Epson Scan Software

[TonyW]

Ray, I agree that there are potential issues in trying to evaluate scans between models in the way shown in the link.  There does not appear to be any detail given on the conditions and in the case of the last image comparison at least different images used.

That’s why I said not the best example, however the differences shown illustrate what I would expect to see from the v700 or v750 when both were set up as well as possible and having seen scans from both the Epson and the Nikon (5000ED I think) using the same image and optimal quality settings.  EDIT: There are two entries for the v700 and I took the first one as the comparison.
 
When I first saw the crops I too thought they were quite different, however If you copy and bring both images into PS as layers and set difference blend mode and move one of the images you should see that they line up pretty accurately as far as size goes.  Quick example attached using the middle image.  The only thing I did was to adjust colour balance a little to get a slightly improved visual match and just crop the whole image down to the v700 size.  Wait a few seconds for the image to change.


I would think that you may actually see the differences better with your 35mm mounted behind glass and accurate focus set for both scanners.

At the end of the day you are getting very good results from the Epson and the question is how much more detail do you need to reveal and would it yield a significant benefit once printed after good sharpening and noise reduction.
 
I agree with artobest that one persons testing methodology on two different pieces of equipment needs to be undertaken with great attention to detail to ensure that both are performing at optimum levels to make a meaningful comparison.        

[Ray]

  EDIT: There are two entries for the v700 and I took the first one as the comparison.
 

Ah! My fault. I clicked on the second entry. That's a good method you are using to demonstrate the differences, Tony.

My interest in this matter is mainly for practical reasons. I've had a Nikon 8000 ED scanner for quite a few years, but cannot now use it with my recent computers, or even the old, original computer, without getting that computer repaired.

The question for me is, should I buy the Silverfast software which enables me to continue using this Nikon scanner? I'm currently using the trial software with numerous watermarks plastered over the image, and I'm trying to find out if there would be any occasion when I might want to use the Nikon scanner to eke out slightly more detail, either for the purpose of making a very large print, or for the purpose of making an average size print from a small crop.

Whilst continuing with my MF scanning of old negatives, I came across a portrait-type image with a shallow DoF and a detailed focusing area in the centre of the frame. I thought this should be an ideal shot to demonstrate any detail and sharpness differences. There should be no problem selecting a focusing point.

It's a shot of a Hill Tribe person in Northern Thailand, standing to attention for his portrait.

In order to avoid any confusion resulting from different sharpening methods from different software and hardware, I scanned the image on both scanners without applying any sharpening or noise reduction whatsoever. If one or both of the scanners are applying some degree of sharpening 'under the hood', or by default, then there's nothing I can do about that.

I was surprised to see that there is a big difference in apparent sharpness and detail between the two scans when both images are unsharpened. The differences are like the difference between an excellent prime lens and a mediocre zoom lens, and similar to the differences shown at the website you linked to, at http://zeux.zlakfoto.ch/scanvergleich/index.html

But again, the question is, does the Nikon scan contain any detail that cannot be brought out in the Epson scan after appropriate sharpening?

In the attached examples, I upsampled the Nikon scan to the same file size as the Epson scan. I then initially applied equal sharpening to both images. I can't be fairer than that. But what I saw was significantly more noise in the Nikon scan on smooth areas, such as skin. This indicates to me that the reason for the 'apparently' unsharpened Nikon scan looking sharper than the unsharpened Epson scan is because some degree of sharpening has been applied by default to the Nikon scan, which is not user-controllable.

I've therefore applied different degrees of sharpening, as appropriate, to each scan, but without application of any noise reduction.

I'm trying to find even one skerrick of detail that exists in the Nikon scan, but not in the Epson scan. I can't see it. Even in the 200% crop which shows the individual threads of the guy's shirt, I can see no additional detail in the Nikon scan.

Of course, I accept the possibility that a Nikon scan of a line chart taken with a state-of-the-art lens, might show a few additional lines, but that's not what I'm scanning.

Cheers!

[TonyW]

Ah! My fault. I clicked on the second entry. That's a good method you are using to demonstrate the differences, Tony.

Still raises the question why two entries and different sizes so you would be correct in questioning the testing methods as in your first reply.  The example was posted to illustrate  what differences I would expect to see between a good quality flatbed (and I think the Epson 700 and 750 are about as good as it gets at the price point)  and a good quality dedicated film scanner such as your Nikon.  I still think there is an issue with the Epson scan as it seems to exhibit a small degree of distortion which suggests that the film not held flat thereby focus not optimum.

