Is there something wrong with my Nikon Coolscan 5000?

[chris.hulley]

I bought this secondhand, and while it works fine it seems the negatives I'm scanning result in a very noisy jpg. The film I'm scanning is a Fuji, marked G-100 E09. The negs are about 25 years old I suppose. I'm scanning with Silverfast 8.8, and using Multiple Exposure, IR dust removal, SRDX, scanning at 4000dpi. The best of the profiles I've found is Fuji Superia 100. This is the result I'm getting:

 

http://www.jera.com.au/Silverfast.jpg

 

I've tried it a number of different ways. I've also got a cheap and nasty scanner, a Glanz (14mp) which produces

 

http://www.jera.com.au/Cheap.jpg

 

Although not as pleasant a scan, there seems to be far less noise evidence?

 

I've tried it with the Nikon software and a 16x multisample, which gives a better result as far as noise is concerned:

 

http://www.jera.com.au/Nikon16x.jpg

 

The reason I wonder if there's something wrong with the scanner is, various opinions that scanner noise is reasonably low in the Nikon, but if multisampling fixes it, it is probably scanner noise rather than grain, which wouldn't be affected.

 

Another way I tried was simply snapping the negative handheld with a Canon macro lens and inverting in photoshop:

 

http://www.jera.com.au/Photo.jpg

 

And finally, a 600dpi scan of a print done from the neg

 

http://www.jera.com.au/Print.jpg

 

Any thoughts appreciated. I've seen comments that dirty optics can be a problem, but that it usually manifests as lens flare which doesn't seem to be a problem.

 

Thanks ChrisH

[rodeo_joe1]

I think what you're seeing is just film grain, or grain "aliasing". Grain aliasing exaggerates the grain of the film due to the digitisation process. It seems to occur when the average dye cloud spacing in the film is close to that of the pixel spacing used in the scanner.

 

Some films, usually 400 ISO or above and always negative type, show it much more than others. The only way round it is not to use those particular films.

 

The reason the low resolution scans don't show it is that their sensors or lenses simply can't resolve the film "grain".

 

You can lessen the effect by using a low radius blur on the scans. The image shown appears not to be very sharp and to contain no fine subject detail anyway.

Edited May 23, 2017 by rodeo_joe|1

[chris.hulley]

I think what you're seeing is just film grain, or grain "aliasing". Grain aliasing exaggerates the grain of the film due to the digitisation process. It seems to occur when the average dye cloud spacing in the film is close to that of the pixel spacing used in the scanner.

 

Some films, usually 400 ISO or above and always negative type, show it much more than others. The only way round it is not to use those particular films.

 

The reason the low resolution scans don't show it is that their sensors or lenses simply can't resolve the film "grain".

 

You can lessen the effect by using a low radius blur on the scans. The image shown appears not to be very sharp and to contain no fine subject detail anyway.

 

Hmmm well, I'm using ISO 100 film. If it's grain, why does it improve with 16x multisample? Also, the noisy effect doesn't seem to go away at 2000px, which is much less than the 'cheap' scanner

[rodeo_joe1]

"Hmmm well, I'm using ISO 100 film. If it's grain, why does it improve with 16x multisample?"

 

The "noise" looks consistent with film grain to me, and does not significantly change in the oversampled scan. The oversample is darker and has a yellow/green cast that masks the effect a little. Depending on how the oversampling is done, there may be a sub-pixel shift in negative position between samples. That could also reduce the appearance of grain.

 

 

"Also, the noisy effect doesn't seem to go away at 2000px, which is much less than the 'cheap' scanner"

 

Changing the scan "resolution" doesn't alter the sharpness of its lens or physical distance between sensors in the Coolscan. Undersampling just skips every other pixel, or bins a cluster of pixels together.

 

Using a lower quality scanner is almost the same as applying a blur filter. The lens and sensor just can't "see" the film grain.

