Noritsu scanner - showing rebate question

[calico]

Does anyone work with a Noritsu scanner at a lab? Have a question about including rebate in scan of 6x6 film.

I have scans made at a lab when I don't have time to scan myself. I have been using a certain high quality lab for a few years now. I pay extra for them to include rebate in the scans.

The rebate is never the same width on all sides, and sometimes only three sides of a frame will show the rebate. Usually, there is at least some rebate on all sides. I have accepted that it isn't even, as they have told me they can't really control exactly what shows of the rebate.

But I got a roll back recently (Portra 160 in 6x6) where only two sides of rebate showed (quite wide), with no rebate at all on the other two sides. Scans from the whole roll were like this.

In the same order, they scanned a roll of Provia 100F in 6x6 where the rebate showed on four sides, though uneven, as usual.

I asked them to rescan the Portra 160 to deal with the rebate issue.

I had always assumed the reason the rebate was uneven was a physical issue -- how the film was positioned when it went through the Noritsu. When rebate not even on all sides, I assumed they didn't (or couldn't) get it positioned perfectly.

But when I asked them to rescan the Portra 160 today, I was told that how much rebate shows is determined by "...... where the exposure falls within the edges of the film itself. Exposure can also cause issues with reading frame lines, underexposure, flares and light leaks can cause misalignment of the scans."

That sounds strange to me. I don't understand how exposure within the image would affect whether rebate shows in the scan.

They also told me today they would rescan the film if I let them cut it into strips, as they can control the film better when in strips, rather than uncut roll. That would indicate the physical position of the film in the Noritsu is the issue, not exposure of image.

A while back, when I questioned why it costs so much more per roll to show rebate in the scan ($7), their response was that they have to put a new film holder in the scanner, and that takes time. This would also indicate it is the physical position of film in the scanner which determines rebate.

(Incidentally, it seems to me one should just have to pay once to have the holder changed if having a number of rolls scanned, not be charged $7 for each roll. But that is another issue.)

Anyone work with Noritsu? Does what I was told re exposure and how much of rebate shows make any sense at all? I don't know what model Noritsu they use.

I do plan to get back into scanning myself. I have a Nikon LS 9000.

Thanks.

 

[koraks]

But when I asked them to rescan the Portra 160 today, I was told that how much rebate shows is determined by "...... where the exposure falls within the edges of the film itself. Exposure can also cause issues with reading frame lines, underexposure, flares and light leaks can cause misalignment of the scans."

That sounds strange to me. I don't understand how exposure within the image would affect whether rebate shows in the scan.

I don't think that's what they're saying, though. The way I read their quote is that they're saying two things:
1: The amount of rebate is influenced by where the exposure falls on the film. The gate and alignment on cameras is not always the same. Some cameras will expose a slightly larger frame, some expose a smaller frame. Some may be slightly offset to one edge of the film.
2: Frame recognition is influenced by exposure; e.g. a frame with one half of severe underexposure may not be correctly recognized and this will affect the scanner's choice of frame alignment. The result may be the rebate being wider or narrower (or lacking) on either side.

Perhaps @brbo can comment on this; I think he has done extensive work with Noritsu scanners.

 

[250swb]

I've yet to find a negative holder for any scanner that gives consistent borders, especially if using different cameras and lenses with wide angle lenses making a larger image on the negative than telephoto lenses. So with a film whizzing through the machine it has little control except to prioritise the image area. I'm sure if it had a brain the Noritsu scanner would think success was to make 36 scans all the same size without a border showing. Perhaps shorter six exposure lengths of film allow the lab to turn it 180 degrees and try scanning it rotated which can change where the border comes, and this is one thing you'd do at home if a border was important. Ultimately there are only three or shots on a film that are keepers if a good photographer is lucky, so nowadays adding a border with software is faster and more cost effective than paying for somebody to struggle and possibly fail.

 

[calebarchie]

Loading your film in camera slightly askew can cause this, also it's just a drop down menu.

 

[brbo]

My experience with Noritsu scanners is limited to LS-600 which only takes 135 and APS film. But since film transport is similar in the holders for their 120 scanners, I can totally see why consistency in showing film borders is definitely not trivial.