The question for me is, should I buy the Silverfast software which enables me to continue using this Nikon scanner? I'm currently using the trial software with numerous watermarks plastered over the image, and I'm trying to find out if there would be any occasion when I might want to use the Nikon scanner to eke out slightly more detail, either for the purpose of making a very large print, or for the purpose of making an average size print from a small crop.

Have you tried the Vuescan demo for comparison as it does offer support for the Nikon 8000ED?  It may or may not bring anything different to the table but a lot of people seem to like it – I do but as I said earlier I tend to use the Epson software for print copying.
Vuescan demo http://www.hamrick.com/vuescan/nikon_coolscan_8000_ed.html  support for Win7 and 8 64 bit.

I was surprised to see that there is a big difference in apparent sharpness and detail between the two scans when both images are unsharpened. The differences are like the difference between an excellent prime lens and a mediocre zoom lens, and similar to the differences shown at the website you linked to, at http://zeux.zlakfoto.ch/scanvergleich/index.html

But again, the question is, does the Nikon scan contain any detail that cannot be brought out in the Epson scan after appropriate sharpening?

At least in theory if one scanner is not able to resolve the finest detail it will result in a mush of that detail which no sharpening technique (that I know of!) will bring back.  Question then would be, is the loss small enough to be acceptable when the image is considered as a whole?  

In the attached examples, I upsampled the Nikon scan to the same file size as the Epson scan. I then initially applied equal sharpening to both images. I can't be fairer than that. But what I saw was significantly more noise in the Nikon scan on smooth areas, such as skin. This indicates to me that the reason for the 'apparently' unsharpened Nikon scan looking sharper than the unsharpened Epson scan is because some degree of sharpening has been applied by default to the Nikon scan, which is not user-controllable.

Cannot see examples however I accept what you are seeing but wonder again about the differences.  Just a few random thoughts and not a suggestion that you are doing anything incorrectly .
Firstly as you upsampled the Nikon scan what upsampling algorithm did you choose if Photoshop and have you tried the same upsampling in either LR or ACR?  It may offer an improvement, although I cannot guarantee this will be the case.  Going through LR or ACR I think could offer you a better sharpening (including masking that you can actually see where it is being applied) and noise reduction at this stage of your capture.  
How does the Epson scan look when downsampled to the Nikon size?
Are you using the same software for both the Nikon and Epson scanners as I would have thought that even if sharpening is going on under the hood then it would be the same for both?

I'm trying to find even one skerrick of detail that exists in the Nikon scan, but not in the Epson scan. I can't see it. Even in the 200% crop which shows the individual threads of the guy's shirt, I can see no additional detail in the Nikon scan.

Seems to me that you are already achieving the maximum detail from your Epson scans and there is just no more to be had even with a theoretically better system.

Bear in mind that my opinion about the difference in achievable quality and although based on previous experience of very good from Nikon vs not as good from flatbed (even the high quality ones) is subjective – I do not have the equipment to hand to try and put theory to test.  Perhaps my expectations are too high for the Nikon?  On the other hand you are able to undertake objective testing based on the two pieces of equipment you have and unless something is out of whack with the Nikon then perhaps this is as good as it gets?

EDIT: After posting I now see the images you refer to.  First look seems that pretty much trying to split hairs here between the two scans.  But it does appear that they were both brought into different applications and I therefore think that it would be better if both were scanned using the same software and whatever gets applied in the application or by your own manipulations would even the field somewhat.  It looks like the Lasersoft applications is adding something in this example that I would prefer not to see.

Again at first sight and looking at the unsharpened comparison.  The left image Lasersoft looks a little sharper however in the process it seems that detail is lost in the nose and artefacting apparent resulting in loss of tonality changes.  I much prefer the starting point image shown on the right as it appears to offer a much smoother transition between tonalities - somehow I think I am betting on the Epson scan here? 
Curious and I hope you do not mind but I took a portion of your unsharpened results and applied a little sharpening to the right hand image only shown in the bottom right and I must say I prefer the result to the Lasersoft example - smoother tonality and highlights showing without clipping.  Still while we can try to compare nuances on screen the final quality is going to be only seen once printed.

[Ray]

Tony,
Thanks for your feedback. Good to know that Vuescan contains drivers to enable use of the Nikon 8000 ED with Windows 7.

