Here are a few bullet points:
- As noted above, grain is supposed to be there. How much grain and how it appears depends on the film you use. Some films are quite grainy, others not so much.
- Electronic noise, while possible, is usually not a prominent feature of negative scans
- Machine scans from the minilab are seldom very good. See below for details.
- The scan resolution (number of megapickles) is not an indication of scan quality. I regularly scan at 1200 dpi and get better results than the OP shared above.
- As with optical enlargements done in the darkroom, the single greatest determinant of reproduction fidelity is the quality of the optical path. This includes focus, lens(es) and film flatness.
- The second greatest determinant is initial image processing.
- The third greatest determinant is the quality of the original negative (huh?)
Way back in about 2009, I started shooting film again and soon became discouraged with the quality of the scan results I was getting from Costco. Despite being 8 megapickle images, they often appeared grainy, had poor contrast and just looked yucky. Why should this be?
In theory, the minilab results should be pretty good. Rather than being a scanner
per se, they use a detector that is essentially similar to that in a digital camera. There is a lens, a light source, a detector and an image processor. There are several PF members who use their dSLRs in similar manner with good results and also several PF members that actually own Pakon (Kodak) F-135 commercial scanners with which they can scan a full roll very quickly and with which are very happy. Clearly something is amiss with either the minilab machine or how it is operated.
Back to my personal history. I did some careful evaluation of my negatives and also tried several different local drug, variety, and big box stores having minilabs (yes, I do have a choice...luck me). Here is what I found:
- There was evidence of poor (inadequate) development at my local Costco. I lay that to inadequate attention to developer replenishment and machine maintenance. Target and Walgreens did a much better job with the negs.
- All of the minilab scans suffered from digital processing artifact (more below)
- Few of the people running the machines were capable of intelligently overriding the default digital processing
The way it was explained to me is that the minilab scanners are optimized to work with average quality negatives (e.g. crummy) to produce nice looking prints from the minilab printer. This means that all images have automated curve adjustment, additional sharpening, additional smoothing (grain reduction), and saturation boost applied. For those labs where the tech was savvy enough to turn that stuff off, the scans were actually pretty good.
My solution went like this:
- I bought a decent dedicated 35mm film scanner
- I started using a local pro lab for all my processing. The price is about the same as the drug store, but the quality much better. I can also request my negatives back uncut and unsleeved.
- I learned the fine points of scanning film
My first scans were a revelation. The images were beautiful! This was even true with the scanner running with auto exposure and auto contrast. With time, I got even better results as I learned the limitations of the scanning process and the data that can reasonable extracted from negative. I won't spend any time suggesting comparison to either my K10D or K-3, except to say that resolution is not my first consideration when choosing 35mm film vs. the K-3.
So...Why do the minilab scans suck? My conclusion is that the sharpening, smoothing, and saturation adjustment is applied with inadequate finess. I have found that things work best if the initial scan is made with
no sharpening, fairly flat contrast, and no saturation boost, with any adjustments applied in post-processing (Lightroom). It is in PP that the difference between a scan and straight digital capture become evident. Scans are much more "brittle" in PP, meaning that artifact appears much more readily. This is probably because of how the image is structured* and is particularly true of sharpening. If one attempts to smooth grain using noise reduction the results get truly yucky. Curve adjustment generally works pretty well. Saturation adjustment requires a gentle hand.
So, what are the practical solutions for today's film shooter wanting a hybrid workflow?
- Buy a high quality scanner ($$ - $$$$) and scan your own --or--
- Use minilab scans for proofs and a custom scanning service for the better shots --or--
- Buy a lower quality scanner for proofs and use a service for the better shots
FWIW, the third option is not a bad one.
I hope all this helps.
Steve
* Image structure on a scan is a difficult thing. With color, what we see on the scan is an interpretation of dye clouds in the various layers of the negative. The size and uniformity of those dye clouds affects the "graininess" of the scan. It is not so simple as "photographing" the dye cloud or grain clumps. Both are beyond the scanner resolution.