Resolved VueScan settings: noisy negatives

[MrCheeto]

Scanning with my Nikon 4000 ED. My images resulting from VueScan just have terrible noise. Basically, the sky or dark spots of these images looks like somebody sprinkled the dust from the bottom of a Froot-Loops box all over it.

Ideally, I'd like to find a setting that gives me low-noise high-DPI scans with enough range that my developing software can do the rest of the work manually. Would that be RAW? What bit-depth? Do I use the multi-exposure scan setting and combine the two images?

Aperture doesn't have a means to invert negatives, so RAW negs adds a major step since I'm trying to limit my process to within Leopard. Yes, I am aware of and use Negative Lab Pro but the Creative Cloud crap makes me nauseous. It is an absolute LAST resort.

If you have experience with VueScan, what is your process for the best scans?

This is NEW, non-expired Kodak Gold 200 shot and developed at box speed this year. Excuse the actual content, I accept my focus isn't great in these two, but there is a lot of sky to judge how accurate the scans came out.

 

[mollyc]

I do use NLP with Lightroom, but I scan with my camera.

I've never used VueScan. So I can't really help here, unfortunately, but wanted to respond because I love converting negatives.

 

[OldMacs4Me]

I think you are confusing film grain with digital noise.

It appears to me that the full frame of the sample you are showing is enlarged to about 50x75 inches. Not sure about the current version of Gold 200 but I would certainly expect it to start showing grain at any size bigger than 12x18 inches and perhaps even as small as 10x15 inches. Solid greys and blues will always show grain at smaller enlargements than portions with more detail.

 

[MrCheeto]

I don't understand your size calculations but what I'm noticing is that well before grain is visible there is a sort of digital chromatic noise. It's just random hot pixels of cyan, red, etc. This is especially evident in the underside of the B-24 which is dark but not clipped by any means. The plane is olive drab but judging by just the image one would think it's dusted with glitter. This reminds me of shooting digital photos in the dark with a CMOS, where the blackness of night just becomes a rainbow of glitter dust.

If you're familiar with printing registration, the hot pixels looks "off" like a printer that needs calibration. My other color scans that were done by my local lab show no such signs of hot specs like this. The pictures they scan show normal grain pattern with similar film with both Kodak and Fuji.

Have you tried VueScan?

 

[mollyc]

I get that you don't want to go down the Adobe/LR/NLP path if you don't have to, and I respect that. However, NLP has a forum where people chat about conversion processes, and there is a subforum for VueScan, which might help you with scan settings, even if you do a conversion outside of NLP.

Vuescan

For any topic related to using Vuescan to scan negative film – specifically for using in conjunction with Negative Lab Pro conversions.

forums.negativelabpro.com

I agree that what you are seeing is digital noise introduced by scanning, but since my workflow is different, I can't help you troubleshoot; however maybe there is something in the NLP forum that will help you.

 

[bunnspecial]

Looks not unlike grain to me, but I'd also be interested in seeing your negatives(just a photo of your negatives, not scans). When you scan underexposed, or "thin" negatives you can end up with a lot of noise. Usually on negative film this shows up more in the shadows than the highlights, but it's there. If you do end up dealing with a lot of less than great negatives, the Coolscan V and 9000 actually made significant improvements in this over the 4000/8000. I see enough difference between 35mm scans on thin negatives from my V vs. my 8000 that I often entertain the idea of getting a 9000(then drop it when I look at prices).

As for file formats-if I want to maximize image quality I usually scan in TIFF, although in the real world I tend to not see a lot of difference between TIFF and JPEG. If scanner RAW files exist, I don't know that I've ever seen them. Certainly I've not seen them from VueScan or from NikonScan. With scanners you have the benefit of being able to adjust it on the front end before capture, so use that and get what you want before scanning.

I don't necessarily see this in your scans, but the optics can get dirty in these Nikon scanners. The results from my III when I first got it, although it mostly showed as low contrast and haloing. It was a night and day difference after I cleaned the mirror in particular, although the lenses need it too. It's not a terrible job and doesn't require disassembly beyond removing the cover from what I remember, but does need a lot of care as the scanner is pretty well junk if you scratch the mirror.