I did use both the Silverfast software and the Epson scan software when scanning this negative on the Epson scanner, but I didn't find any noticeable difference in sharpness and detail in the two scans when  no sharpening or noise reduction had been applied to either. But there can be a slight difference in tonality, brightness and contrast, unless one is very meticulous, because the adjustment tools in the different software are different.

The upsampling algorithm I used for the Nikon scan was Bicubic Smoother (Best for Enlargement). However, I notice that Photoshop Creative Cloud has introduced a new option for upsampling, called 'Preserve Details (Enlargement)'. I've just tried it, and I see a very marginal improvement in contrast and detail when using 'Preserve Details'.

Which brings me to the next point. I think my reasoning is flawed when I draw the inference that the additional noise and appearance of oversharpening in the Nikon scan (after applying the same amount of sharpening to the upsampled image as applied to the Epson scan) might suggest that some default sharpening has been applied by the scanning software to the Nikon scan.

I don't think this is necessarily correct. I think it's more likely that the lens in the Nikon scanner is a more contrasty lens than the lens in the Epson flatbed.

Image resolution is always a combination of lens and sensor resolving power. I believe the combination of the lens and the sensor in the Nikon scanner is simply producing a more contrasty image than the Epson scanner, but not necessarily a more detailed image.

Now, as I understand, as long as detail exists it can be brought out with sharpening. If detail doesn't exist even faintly in the scan, because either the lens or the sensor lacks sufficient resolving  power, then nothing can be done, short of using some complicated algorithm which might make a guess as to what the detail might have been.

The fact that I have so far been unable to discern any detail in the Nikon scan which I cannot bring out in the Epson scan with appropriate sharpening, or even inappropriate sharpening, such as severely oversharpening both images, suggests to me that the main advantage of the 8000 ED is that it produces better 'apparent' resolution straight out of the box, whereas the Epson V700 needs more work in post processing, particularly with regard sharpening.

To be more certain about this, I guess I would need a negative or slide of a high quality resolution-chart to scan. However, I find it difficult to believe that the fine hairs, skin pores, film grain, marks, blemishes, dust spots and fine scratches on the film surface do not provide sufficient detail to make a good comparative assessment of the resolving power of the two scanners.

Searching for the faintest scratch on the negative, I found one positioned over the Hill Tribe guy's neck. This scratch is significantly thinner than a single strand of hair on the guy's chin.

The attach image compares 100% crops of the original 6400 dpi scan from the Epson V700, using the Silverfast software, and the Nikon 8000 ED scan after upsampling using 'Preserve Details'.

Both images have been excessively sharpened to bring out the maximum detail, but the Epson scan needed significantly more sharpening to get the detail appearing the same.

I can't see the need to scan a resolution chart. Surely there can be no finer detail than the grain of the film, and that fine scratch rising from the bottom of the frame, left of centre. 

I should also mention that these file sizes are around 380 MB in 16 bit greyscale. In 48 bit color they would be more than 1 GB and any print of the entire image, that would be large enough to reveal that fine scratch as you see it on your monitor, would probably not fit on your wall, from floor to ceiling.

Have I made my case? 

By the way, you have to be logged on to see images which are attachments.

[Alan Klein]

Why do the right pictures have the bright areas on the nose?  Why do the left pictures have these two "tags" on the nose?

[Ray]

Why do the right pictures have the bright areas on the nose?  Why do the left pictures have these two "tags" on the nose?

Alan,
The bright area on the nose in the right image is a highlight which was preserved as a result of the levels, brightness and contrast adjustments in the Epson scan software which I used with the Epson scanner.

When scanning the same negative on the Nikon 8000 ED (left image) that highlight was reduced in intensity as a result of slightly different brightness/contrast adjustments I made with the different scanning software, which was Silverfast.

I presume the two tags you are referring to are the two small white spots on the end of the nose which are missing in the right crop. These are dust spots which would have been removed through my normal practice of brushing and blowing the negatives after inserting them into the holder, and before placing the holder onto or into the scanner.

As I recall, the first scan would have been on the Epson scanner using the Epson scan software. After examining the result, I decided to use the negative to compare detail from the Nikon 8000 ED, but first I would have rescanned the image on the Epson without sharpening or noise reduction applied. I would then have removed the negative to scan on the Nikon.