You have to consider they might have different scanners (each will show more or less rebate because tolerances will come in play), as mentioned different cameras will have different frame sizes, even lens with different focal lengths might have slightly different frame sizes exposed on film, orientation of the film when fed into scanner will matter, film transport in camera and in scanner.

But it mostly comes down to how the Noritsu scanners operate. Even if operator sets positioning per every single frame during the preview, those marks will never be hit exactly during scanning. Film transport in Noritsu scanners relies on rubber rollers in contact with the film. Any of the many rollers not perfectly round, grippy enough, etc. will cause that scanner won't be able to hit the mark set at the preview stage. You would be surprised how different the geometry of two scans of the same frame can be on this type of scanners (same goes for Flextights, for example).

And, as @250swb mentioned, the goal with those scanners was to avoid showing borders at any cost. You can get some additional frame estate from Noritsu raw files but that is some fuckery no lab will do for you even if you paid them $100 extra.

 

[koraks]

You would be surprised how different the geometry of two scans of the same frame can be on this type of scanners (same goes for Flextights, for example).

It's somewhat offtopic, but I've been operating a Flextight for a few months now, and it consistently hits the alignment within fractions of a mm on the device I've been using, which is an old Precision II unit. Maybe the transport was changed in later versions? Initially I allowed for a fairly liberal border around e.g. a 35mm frame, but I'm now entirely comfortable drawing a very tight bounding box just around the aperture in the knowledge that I'll get precisely that crop in the final file. I was actually surprised at how well this works.

 

[brbo]

@koraks, it would be odd if newer versions would change film transport to a less precise method, although targeting faster scan times might have an effect on precision.

If you would have the time to make two consecutive scans of the same frame and then make a gif toggling between those two scans, I would be very surprised if it showed no difference in registration. This “problem” is fairly well known*. But it may be limited to scanners that are not maintained well, though.


* I believe seeing a post even here on Photrio (from @dokko if memory serves me right) demonstrating this (I'll try to find the post later)

edit: here is the link to @dokko post showing differences between consecutive scans on 848

 

[foc]

I had always assumed the reason the rebate was uneven was a physical issue -- how the film was positioned when it went through the Noritsu. When rebate not even on all sides, I assumed they didn't (or couldn't) get it positioned perfectly.

+1

Although I never operated a Noritsu scanner, I have done so on Frontier. While different equipment, they operate basically the same.

As has been mentioned, the native negative carrier was designed NOT to show any rebate. Custom made carriers can be ordered that will show the rebate on the negative. Also the software must be adjusted to show the negative image and the rebate area in the scan. While it involves a bit of investment in time and money on the lab's behalf, once set up, it should run smoothly (famous last words!!!)

I know that some 120 film cameras expose a slightly larger negative image area than others but it would only be 1 or 2 mm, I suspect, at the most.

I think the uneven rebate surrounding in the scans may be down to operator error. The negative is simply not positioned in the centre of the carrier.

When I operated Frontier 120 scanning, many moons ago, the negative carriers were all manual. The negative advance and positioning was manually adjusted by the operator.

I can understand the lab charging extra for including the rebate in the scan, besides in the initial carrier/software time investment, the positioning the image to show an even rebate would be more time consuming than normal scan.

 

[brbo]

@foc, later Frontiers all used area CCD sensors which are immune to distortions present in line CCD scanners with less than 100% perfect film transport during scan time.

 

[koraks]

If you would have the time to make two consecutive scans of the same frame and then make a gif toggling between those two scans, I would be very surprised if it showed no difference in registration. This “problem” is fairly well known*. But it may be limited to scanners that are not maintained well, though.

There's a minimal shift that in practice presents no problems for me. The unmodified film holder for this scanner does not allow scanning the rebate. If one were to modify it (I assume this can be done without problems), the scan margin can simply be set wide enough to accommodate the shift.

Note that the maximum shift between these 3 frames was around 4 pixels, or a 160um distance. Note also it's only along the longitudinal axis, which makes sense as that's pretty much the only one on which any shift can be produced due to longitudinal transport inconsistencies.