Last thing-you have PPC Macs, don't you? I'd highly, highly recommend not messing with Vuescan and instead just use Nikon Scan on PPC(or on an Intel Mac running Snow Leopard). Unfortunately virtualization won't work with your 4000 as you can't pass through Firewire to a VM. NikonScan isn't as user friendly as VueScan, but the results are worth it especially given that you get real Digital ICE and not Vuescan's version of it. I bought my first Vuescan license in 2007 to run the Canon flatbed I was then using, and used it for a long time but when I started using the Nikon scanners and comparing side-by-side I gave up on Vuescan even though I knew my way around it well at that point.

 

[OldMacs4Me]

The enlargement factor was easy to calculate. The tail section you blew up is about 1/10th the width of the full image or about .15 inch on the original neg. The width of the tail section is 7.5 inches (on my screen), which is about a 50 times enlargement from the original. I may be off a little but it will certainly be in the 45-55 times dimensional enlargement range. Way more than an ISO 200 color neg film can reasonably expected to handle. I would expect that film to deliver 10x, and be delighted if it would cleanly go beyond 15x.

 

[MrCheeto]

I gave up on Vuescan even though I knew my way around it well at that point.

Welp that's good enough for me. I'll try to track down the old software. May not be all that easy.

I'm running this on a Mac Pro.

As long as the software spits out something high-DPI, low-noise, and nearly color-correct I can handle the rest in developing software.

 

[bunnspecial]

Welp that's good enough for me. I'll try to track down the old software. May not be all that easy.

I'm running this on a Mac Pro.

As long as the software spits out something high-DPI, low-noise, and nearly color-correct I can handle the rest in developing software.

I think I put it on the Garden at one point. If not, it certainly out there and you may still be able to find it on Nikon's website. If it's not on the Garden I'll put it there(I need to upload a bunch of old photo software there since some is getting impossible to find).

You want a 4.x version. It will control all the USB and Firewire scanners(which means the 4000, V/5000, 8000, and 9000) and is Carbon so will run in OS X or OS 9.

I've done a lot of scanning with Nikon Scan on my Mac Pro 5,1 running Snow Leopard. I have it on a PCIe SSD and it's quick to say the least. When I first got my Coolscan 8000, I was running the whole set-up on a dual 2.7 G5(fastest PPC for that software since it's single threaded) but even that gets sluggish with a full resolution scan of a 6x6, much less 6x7. I've run Nikon Scan both as stand-alone software and as a plug-in in Photoshop CS2 and CS4. I actually kind of prefer the latter as you get the image in Photoshop immediately.

EDIT: Download from Nikon. I should still drop this on the Garden in case it goes away

https://www.nikonimgsupport.com/eu/BV_article?lang=en_GB&articleNo=000042176

 

[MrCheeto]

Thank you for serving this one right up! I did my best to make it work but it simply will not load. It crashes under different circumstances but let's just say the scanner being on and the software loading cannot happen at the same time. I notice it's also PPC. For whatever reason even after uninstalling several times, deleting prefs, etc it simply doesn't work in Leopard on intel.

OK look at attached pic. One is the Nikon with VueScan, the other is from the photo lab. I just don't think I can compete with their scanner. Look at the plants. The difference in quality is like a difference in competency. I can't have underexposed film considering the number of rolls I've put through my T70 with not a single one exposed improperly. The noise is certainly there and certainly a problem. I can see the grain on their scan and it is what I expect to see. Yes I am zoomed in farther than I would ever really need to be but that counts as a loss in image quality.

There are sprinkles of hot pixels ALL OVER the image on the left whereas the one on the right is super sharp, shows normal grain, and recovers more detail in the plants which were in-fact green not reddish brown.

Is this the best I can expect from this hardware? Does anybody have better samples from a similar scanner? If anything I hope I have a dud and it can be serviced or something and I don't think it's a dirty mirror in this case.

 

[bunnspecial]

First of all, what's your full hardware set-up you're trying to run it on? I think I've hit a wall of where Nikon Scan will work or not work, and that place seems to be 2011-era Macs. I also had some issues on my Mac Pro 5,1 with dual X5690s. It would crash/not load with the scanner on, as you've described. Others, like my MBP 8,1, were not stable and would sometimes crash mid-scan. On the 5,1, though, funny enough it wouldn't run as stand alone software but would work fine as a Photoshop plug-in so that's how I did it(and actually found I prefer doing it that way).