 

The other thing that makes me think it isn't scanner noise is that the speckling is visible across the entire tonal range. Sensor or electrical noise is signal level dependent, and varies from highlight to shadow of the scan.

 

I suggest you scan a good quality slide (positive transparency) and look at the deepest shadows, which will show any true noise differently from the midtones or highlights.

 

I believe you can manually override focus in the Coolscan. If you defocus the scan and the noise goes away, then it wasn't noise to begin with.

 

Grain issues, noise and grain aliasing were hot topics of discussion 20 or so years ago in scanning threads and fora. That was back in the heyday of scanning.

Edited May 23, 2017 by rodeo_joe|1

[rodeo_joe1]

I've just noticed. No need to do the defocus test: There's a small area on the very lefthand edge, near the brown path and rocks, where the scan is slightly out-of-focus. The speckling disappears completely in that region, showing fairly conclusively that it's grain and not noise.

 

That's probably due to a slight bulge or depression in the negative.

[danny_weber]

About ten years ago I used a CS 5000 to scan about 60,000 slides and several hundred strips of negatives. Although it was probably the best scanner available, it has a reputation for accenting grain because of its hard light source. I found this to be true, so I scanned to TIFFS to allow post processing.

 

Danny W.

[Wouter Willemse]

Maybe another variable you might want to get out of the comparison is the software used. If you use a film profile in Silverfast, for sure this profile will affect colour, sharpening and such. Scans made with other scanner with other software will apply different values, or none at all.

Fairest comparison might be using VueScan, as that has the same level of options available as Silverfast and will work with loads of scanners (while silverfast is usually licensed per scanner). I ran a comparison between a CoolScan 5 and my own simple filmscanner recently, and found that the above remarks on AF playing a role, and the CoolScan accenting details and hence grain both ring true, but avoid using sharpening early in the process, and the issues (and difference between the 2 scanners I tried) are pretty small and easy to manage.

Frankly looking at the first two images you linked to: the Nikon one looks ever so slightly oversharpened, the second one (Glanz scanner) looks like it could take a good deal of sharpening though it might be too soft to start with. The one done with the Nikon software at 16x sampling looks right in terms of details/sharpening to me.

 

Of course, you'll always balance between grain and extracting details. It's inevitable that trying to get the most details out of your negative means rendering grain more visible.

[Ed_Ingold]

Multiple scans are intended to reduce noise in dense areas, mainly for high-contrast reversal film like Velvia or Kodachrome - Thermal noise as opposed to grain "noise". You can sometimes reduce grain through a higher setting for ICE, but at the expense of overall softness. The fact that you see gain in mid-tone regions like faces attests to the high resolution of the LS-5000. The medium format version, LS-9000, seems to have softer light, and shows less grain. It's there, but not as obvious.

 

You have limited flexibility with focus in a Nikon scanner. You can turn it off completely, or reset the focus area. I set mine about halfway between the center and the corner, to accommodate the normal curvature in a non-glass mount. For unmounted film, I use a glass holder in a medium format scanner. an LS-8000.

 

Our taste is somewhat jaded by nearly two decades of digital photography, which has much less noise (grain) than 35 mm film once the resolution exceeds about 12 MP (or less). We would like film to resemble how we remember it, not as it actually is. You can capture some of that feeling with medium format film, but at the cost of about $20/roll with processing only. I recently inherited a Kodak slide projector, which I donated to a worthy organization, "Good Will", with some sadness, who will give it a new home. I have thousands of slides which are better scanned and viewed or shared digitally.

Edited May 23, 2017 by Ed_Ingold

[Didier Lamy]

Maybe the scanner mirror need some cleaning, this is an important factor for good results. Get is checked by a professionnal only. In Europe Nikon Dusseldorf is quite helpful for this operation.

I have been using a 5000ED with Nikonscan since it went on the market. I always kept processing by the scanner at the minimum, i.e. 4000ppi, 1 pass, autofocus at the beginning of the scan, autoexposure, no sharpening etc.. (and 16 bits B&W). And I did all the post processing later with a photo sofware. I am quite happhy with that, although sometimes clouds with Tmax400 end up with too much grain.