What's interesting furthermore is that the actual aperture of the film holder does not shift - only the image frame inside it. This suggests the film shifts a little back & forth within the holder. I think this is due to the unique approach of the FlexTight (mind the name!) which bends the film + holder during scanning to create a slit/line of perfect focus. The side effect is that the film can move about a tiny bit as it's not a solid fit within the film holder. This also sets the phenomenon conceptually apart from what you'd see in a Noritsu-type scanner that evidently does not have this kind of holder. In this scanner, the inconsistencies will indeed be mostly due to roller transport. I emphasize that in the FlexTight example shown here the film gate remains in absolutely perfect alignment, suggesting that the problem is really within the film/holder assembly and not a roller inconsistency. So it really is a very different phenomenon.

Sorry for the offtopic; insofar as we can derive something from this example, I think it boils down to "if possible, scan with a margin to account for virtually inevitable inconsistencies". I've not (yet) repeated the experiment with my Scan Dual IV, but I'm sure there will be a small inconsistency in that one as well, however small.

A further perhaps marginally relevant remark is if you scan the same negative on different scanners, you'll get files that won't perfectly overlap. There will always be a degree of longitudinal or lateral compression or expansion going on. The exception of course is any well-aligned and well-compensated system that captures the frame in one go instead of scanning it, assuming that any geometrical aberrations are eliminated or at least corrected for.

I think the main problem in the end when it comes to rebates is that most scanners just aren't designed to scan that area so you're pretty much always working with a film gate/aperture that's a little too small to do what we want.

 

[brbo]

Note that the maximum shift between these 3 frames was around 4 pixels, or a 160um distance. Note also it's only along the longitudinal axis, which makes sense as that's pretty much the only one on which any shift can be produced due to longitudinal transport inconsistencies.

It feels like more than 4px or is the scan resolution set very low so that 4px mean quite visible shift in frame content? Anyway, I think even if OP is getting a variation this small it will be easily visible in the amount of rebate that he gets from the lab.

Sorry for the offtopic; insofar as we can derive something from this example, I think it boils down to "if possible, scan with a margin to account for virtually inevitable inconsistencies". I've not (yet) repeated the experiment with my Scan Dual IV, but I'm sure there will be a small inconsistency in that one as well, however small.

All my scanners with geared transport show very little variation. Drum scanners also has very very little variation...

A further perhaps marginally relevant remark is if you scan the same negative on different scanners, you'll get files that won't perfectly overlap.

... and I observe huge variation between different scanners.

I donated my Microtek 4000tf and Canoscan FS4000US so I can't do further test between the two, but the last scans I did on them showed absolutely huge differences in geometry:

 

[koraks]

It feels like more than 4px or is the scan resolution set very low so that 4px mean quite visible shift in frame content?

I offered a translation into real-world shift, which is around 160um. Of course this is visible, but as said, given a sufficiently large aperture, this could be allowed for. The problem is likely that the Noritsu doesn't easily offer this opportunity. IDK, I've never used one.

I donated my Microtek 4000tf and Canoscan FS4000US so I can't do further test between the two, but the last scans I did on them showed absolutely huge differences in geometry:

But a pretty good job color matching those scans!

 

[brbo]

I offered a translation into real-world shift, which is around 160um.

I know. It's just that my math was different, I came to the difference of 1% (4px in 400px width) --> 0.35mm for a 35mm frame, so I wasn't sure whether your real-world number was a bit off or (more probably) my math was flawed

Anyway, my original contribution was merely that there are scanners where pixel perfect repeatability across different scans is impossible. I'm quite sure that many scanners have inconsistencies in the range of 1px (at full resolution) or less*. Flextights and Noritsus are sadly not among them.

My message to OP therefore was that lab is probably not inventing excuses or just getting more money from him/her without doing any actual work.

* for example, my Minolta 5400 that people regularly call a flimsy plastic scanner, has virtually zero shift if you know what you are doing. Links to 2 full res files done one after another: file1, file2 (more interesting bit from about this scans is that they are form the new Orwo NC200/Opticolour 200 preproduction film; sadly this film might not be what I was hoping for after looking at sample scans from Optik OldSchool)

 

[koraks]

I know. It's just that my math was different, I came to the difference of 1% (4px in 400px width) --> 0.35mm for a 35mm frame, so I wasn't sure whether your real-world number was a bit off or (more probably) my math was flawed

Sorry, I should have been clear; I downsize the file for display here. The difference on the original 600dpi scan was 4px.