To your scans-I've spent a few minutes looking at them at them and to my eye, the Nikon scans seem sharper and generally more detailed than the film lab scans you've shown. I'm looking at the plant to the center/right of the frame in particular since it seems to be in sharpest focus. What I DO see are blown highlights and generally an overexposed look(overexposed in the scan, not the negative you're working with) that's not present in the lab scan. That can be fixed in your scan settings. The color can be fixed also. Forgive me if I'm mentioning things you're already doing, but be sure you're selecting an appropriate film profile in Vuescan. I could have sworn there were some Kodak Gold profiles available. You can also fine-tune the colors in there. If you want to get fancy, photograph a color target and then profile the scan of it.

There are a few other considerations, too. The 35mm Nikon scanners don't really have the flatness issues that the bigger medium format ones do(that's enough to make you want to give up on even trying until you get one just right) but some holders are better than others. The absolute best is the single-frame strip that works with the slide holder, although it's inconvenient enough to use that I rarely bother. The automated strip feeder is fairly good(and super convenient) but isn't quite as good as the single frame holder. Mounted slides are a mixed bag and it depends more on the quality of the mount. If you are using the strip feeder and haven't done so, be sure you take it apart and clean the rollers to help avoid scratching.

To continue with that, though, don't forget that the scanner does need to be focused. Autofocus is generally good on the Nikons and they have a decent range(unlike the also popular Epson flatbeds, where you can spend an eternity adjusting your holders to try and get optimum focus....) but I seem to recall that Vuescan lets you fine-tune focus. If you want to go nuts on this, you can buy a resolution test target in a mounted slide but I've never gone that far. Also, dust on the lens in particular can obscure fine detail.

I can't show scans from a 4000 as I don't have a functioning one. I bought one several years ago and brought it home to find the Firewire chip was dead(if you're not aware of this issue, just know it can happen with the 4000, 8000, and 9000 and hot plugging/unplugging seems the biggest culprit). I've always meant to have it fixed since, unlike my V, it can handle the bulk film/whole roll feeder, but I never wanted to spend the money on one of those either.

With that said, I've used the very similar V quite a bit and can show samples.

Here's one I grabbed at random. This would have been c. 2009 and taken with my Canon T90 and 20mm f/2.8 on Elite Chrome 100.

And a 100% crop

Here's a slightly more recent one, although for whatever reason I seem to have scanned it at lower resolution(or maybe did it on the Coolscan III). This I have in a folder labeled "2019". Funny enough the details aren't as clear to me as above(although the occasion wouldn't have been as memorable) but knowing my normal kit then for this sort of stuff, or rather what I normally walked around with in this location(or still do when I visit for that matter) I can just about guarantee it was my Nikon F2SB and Velvia 50. I'm trying to think what the lens would have been, as my usual walk-around kit those days was a 24mm f/2.8, 50mm f/1.4, 105mm f/2.5, and 200mm f/4(I don't think I had my 35mm f/1.4 in April/May 2019 when this would have taken, otherwise I'd say that was it). Given the location and FOV, I'm going to say it was the 24mm...

One last one-this was Ektar 25 shot 2018 or 2019ish in a Nikon FM2n. Given the rather extreme perspective, I'm inclined to think it was either my 18mm f/4.5 AI-s or 18-35 f/3.5-4.5 AF-D zoom. With the barrel distortion I would lean toward the latter, plus I don't remember if I even still had the 18mm prime then!

I'm happy to do other, more recent ones although it may be next week before I can get to it as I'll have to actually get my Mac Pro set up.

 

[MrCheeto]

Thanks for the thorough reply. I immediately wanted targets for each film stock but I have read that these scanners are extremely temperature sensitive and can lose calibration in the middle of scanning a single image.

What I’m especially noticing in your images is that there is NO color aberration or what I call hot pixels. The shadows in your black steam engine are BLACK. There may be individual grains that appear but nothing is cyan or magenta. I can especially make out the grain in the stone architecture but that’s all I see. It’s grain, not noise.