As an afterthought, after scanning the negative on the Nikon scanner, I then inserted the negative a second time into the Epson film holder in order to scan it again using the Silverfast software, which was included with the Epson Scan software that came with the scanner, just to see if the different software had any effect on sharpness and detail when used with the same scanner. It didn't.

Those two dust spots on the end of the nose would have been removed as a result of the additional blowing and brushing of the negative before it was inserted a second time onto the Epson scanner.

I think I am now ready to apply for a job as a forensic scientist. 

[TonyW]

Ray,
As you said in your final statement ‘Have I made my case’, accompanied by a smiley I hope you realise I am not trying to be argumentative or anal (I used to get paid for this level of anality in image evaluation  ).  My interest is that you do not seem to be getting optimum results from the Nikon and if it was my unit I would be investigating a little further as to the reasons.

I really think that it is not a good idea to compare scans from two different applications regardless of setting both to no noise reduction and no sharpening.  We just cannot be certain of what is happening under the hood with the initial algorithms for each application – they are likely to be quite different.  Similarly when we do apply changes within the particular application e.g. sharpening, noise reduction, curves or levels the application programmers will have used their own algorithms to achieve the effect

It is my belief that apart from any contrast differences contributing to perceived resolution the Nikon optical system is superior to the Epson therefore it is actually capable of resolving more detail as long as that detail exists in the original.  How much added resolving power is the $64,000 question and batch differences may be observed between models from the same manufacturer.

If detail exists in a scan it can be enhanced with sharpening of course and to a degree any losses is the digitization process can be at least disguised if not completely eliminated.  Slight digression here but a few years ago I was considering if I should jump to all digital capture or utilise my existing 35mm and MF kit.  I had thought about finding a used Nikon scanner either 35mm or even MF like the 8000ED or go out for high quality scans to a bureau.  In the process a colleague sent me a comparison scan (attached) from his Nikon 5000 vs either Epson 700 or 750 (cannot remember which).  I was assured that the conditions for the scanning was optimum in both cases and the test slide was a glass mounted 35mm SMPTE RP40 (usual use projector alignment).  In the case of the Epson the film height adjusted to optimum.  Although there is some difference in contrast (the centre dot of the star matches density exactly between images) there is a discernible loss of detail  in the separation of some of the bars that cannot really be recovered by sharpening as this detail has been lost in a single grey mush.  I know that the 5000 is a different beast being 35mm to the 8000 but the optical systems I understand are similar enough to consider comparison appropriate.


The question of course is how important this loss is to the impact of the image as a whole considering that this is a very small section from a 35mm test object and may not be as apparent in a MF shot?

I doubt the value of using a test object over a normal film other than if you need to establish quality differences to the absolute degree between units or need to confirm a particular unit is performing consistently or within a given specification.

Your results from the Epson are very good indeed and this may be all you need, but you did suggest that you were wondering if you could eke a small amount of usable information for either image crops or large images by using the Nikon.  So I can only say if this was my system I would be investigating a little further to see what may be causing possible losses. 
I wish I could tell you that by taking a certain path would lead you to a noticeable improvement for the Nikon – quite simply I cannot and you may be wasting your time and effort looking any further but...

One thing that will effect quality is dirty scanner optics, lens, mirror or both.  Worst case that should be very noticeable is blooming but prior to things getting that bad a slight coating on either mirror or lens can result in a loss of resolution.
These may be of interest:

Dirty Scanner Optics - http://www.vad1.com/photo/dirty-scanner/
If you are brave and feel competent to undertake it self cleaning may help? - http://www.marginalsoftware.com/LS8000Notes/cleaning_the_optics_of_the_ls.htm

[Alan Klein]

The reason for my question is that because there was a white glare spot on only one set, then the images were processed differently.  Once you introduce different adjustments, it's hard to determine which is doing what.  There are so many variables.  Probably the best way to compre is too try to max out the processing as best you can with both scanners.  Then print both results and compare.  After all, it's the final results that count.

[Ray]

My interest is that you do not seem to be getting optimum results from the Nikon and if it was my unit I would be investigating a little further as to the reasons.

Tony,
Thanks very much for those links to 'dirty scanner optics' and how to disassemble the 8000 ED for cleaning.  I think I should try to do this. I'm going to print out the instructions so I can refer to them whilst I try disassembling the scanner, and hope I don't make a mess of the job. 