[chris.hulley]

Thanks one and all for all your helpful comments. Rodeo_joe, here's a scan of the IT8 calibration slide

 

http://www.jera.com.au/IT8005.jpg - 'basically, everything off', done on the Nikonsoftware

 

and one done at 16x http://www.jera.com.au/IT8006.jpg

 

Basically the difference I see here is that there is less grain, but also it doesn't vary much with the oversampling. Of course, negatives are inherently darker than positives, due to the orange colour cast. I guess I'm wondering if there could be an effect from just simply being a crappy amateur photographer, and for darker, compensated photos it's going to bring the grain out, of whether there's something inherently wrong with the scanner, e.g. it's not getting enough light to the neg for darker photos, so the noise becomes more prevalent, and I guess what most here seem to be saying is, not, this is about what you can expect from the scanner, if you can't take a better shot.

 

Certainly, some shots seem to come out better, e.g. http://www.jera.com.au/Europe213.jpg

 

Cheers ChrisH

[Ed_Ingold]

Dust on a scanner mirror is one cause of flare. Dust on the sensor may cause broad longitudinal streaks. Dust would not exaggerate grain, if anything, make grain less distinct. Over or underexposure in the original definitely exaggerates grain.

[chris.hulley]

Dust on a scanner mirror is one cause of flare. Dust on the sensor may cause broad longitudinal streaks. Dust would not exaggerate grain, if anything, make grain less distinct. Over or underexposure in the original definitely exaggerates grain.

So, crappy photography it is, then lol .... Sad thing about all this is, when I take the Nikon scan and apply all the filters necessary to get the noise level down to the point where I am ok about saving it into my 'memories' folder, I've lost so much detail it's not very different from the 'cheap' scanner result, or indeed the scan-of-the-print.

 

Of course, if I have a good shot it can do better. Ho hum.

[rodeo_joe1]

"Of course, negatives are inherently darker than positives,"

- Not if they're properly processed they're not.

 

The maximum density you should get on a colour negative, including the mask, is about 2.2D. Whereas an Ektachrome transparency might reach 3.3D. Velvia apparently has a Dmax of 3.6 officially. That's too opaque for my densitometer to get a reliable reading from!

 

So negatives in their densest regions are actually about 3.5 stops, or over 10 times lighter than an average slide film.

 

The issue with negatives is actually their lack of contrast, which needs considerable boosting by the scanning software, as well as the removal of the mask. The contrast boost is what tends to exaggerate grain, and can leave the colours from negatives a bit posterised.

 

Anyway, Chris, I don't see anything wrong with the scanner here. The IT8 isn't the best dupe I've seen (a cheap and fuzzy knockoff of the Kodak version) but it shows there's not a lot wrong with the scanner. There's no noise visible in the highest densities, the shadow tones are well-separated, and the surrounds of the lighter areas aren't flared or bloomed. All in all it's one of the better scans I've seen.

 

BTW, oversampling's main use is to give better separation of tones in the densest areas of a transparency. Not to reduce grain.

 

I really don't see any need for oversampling based on the scans you've shown. There's no shadow noise, and the tonal separation is already good without it. I'd save myself some scanning time and leave it off unless you have an underexposed transparency to deal with.

Edited May 24, 2017 by rodeo_joe|1

[chris.hulley]

Hi Rodeo_JOE, yes I'm sure you're right about the greater dynamic range of slides. However, there is another factor at work: on a slide, the darkest bits are where there is the most emulsion. Less light will get through, and any grain will tend to get 'lost' in the darkness. If you boost the light to bring out the shadows, you'll see the grain for sure - but as the regions are darker, you can apply more strenuous noise suppression, at least locally, without too much of a problem. In the areas where it's lightest, there is the least amount of emulsion left, and there's no real grain in the transparent base.

