. Second grain size is relatively small even with higher speed films relative to how many pixels that you have for the scanned image. In this case the film dynamic range matches up to your scanner pretty close and grain noise will not be a huge issue, it will work/look similar to analog print output at any given size, not really but close enough not to fuss with.

The issue comes based on two factors, contrast manipulation of the scanned image and what digital interpolation or resizing algorithms do versus analog enlargement. Analog enlargement actually makes every single thing optically bigger the more you enlarge the image, conversely the less you enlarge the physically smaller it makes everything - including the grain noise. Now for the punch line - you are going to have to think about this for a minute to wrap your head around it - your first reaction is going to be "bullshit". When you scan/digitize film you are starting out with the biggest image/enlargement you are ever going to get and when you resize downward (less magnification) the grain does not get smaller, well it may if it is HUGE but if it is a pixel or so big it cannot get smaller it stays the same but it's size in relationship to the image as a whole may become gigantic as you make your image smaller and smaller, like for screen or web display. In effect this is exactly the reverse effect of analog optical printing/display.

If your original scanned image matches the output resolution for a print/print size that you want to make within reason there is no reason to worry about this even if it looks like shit on your screen resized to 75-100dpi if your output medium is more like 300+ dpi. Why? Well, because your grain is still only a pixel or few wide in the file that is very very very many pixels - optically it will be small. Resize that image to 500 pixels wide and guess what now the whole image is only a few pixels and the grain is still a pixel of a few wide. Still this is not usually a big issue if your negative/positive that was scanned matched the dynamic range of your scanner and you did not jack the contrast up. The two things that really aggravate this effect are really flat/low contrast negatives and really really giant hi-res scans of big film that has had the contrast jacked up. Why - you are going to make the noise more contrasty as well - the higher the contrast the more weight the digital interpolation when making the file SMALLER will give to the grain noise and through the low contrast pixels away. This is how resizing works. The resize has no freaking idea what is detail and what is grain and what is noise.

As I said for small film of normal contrast range with a average lab resolution film scan this is not a giant issue - the web size/display size stuff still emphasizes the grain noise more than it should compared to an optical/analog print but nothing to be really upset about and in prints of the same linear dimensions it should not be an issue at 300dpi or above. The real issue come in when you are scanning medium and large format films at very high resolution and need to resize to a much smaller file and when you are dealing with low contrast negatives that need to have the contrast boosted in a big way.

So what to do beyond having someone else deal with the scanning and file sizing, etc?

1 -One only worry about this when you are downsizing a file in a significant way if you have the right or close to right number of pixels to make the print you want at the real resolution of the printer you will never see it in the prints. When I say "you" I also mean the printer itself - I have seen bad things (grain in LF scans) this happen when sending a file that is way way too big for the print size to an ink jet printer - ie. 4000dpi scan of 8x10 negative to make an 8x10 print. Guess what - grain.

2 - My policy is to capture as much information as I can when I scan - highest resolution, etc. If I need to resize downward huge scan files I do a noise reduction on them using Nik Dfine either before the resize. The visual effect of a screen sized image is the same. I also do the same thing if I am printing significantly smaller than the optimum size for the format/scan. I do not remove the noise from the file at scan time, I do it based on output intent so that the visual effect is the same I get from analog optical prints.

3- If I need to jack the contrast up for a given negative and I know I am going to do a down-sample I do a noise reduction on the un-manipulated layer. This usually works the best but occasionally I find that noise reduction after the contrast adjustment works better. I have not narrowed it down to set of specific parameters like film type etc because I almost never have to do this.

I hope this may answer some questions for those of you with gigantic scans of gigantic negatives that print beautifully optically but look like shit when printed digitally at small sizes or computer display sizes. I do believe that this is not common knowledge as I routinely see books published with large format negatives where the image sizes are actually smaller than the negative size and I can see grain in them - same goes for web pictures I see of large format negatives. There is NO WAY for  this to happen in optical analog printing with negatives that size - no matter how big the grain.

RB

Ps. I also think this effect is part of the reason grain simulations in digital are so outrageously bad and look nothing like real film and real print sizes in the analog world. Tri-X in 35mm is virtually grainless at 5x7 and even 8x10 depending on how it was processed but digital simulations look like Tri-x pushed to 3200 and printed 16x20 even on a web sized picture that measures like 3" wide on the screen - this is a ridiculous digital distortion that happens with a less than informed work-flow and does not happen with real film and real analog prints.

blog comments powered by Disqus