When I decided to get this Nikon scanner operational again, I was kicking myself for not having wrapped it in a plastic bag to protect it from the dust. However, I can't say that I've noticed any of those tell-tale signs of dirty scanner optics mentioned in the article, at least not in any significant way. Perhaps this is because most of the film I am scanning is quite old and faded with a few scratches and marks. One doesn't expect a perfect result.

However, I do have some scans of Kodachrome slides I made when I first acquired this Nikon scanner back in 2003 when it was brand new. I'm now going to search for some of those scans, and the original slides which I'll re-scan on the Epson V700. Watch this space. 

[Ray]

The reason for my question is that because there was a white glare spot on only one set, then the images were processed differently. 

Alan,
As far as I understand, the images are unavoidably processed differently because I'm comparing different scanners. If I were to compare images from a Nikon and a Canon DSLR, the images would also be processed differently in the RAW converter.

One could spend ages comparing different RAW converters, and one might find that one particular converter produced better results with the Nikon DSLR and another converter produced better results with a Canon RAW file.

This is why I reinserted the negative in the Epson scanner to scan it again with Silverfast. However, the Silverfast program is specific to particular brands of scanners. I can't see why one would assume that its operation will be identical whatever the scanner. As regards resolution, I could not see any difference between the two scans from the Epson scanner after using the two programs, Silverfast and Epson Scan, with the same negative.

Probably the best way to compare is too try to max out the processing as best you can with both scanners.  Then print both results and compare.  After all, it's the final results that count.

Alan, I'm doing archival scanning and simply trying to extract the most detail from the film. I don't know what the final result will be used for. I take it for granted that whatever I see on my monitor will be what I see on my print, allowing for the inherent differences between a transmissive image and a reflective image, and assuming that the print is large enough to show the detail at the same size it appears on the monitor.

That film scanners nowadays do not seem to be on that trajectory of continual improvement, gives me some urgency to get all my old film digitized. I'd much prefer to have all those boxes of slides and negatives, which take up quite a bit of space, preserved on just a couple of 2 TB pocket drives.

If I decide to make a print of any one of those thousands of negatives or slides, I can then do whatever processing is required, in Photoshop, to get the best or most pleasing print.

[Alan Klein]

As an aside,  I believe ICE won't work with Kodachrome color and BW film due to the way it's layered.

Also, if you plan on just scanning for archiving, have you considered just scanning flat?  Leave it for others later on to apply all post processing.  That way they will do what  is necessary at that time to get the best out of image files.  Also, processing techniques will be better in the future.  If you apply the processing now, they might not be able to take advantage of future processing improvements.

[Alan Klein]

PS:  If you scan and archive flat with no post processing, the whole archiving process will go much, much faster.

[TonyW]

Ray good luck with your scanner cleaning.

... However, the Silverfast program is specific to particular brands of scanners. I can't see why one would assume that its operation will be identical whatever the scanner....

Ray, I had forgotten that Silverfast is scanner specific and of course if you wanted to use it you have to purchase a Crossgrade version which still allows you to use your original scanner.  Silverfast crossgrade shows no price until you put in your Silverfast serial number into the order box - sounds like it could be costly?
https://www.silverfast.com/buyonline/en.html?productline=scanner&upgrade=1

On the other hand Vuescan should allow you to use any scanner that it supports by selecting from the input source in the dropdown menu - at least that is my understanding.
http://www.hamrick.com/vuescan/html/vuesc28.htm

[Ray]

The question of course is how important this loss is to the impact of the image as a whole considering that this is a very small section from a 35mm test object and may not be as apparent in a MF shot?

Tony,
This has occurred to me also. At extreme levels of pixel-peeping, after having applied the optimum degree of sharpening to each image, I believe the Nikon 8000 ED should have the edge, at least with 35mm format.

I haven't attempted to clean the scanner yet. I wanted to first assure myself that a clean 8000 ED is capable of producing better results to a significant degree, so I've searched for some of the early scans I made with the 8000 ED, back in 2003 when the scanner would have been new and clean.

The attached scan is of a scene of village life in Nepal about 50 years ago, showing the wife teaching her children how to weave fabric. Mahatma Gandhi would have been proud of her. 

The image is about as detailed as any of my old 35mm slides and negatives get.

The first comparison shows how soft the Epson image is without any sharpening applied. This difference is typical of the differences one sees in comparison scans on the internet, including the comparison of the test chart that you have shown in your post above.