 

For a negative, the darkest areas are going to get inverted to being the lightest areas, and the grain's going to come out because what's there will become light to possibly even white, and the grain will show as a variation on that. That's what I'm sort of talking about with my technically vague and inaccurate saying of 'darker'. I think!

[chris.hulley]

Thanks everyone for the tips and help. Life is weird, or maybe it's just me ( :-) ) This morning, looking through my scans (500 so far), I find the grain less objectionable than I thought. I'm going to keep on scanning, then discard the obvious ones, then look at the rest, and for critical memories I want to keep where the grain is visible, intensively rescan with various parameters to get the best result I can - probably what any of you would have advised me to do in the first place, no doubt!

[Ed_Ingold]

So, crappy photography it is, then lol .... Sad thing about all this is, when I take the Nikon scan and apply all the filters necessary to get the noise level down to the point where I am ok about saving it into my 'memories' folder, I've lost so much detail it's not very different from the 'cheap' scanner result, or indeed the scan-of-the-print.

I didn't say you did a crappy job. I'm just explaining some issues with scanning in general. Film has grain, and you're going to see it when using a Nikon scanner. As to the quality of exposure, nothing can be concluded from the scan alone. I don't see anything that would indicate a problem with the scanner, however.

 

Scanning is so time consuming, including preparation, loading and unloading the film, that it makes sense to get the best quality as possible the first time, including maximum resolution in a non-lossy TIFF format. I use ICE, but only at it's lowest setting, to avoid artifacts inherent in its use. Memory is cheap, and you can always buy more. Time is ... well there are a lot of folksy things to say about time.

[chris.hulley]

Thanks Ed_Ingold ... No, you didn't say I did a crappy job, that was me ... I think the photo's just not very good, if the exposure was better I think there wouldn't be so much grain visible, as there isn't in quite a lot of the others. Thanks for all advice; I've just got so many negs to scan I'm just batching them through. I'll probably discard 7 out of 10 of them (especially the diving ones, there one shot worth keeping on a roll was doing well in the old days), and I just can't tell enough from the thumbnails to decide, and the previews take so long. Of course, what I wanted was a scanner that would do about the same job as a commercial film printer, i.e. produce a reasonable result with minimum effort from negs that are under or over exposed!

[Ed_Ingold]

Scans from commercial mini-labs are nothing special. They have low resolution and grossly hyped color and contrast. I prefer to start with an image less exaggerated and work from there. If you think about it, a wide dynamic range in an image will result in a understated appearance when viewed or printed. That's also why I scan to 16 bit TIFF files, which have more room for adjustment.

 

You can produce something closer to mini-lab results by treating those images to AUTO COLOR, optimizing color by channel, in Lightroom or Photoshop - similar but actually far better. An inexpensive dye-sub printer (e.g., Canon) produces prints as good or better than a mini-lab, with high gloss, and more snappy than any inkjet I've used (but with far less flexibility and control). Mini-labs can do a very good job printing images you've scanned and adjusted.

 

I have found that negative film, either color or B&W, gives better results if I overexpose between 1/2 and a full stop. I think we still suffer the effects of the ISO wars of the mid-60's (aka ASA Wars), which saw the doubling of most ratings without any evidence of a change in the emulsion. I never owned a densitometer, but I'm sure there is data tu substantiate this observation. I also find that incident readings are more reliable than reflected exposure readings for closeups, portraits and group photos when using film.

 

Which are keepers? I think it's too hard to tell from contact sheets or thumbnails. You can use thumbnails in Nikonscan (and Silverfast) to set cropping and exposure, then scan the entire strip while you do something else. That also makes it easier to log the frame numbers against the scan for cataloguing. I'm fortunate to have a 40 exposure roll attachment, but all of my negatives have since been cut into 6 exposure strips for storage, and remain unscanned. In my "golden age" of film, I scanned new rolls before cutting.

 

Another thing I have learned is that in music and photography, your first impressions are not always the best. Scan everything and decide later.