Noise On Scanned Negs

vidrazor

TPF Noob!
Joined
Aug 22, 2009
Messages
16
Reaction score
0
Location
North Bergen NJ
Can others edit my Photos
Photos NOT OK to edit
Hi folks, hopefully I'm posting this in the right section here.

For some reason I get some pretty nasty noise, specifically in the shadow areas, when scanning negs, and I'm at a loss as to why. Looking at the negs (old Kodak CL 200 5093), you CAN see grain, but the film is just not THAT grainy, and the old prints made from it look similarly grainy to the negs. Here's the neg, print and scan sample so you can see:
neg.jpg

print.jpg

scan.jpg


I didn't have an effective way to shoot the print, so you'll see some room reflections on it, although you can clearly see the print grain. The neg is a photo-macro shot.

This scan sample here was done with a Minolta dImage Scan Elite 5400 using it's native software, although I also scanned this image with a Polaroid Sprintscan 4000 with it's native software, as well as both scanners using Vuescan. I tried with and without multiscan modes as well as hi-bit modes. I always got the same result.

Looking at my scan, you'd swear I used sharpening on it, but no sharpening was used at all, either in scan or post. It's strange, I can see getting noise like this from a dense positive, but being as the sensor is getting plenty of light through even the highlight areas of the neg, and the Minolta's sensors have fairly decent dynamic range, I'm at a loss as to why I'm getting such nasty noise from these scanners.

Even stranger is that back in the mid-90's I used to scan negs with a Polaroid Sprintscan 35 scanner and I would get awesome results from that old bugger, and the dynamic range and scanning res on that was significantly lower than either the 4000 or the Minolta!

So if anyone has any clues why I get this, please let me know. I'd love eliminate this from my old scans. Thanks for your time.
 
What format? The Minolta dImage Scan Elite 5400 is a 35mm scanner. The image was scanned at the scanner's full 5400 dpi resolution. It doesn't matter much if it's scanned at a lower resolution, unless you scan at something like screen res, it still looks the same.
 
Where and when did you aquire the Minolta Scanner?


I almost have to say that the device needs to be overhauled, the censor is not working properly, My Canonscan FS 2710 provides some noise but nearly as much despite my heavy use over the last couple years and several years of use prior to my aquisition of the device. I simply can not believe Minolta would allow such poor quality out their doors new from the factory.
 
Everthing I am reading about the Minolta dImage Scan Elite 5400 sais the scanner is spectacular....I'm even more inclined to believe yours is suffering from ware and tare or is a QC missfail....
 
...This scan sample here was done with a Minolta dImage Scan Elite 5400 using it's native software, although I also scanned this image with a Polaroid Sprintscan 4000 with it's native software, as well as both scanners using Vuescan. I tried with and without multiscan modes as well as hi-bit modes. I always got the same result....

Without seeing a full image to judge the degree of cropping its somewhat of a guess, but it looks like film grain, or more correctly the dye clouds that exist where the silver grain was, and not electronic noise. Also, if your statement that two different scanners produce the same results is correct, then that rules out noise. If it was noise, the two scanners would yield different results.

The image looks significantly underexposed, judging from the negative image and from the print image. There is too much blank shadow detail and too little density difference between the thinest shadows detail and the lips and cheeks for the neg to be anywhere near properly exposed. Color negative film gets quite noticably grainier when underexposed. Also, the expanded tonal curve used to produce the "normal" looking image from the scan will also increasee the visibility of the gain.
 
I bought the dImage new several years ago, and I've always had this from negs, and as I said, not just the dImage.

The neg was not so much "significantly underexposed" as it was improperly lit for a portrait, the light source being a strong hard sidelight (from my early experiences in studio photography back in the early 80s ;))

It does not look like grain to me, it appears as noise. As a matter of fact, these Photobucket re-compressed samples you're looking at here are a bit SOFTER than the real images. You can see the actual grain in the print shadows (and in the neg, for that matter). Here's a 200% image screen grab:
red.jpg


If you look carefully you can see the real grain versus the noise, which is mostly in the red channel.

In comparison, here's a sample of a positive scanned at the same resolution:

truetv.jpg


As you can see, even in the dark shadows the detail is that of natural film grain (as you see in the print and neg samples above), not noise. I find it odd that a neg, where the "shadow" detail allows the sensor the luxury of getting proper illumination would cause the transform to look so noisy. The old SprintScan 35 I used to use in the 90's would do this with dense POSITIVES, but NOT negs. Go figure.

As far as I can deduct it's something in the digital negative-to-positive look-up-tables that's causing this. So I'm at a loss as to how to avoid it. My only recourse is additional post using tools like Kodak Digital GEM:
Kodak's Austin Development Center - Photoshop Plugins & More!
Even with this I have to sectionalized the shadows because the process creates additional problems, as you can see in this example:

junk.jpg


So I'm damned if I do and damned if I don't.:mrgreen: Thanks for your replies.
 
Based on the near lack of tonal separation in the shadows and the amount of expansion shown in the scan positive image, the negative was definitely significantly underexposed. If the negative is the way it is because of intentional harsh lighting then the scan postive is a very poor reproduction of the original intent because is shows a definite attempt to expand the shadows to reveal detail.

The enlarged shots posted do reveal electronic noise. My guess is that the scanning software is failing to auto adjust, or you are not manually adjusting, for the limited tonal range you wish to expand across the final image. This results in the desired portion of the originals tonal range to exist in only a limited portion of the digital scan's range. When you expand this limited portion the noise is made more evident. Either that or the original's lack of shadow detail is too severe to expand without revealing noise.
 
Yeah, well, it was bad lighting technique on my part. :mrgreen: However, tonally the print looks more or less the same, and you CAN see the film grain, but even close to midtone data there's a surprising amount of noise in the scan, I don't get it with positives.

So I don't understand what the scanning software, whether it's the Minolta, Polaroid, Vuescan, et al, is doing to the data it receives off the sensor. You may be right that the software may be trying to expand limited data, but I'm not sure how to stop it from doing so if that is indeed the case. I would've thought that multipass scanning would lessen this effect, but it doesn't.

I'll have to try some test with modern-day neg film at various exposure settings to try and better understand what the software is doing. These particular images are merely historical for me, so it's no great loss that they're coming out like this, but how these programs are treating neg data versus positive data is a bigger concern for me.

Perhaps I need to seek out an old Polaroid Sprinscan 35. :D
 
What format? The Minolta dImage Scan Elite 5400 is a 35mm scanner. The image was scanned at the scanner's full 5400 dpi resolution. It doesn't matter much if it's scanned at a lower resolution, unless you scan at something like screen res, it still looks the same.

I know it's a 35mm scanner.. you generally still have to pick an image format for the scanned negatives to save as on your computer; .jpg, .tiff, .png, etc..
 
Oh, what FILE format? Uncompressed tiff, either 8- or 16-bit.
 

Most reactions

Back
Top