In the plant I posted, I can make out a hot pixel here and there. What I notice is that this is regardless of the actual content of the image. They’re just random hot pixels not blown highlights or crushed blacks (ie thin/thick emulsion).

My hardware is a 4,1 Mac Pro with x5500 CPU and 10.5.8. I can try to run this software on my G4 and see if it’s just a software issue. The fact that the pixels are cyan and magenta makes me think that perhaps the sensor isn’t focusing the three-light passes into a single image, like the misaligned print register I posted.

I’d be absolutely satisfied if my film could come out looking like yours. Not the composition, just the scan quality hah. The plan is Fujicolor 200 and the others are all Gold 200.

Thanks.

 

[bunnspecial]

Thanks for the thorough reply. I immediately wanted targets for each film stock but I have read that these scanners are extremely temperature sensitive and can lose calibration in the middle of scanning a single image.

What I’m especially noticing in your images is that there is NO color aberration or what I call hot pixels. The shadows in your black steam engine are BLACK. There may be individual grains that appear but nothing is cyan or magenta. I can especially make out the grain in the stone architecture but that’s all I see. It’s grain, not noise.

In the plant I posted, I can make out a hot pixel here and there. What I notice is that this is regardless of the actual content of the image. They’re just random hot pixels not blown highlights or crushed blacks (ie thin/thick emulsion).

My hardware is a 4,1 Mac Pro with x5500 CPU and 10.5.8. I can try to run this software on my G4 and see if it’s just a software issue. The fact that the pixels are cyan and magenta makes me think that perhaps the sensor isn’t focusing the three-light passes into a single image, like the misaligned print register I posted.

I’d be absolutely satisfied if my film could come out looking like yours. Not the composition, just the scan quality hah. The plan is Fujicolor 200 and the others are all Gold 200.

Thanks.

As the black goes-

One thing to bear in mind that aside from the last one, all of these are transparency film. Velvia in particular is know for its blacks. Negative film is of course the opposite. A perfect black on negative film is the density of base+fog, and a blown highlight is Dmax. Slide/reversal film is the exact opposite. There's a reason why back in the day landscape shooters often preferred slide film despite the difficulties in exposing it and the limited dynamic range, and often that was because blacks are truly black. It was also one reason Kodachrome is so legendary despite what to me are kind of dull colors(aside from reds, and of course beautiful caucasian skin tones). Kodachrome blacks are so dense that you can actually see them in relief on the film if you hold it an an angle.

It certainly is possible that you have a misalignment. I think the optics of the III on up are all similar(the II is a different beast-in fact there's a version I've kind of hunted for but haven't found that can fit in a 5 1/4 drive bay). I have ventured into the optics a bit but never enough that I had to worry about knocking things out of alignment so can't comment on that. I suppose a mirror out of place could do that...

I'm not sure if this is what you're seeing or not, but one thing to keep in mind with especially fine grained film is a phenomenon called grain aliasing where the grain appears larger than it actually is. I'm not sure that's what's going on here. You can see it especially in fine grained slide, and I don't know that it's not what you're actually seeing in the Ektar 100 I posted. That particular roll of film was probably made before my wife was born, so it may well have been 30 years old or close to it when I shot and developed it. Slow films like this do tend to age fairly gracefully, but I remember it having a lot of base fog. Films do also tend to get more grainy as they age.

One last pedantic thing that I've been guilty of/sloppy about in this thread. Color film doesn't have grain, it has dye clouds. "Grain" means the metallic silver crystals that are produced in the developer. Color film goes through a "bleach" during development that removes the silver and leaves behind the dye that was "coupled" to it(Kodachrome aside, where the color is added during development). Small point and I still usually call what we see in color film grain even though it's technically not.