However, the Epson V700 scan is a significantly larger file and can take more aggressive sharpening before noise becomes objectionable. What I find is there's always a balancing act one has to go through, between acceptable sharpening and acceptable noise. In this case, after sharpening the Epson scan and then downsampling, I noticed that, over all, the Epson scan seemed a bit sharper but had more noise. So I applied some more sharpening to the Nikon scan in an attempt to equalize noise..

I believe the result (in image 03) shows that the Nikon scan, in certain parts of the crop, has very marginally more detail, and perhaps even very marginally less noise in the skin tones.

However, I think it's likely that such small differences will get lost when scanning MF film. First because MF film generally doesn't have the resolution of 35mm film (in terms of line pairs per mm), and secondly because the less obtrusive grain in MF film allows for a greater degree of sharpening before noise becomes objectionable.

What do you think? Is it worth taking the pains, and slowing down the whole scanning process, in order to get such marginal benefits? 

[Ray]

Also, if you plan on just scanning for archiving, have you considered just scanning flat?  Leave it for others later on to apply all post processing.  That way they will do what  is necessary at that time to get the best out of image files.  Also, processing techniques will be better in the future.  If you apply the processing now, they might not be able to take advantage of future processing improvements.

That's a good point, Alan. The Silverfast software has a 64 bit HDRi option which allows one to scan the negative or slide as it is, without any adjustments, to produce what could be described as a RAW file. This means that the negative, after scanning, is still in 'negative' format. If it's a color negative film, one has the problem of dealing with the orange mask in post-processing.

I downloaded the Silverfast HDR (64 bit) trial software, which is designed specifically to process such files. I wondered if I'd be able to see any advantage, and what difficulties there would be in processing such RAW scans. So far, with the trial software, I've had difficulty getting satisfactory tonality and color. The post-processing seems to take far more time than the few basic adjustments in the scanning software plus further adjustments in Photoshop. So I'm not sure if this is the best approach.

[Alan Klein]

Have you tried to scan flat with the native Nikon and Epson software then use Photoshop afterwards?

[TonyW]

I haven't attempted to clean the scanner yet. I wanted to first assure myself that a clean 8000 ED is capable of producing better results to a significant degree, so I've searched for some of the early scans I made with the 8000 ED, back in 2003 when the scanner would have been new and clean.

So I understand correctly the scan on the left was produced when your Nikon scanner was new approximately 13 years ago and the Epson scan on the right was made now you have located the original?
Have you tried scanning this image again with the Nikon to observe if there is any noticeable degradation in quality?

The first comparison shows how soft the Epson image is without any sharpening applied. This difference is typical of the differences one sees in comparison scans on the internet, including the comparison of the test chart that you have shown in your post above.

IMHO this is the way it should be compared and judged.  The whole point is the native capability of the scanners optical system to resolve detail.  Clearly there is a large difference between the two scans shown here and the Epson just doesn’t quite cut the mustard for me as the fine detail has been lost in mush.  Sharpening will not regain lost detail just increase micro contrast and may give an overall impression of sharpness but clearly cannot resolve the fine detail contained in your image

However, the Epson V700 scan is a significantly larger file and can take more aggressive sharpening before noise becomes objectionable.

But it is still not possible to bring back information that was lost due to the scanner not being able to resolve the detail and even at 6400 SPI the Epson is not resolving any more information contained within the original image i.e. it has hit its limits which still probably equate to published figures

 

What I find is there's always a balancing act one has to go through, between acceptable sharpening and acceptable noise. In this case, after sharpening the Epson scan and then downsampling, I noticed that, over all, the Epson scan seemed a bit sharper but had more noise. So I applied some more sharpening to the Nikon scan in an attempt to equalize noise

I think there are a few things to be considered here about sharpness, noise and resolving power.  Sorry I know I am stating the obvious but sharpness does not mean that much if you are losing detail due to lack of the resolving power of either a scanner or one lens vs another.  It may mean that you are able to improve the look of the poorer image and get closer to optimum or acceptable result.  But the same improvements by judicious sharpening would apply equally to an image holding more detail in the first instance in this case being at the time of acquisition via your scanners.  Therefore the distance in quality between original unadulterated captures for both systems should remain pretty much the same when optimum sharpening applied to both images.

At this stage I would not be worrying too much about noise as I would expect to take care of it along with capture sharpening probably in either LR or ACR taking advantage of the visible masking in noise reduction panel.