(more interesting bit from about this scans is that they are form the new Orwo NC200/Opticolour 200 preproduction film; I'm sadly not a fan)

I hope you'll post those elsewhere as well; I'd like to see the scans. Btw, I get an 'access denied' on Drive when I try to open the links.

 

[brbo]

I hope you'll post those elsewhere as well; I'd like to see the scans. Btw, I get an 'access denied' on Drive when I try to open the links.

Can you check if the links work now? Thanks!

Yes, I was planning to write a short post about this film after I've also made a RA-4 contact print to see how it does there, but kind of a lost interest because I already know it's not going to look better when scanned. I'm also not too keen to post my not so enthusiastic thoughts on this film since it's (a) preproduction, (b) my film spent some days stuck in an airport at who know what temperatures and (c) it could be just my processing. Might still do it anyway...

 

[koraks]

Yeah, the downloads work fine now. And those files do indeed look absolutely identical!

PS: I think I see what you mean w.r.t. this film. Hue purity is kind of...limited. And it's on the grainy side for its speed. Might be nice if you're looking for something really muted.

 

[calico]

I don't think that's what they're saying, though. The way I read their quote is that they're saying two things:
1: The amount of rebate is influenced by where the exposure falls on the film. The gate and alignment on cameras is not always the same. Some cameras will expose a slightly larger frame, some expose a smaller frame. Some may be slightly offset to one edge of the film.
2: Frame recognition is influenced by exposure; e.g. a frame with one half of severe underexposure may not be correctly recognized and this will affect the scanner's choice of frame alignment. The result may be the rebate being wider or narrower (or lacking) on either side.

Perhaps @brbo can comment on this; I think he has done extensive work with Noritsu scanners.

I should have mentioned in my original post that the two rolls in this recent order -- the Provia where the rebate showed on all four sides in all frames, and the Portra 160 where the rebate only showed on two sides and was very wide -- were both shot on same camera with same lens (Hassy 201F and 110/2). So frames should've been same size and fallen in same area of film. I don't remember if I used the same film back. Seems the back wouldn't be factor in this situation.

The notable difference is that one was positive film and one was negative. Or, as someone else mentioned, maybe two different people scanned the two rolls, or maybe they were scanned on different machines.

Your number 2 point is interesting and seems to be what the lab was telling me -- that the scanner is choosing frame alignment based on how film was exposed, not based on how film was placed in the machine.

This is a strange concept for me, as I'm used to my Nikon LS9000 where how I place the film in my glass carrier is what determines what is scanned. I'm able to get nice, even rebate edges if I place the film properly on the glass in carrier (I have to put a little of piece of tape on edge of film so film won't move when I put top of carrier down). Then the scanner scans as I've placed it. I can crop in Vuescan or Nikon Scan before final scan, of course, but what I see in preview is a result of how I placed the film in carrier.

I realize my scanner totally different animal than a lab scanner. The lab scanners are shooting film through via a transport.

Thank you for your input.

 

[calico]

I've yet to find a negative holder for any scanner that gives consistent borders, especially if using different cameras and lenses with wide angle lenses making a larger image on the negative than telephoto lenses. So with a film whizzing through the machine it has little control except to prioritise the image area. I'm sure if it had a brain the Noritsu scanner would think success was to make 36 scans all the same size without a border showing. Perhaps shorter six exposure lengths of film allow the lab to turn it 180 degrees and try scanning it rotated which can change where the border comes, and this is one thing you'd do at home if a border was important. Ultimately there are only three or shots on a film that are keepers if a good photographer is lucky, so nowadays adding a border with software is faster and more cost effective than paying for somebody to struggle and possibly fail.

Thanks for your input. Both rolls were shot with same camera and lens, so the size, position of frame on film shouldn't have been factor (one roll had good rebate, other didn't).

 

[calico]

You have to consider they might have different scanners (each will show more or less rebate because tolerances will come in play), as mentioned different cameras will have different frame sizes, even lens with different focal lengths might have slightly different frame sizes exposed on film, orientation of the film when fed into scanner will matter, film transport in camera and in scanner.