 

[bunnspecial]

Last thing and just kind of a peripheral thought-if you are going to shoot film it's worthwhile to at least get some practice looking at film. Even a small light box(a 4x6 one or so with LEDs) will do fine and not take up much space for looking at film. More important is a quality loupe. Back when I was first getting serious about photography in the mid-2000s as a high school/college student and staking my claim that I was going to use film, a loupe seemed an almost insurmountable obstacle. Schneider and Rodenstock ruled the roost, new they were $500+, and used didn't honestly save a ton over that. I was told I really needed a 3x or so for general viewing and a 10x for critical. These days there are decent Chinese options, and the good German ones have come way down in price second hand.

In a pinch, though, and what I did for a while, you can use camera or enlarger lenses. A 50mm lens is just about right for 35mm film, and a 150mm or so is great for critical viewing. Enlarger lenses have the benefit of being much lighter and also tend to be fairly inexpensive aside from of the more desirable ones like the 40mm Leitz Focotar(if anyone has one laying around separate from an enlarger or that they'd part with separate from it, let me know). Even the El-Nikkor lenses, which were "budget" enlarger lenses, are great as loupes(and I've made 8x10s from 35mm negs I'm super happy with using a 50mm f/2.8 EL-Nikkor).

 

[MrCheeto]

After following all of the advice from @bunnspecial I think I'm ready to mark this one solved.

I believe it has less to do with the scanner or software and more to do with the treatment of the scanner image data.

I am aware of "shooting negatives" and I even have a lens/adapter for it. I just am not a fan regardless of final results.

I believe the noise is just noise and your original assessment of the thin emulsion is correct. I notice this noise only in underexposed areas for the most part. I've dialed in the settings as far as black level and avoiding clipping and I have to say I'm really fond of the initial improvements.

The noise is certainly some kind of noise because the "hot" pixels are either green or red or cyan which leads me to believe some sort of interpolation is going on. As I'm writing this, I'm now wondering if I should try scanning at something below 4,000 dpi.

According to my results, 4,000 dpi definitely shows more grain than detail. It exceeds the limit of detail in 35mm film even with fine grain.

My scans now look much closer to the IQ of your last photos posted, which I consider acceptable. They show apparent grain but no or low noise at all.

I will have to make time to clean the mirrors and perhaps sensor/lens but, for the time being, I have cleaned the roller mechanism which immediately improved dust and scratches on the stock.

I wish I had me some of that beautiful caucasian skin. It's not just Kodachrome but most color Kodak in general. I've used Kodak my entire life (I am including disposables, of course) and having olive skin means Kodak brings out the most unflattering green in my face. I look ill in all of those pictures back then. Fuji film already has a green cast, so it's no help for me. Only if you happen to have east asian or black skin tones, go figure. I know there are films that suit me but not in those disposable cams and not vintage Kodak!

The colors in that desert comparison are more accurate using the Nikon than whatever the lab uses. Their color process just made it look like what JJ Abrams did to all of that "film" he claims was used in Star Wars. The scan I did looks exactly like the sand and sky I saw and took be right back out to the desert. That's the only time I've been pleased with Fuji color.

I learned a lot from your posts. I now know why the books of my youth had such a peculiar look when it comes to the landscape photography. Would have never figured that one out. Now I'm sure that answers a lot of questions from back then, actually.

I tried Nikon Scan on the Mac Mini and...eh. It's pretty dumbed down, clunky, and not straight forward. I'm really beginning to love VueScan and I think very soon I will decide which version to purchase.

As for flipping negatives, I think I might have better results just taking the raw negative scans and using Photoshop to invert them. I know NLP is ready to serve up any and all film negatives with presets but I'm hoping I can stay within Leopard by finding a means to do so with Photoshop CS5. Not a deal breaker but immensely gratifying.

Here I have re-scanned and added a few new scans. Let me know if you find these results acceptable, ignoring the photography itself of course. These are straight from VueScan, no PDS or editing.

 

[MrCheeto]

After really digging into the settings and seeing how it reacts in especially thin areas of film, I have decided this is completely acceptible. The thinnest transparent parts of the negatives are apparently just resolved as noise whereas grain/dye is properly and accurately resolved. I've purchased VueScan and can't recommend it enough. This software has everything I was hoping for and nothing less. It makes batch process in both scanning and processing extremely easy. I can also recommend the Nikon scanner. I'm done going back to the lab and I'm very very happy with my scans.

Of course, I had to follow all of the advice above to figure out these points.