Just fyi -- both rolls (the one with good rebate and other with bad rebate) were shot with same camera and lens.

But it mostly comes down to how the Noritsu scanners operate. Even if operator sets positioning per every single frame during the preview, those marks will never be hit exactly during scanning. Film transport in Noritsu scanners relies on rubber rollers in contact with the film. Any of the many rollers not perfectly round, grippy enough, etc. will cause that scanner won't be able to hit the mark set at the preview stage. You would be surprised how different the geometry of two scans of the same frame can be on this type of scanners (same goes for Flextights, for example).

That all makes sense to me, the physical issues of running film through a machine. It is the idea that scanner is basing framing on how the image is exposed in camera that is alien to me, as others have suggested.

And, as @250swb mentioned, the goal with those scanners was to avoid showing borders at any cost. You can get some additional frame estate from Noritsu raw files but that is some fuckery no lab will do for you even if you paid them $100 extra.

I believe the lab has a special film carrier which allows for rebate to show, which is reason for the $7 charge to include rebate -- time to put different carrier in machine.

 

[calico]

+1

Although I never operated a Noritsu scanner, I have done so on Frontier. While different equipment, they operate basically the same.

As has been mentioned, the native negative carrier was designed NOT to show any rebate. Custom made carriers can be ordered that will show the rebate on the negative. Also the software must be adjusted to show the negative image and the rebate area in the scan. While it involves a bit of investment in time and money on the lab's behalf, once set up, it should run smoothly (famous last words!!!)

I know that some 120 film cameras expose a slightly larger negative image area than others but it would only be 1 or 2 mm, I suspect, at the most.

I think the uneven rebate surrounding in the scans may be down to operator error. The negative is simply not positioned in the centre of the carrier.

When I operated Frontier 120 scanning, many moons ago, the negative carriers were all manual. The negative advance and positioning was manually adjusted by the operator.

I can understand the lab charging extra for including the rebate in the scan, besides in the initial carrier/software time investment, the positioning the image to show an even rebate would be more time consuming than normal scan.

I tend to think this, too -- it was operator error, the negative not centered in the carrier. Or the adjustment in software you mention wasn't done. Especially as both rolls (the one where rebate turned out okay and the one where it didn't) were shot with same camera and lens. Maybe two different people scanned the two rolls and one person was more careful than the other.

The fact they said they could do better if I let them cut the film into strips for the second try points to positioning of film in carrier, operator control issue. Cutting in strips wouldn't have anything to do with how my frames were exposed in camera (which is explanation they and some people here give for alignment issue).

Thanks for your input.

 

[calico]

I want to thank everyone for taking the time to offer such long and thoughtful responses to my post. Really appreciate the helpful folks in this forum.

Bottom line re my lab scans: I need to get back to scanning my film myself.

I also need to relearn how to develop my own b&w film (did it many, many years ago). Another lab messed up a roll of FP4 film recently. Dealing with labs can be such a pain.

 

[koraks]

Yeah, it's at least convenient to have the capability to do it yourself, and generally it's also more rewarding. And if you are in a position to invest sufficient effort into the whole thing, the end result can be better, too.

 

[calico]

Yeah, it's at least convenient to have the capability to do it yourself, and generally it's also more rewarding. And if you are in a position to invest sufficient effort into the whole thing, the end result can be better, too.

Agree -- my scans from my LS9000 much better than their scans.

I don't like that the lab uses sRGB color space, rather than RGB 1998. Also, they only offer jpegs, not tiffs (which could be 16 bit).

They are geared to people who make prints at their lab, where those things probably don't matter. Or people who just post on internet.

Problem has been my lack of time recently. But will get back to scanning myself asap.

Thanks.

 

[koraks]

Their not offering tiff would be a dealbreaker for me; 8 bit color space is a problem. The sRGB vs Adobe RGB isnt much of an issue; there's very, very little real world gain in that and arguably, working with color negative film, there's no benefit at all to adobergb. What's more relevant is that they probably send you inverted, color-balanced digital files, which means that most of the flexibility is already gone from whatever they send you.

 

[calebarchie]

So many words, yet I work with these machines everyday.