I believe the result (in image 03) shows that the Nikon scan, in certain parts of the crop, has very marginally more detail, and perhaps even very marginally less noise in the skin tones.

I see more than marginally more detail but I really believe that due to trying to sharpen one to equal another and control noise that this may not be the best way to compare.

However, I think it's likely that such small differences will get lost when scanning MF film. First because MF film generally doesn't have the resolution of 35mm film (in terms of line pairs per mm), and secondly because the less obtrusive grain in MF film allows for a greater degree of sharpening before noise becomes objectionable.

I am not sure I can agree with the assertion about the differences in the ability of a particular film to resolve detail are higher with the same film stock between 35mm and MF.  In theory they should be pretty close if not exactly the same as the same emulsion brew is coated onto a film base which of course is likely to vary between 35mm, 120 and large format – I think any differences are therefore probably minimal.  That is until you start to make enlargements to the same size print from both formats then the difference can be considerable including sharpness how fine detail is resolved and of course grain.

I doubt very much that you are going to see any more from using HDR software at least as far as resolving more detail.  But I would urge you to also try the Vuescan software once again to compare both scanners through the same application.  This software also offers the ability to produce so called raw files

What do you think? Is it worth taking the pains, and slowing down the whole scanning process, in order to get such marginal benefits?  

The important thing is what do you want and is the potential gain enough justification or any added benefit?

FWIW what I think is:
Original Nikon scan superior to Epson without any processing applied therefore must be a better starting point for any sharpening tricks you may want to employ

Nikon scanner optics suspect due to now being over 10 years old without cleaning or service.  Even carefully wrapped and stored it is possible that lens and or mirror may have become even slightly coated due not necessarily dust but just atmospheric conditions around the storage area.  Therefore unknown at this time if still in pristine operating condition.

I would be rescanning the original image on the Nikon as a first test to see what may have changed then depending on the results consider cleaning – I might after all this time consider it appropriate anyway.

If I was quite happy that quality good enough for purpose then I may give up on the Nikon in favour of the Epson – but I do not like the thought of leaving image quality on the table so I would probably consider the cleaning route.  YMMV  

[Ray]

So I understand correctly the scan on the left was produced when your Nikon scanner was new approximately 13 years ago and the Epson scan on the right was made now you have located the original?
Have you tried scanning this image again with the Nikon to observe if there is any noticeable degradation in quality?

Hi Tony,
That's correct, but not quite 13 years ago. Closer to 11. The attached image of the metadata, image 01, shows that the file was created on 26/8/2013, which is very misleading. That would be the date I transferred the file from an old hard drive to a new hard drive in the process of reorganizing my images.

The date the file was modified, 3/05/2003, would have been around the time I made the scan.

I've now re-scanned this Kodachrome on the same Nikon scanner. If one is going to pixel-peep, then let's do it properly at 200%. 

I've chosen this particular area of the film because the lady's head scarf brings out the resolution differences in the most obvious way and to the greatest degree.The comparison crops (image 02) show the detail to be very close. In fact, the recent scan, without sharpening or noise reduction applied, appears very marginally sharper than the 11-year old scan, but also very marginally noisier.

This would indicate that I probably applied a bit of noise reduction to the older scan, so such a comparison is not rigorously scientific. However, the fact that the recent Nikon scan appears very marginally sharper (perhaps not as noticeable on the jpeg) gives me confidence that the scanner is not producing suboptimal results because of possible dust on the lens and mirror.

IMHO this is the way it should be compared and judged.  The whole point is the native capability of the scanners optical system to resolve detail.  Clearly there is a large difference between the two scans shown here and the Epson just doesn’t quite cut the mustard for me as the fine detail has been lost in mush.  Sharpening will not regain lost detail just increase micro contrast and may give an overall impression of sharpness but clearly cannot resolve the fine detail contained in your image.

That's true. Sharpening cannot regain detail that was never captured. However, any detail that appears to have been regained through a sharpening process, must be detail that was originally captured. If in the process of bringing out such detail, one also gets sharpening artifacts and noise on smooth tones, then that's something to take into consideration. Sharpening is a complicated subject in its own right with lots of programs specializing in the task. I don't claim to be an authority or particularly skilled with sharpening techniques. What I've been trying to determine with these experiments and comparisons, is whether or not there is any detail in the Nikon scan which cannot be brought out in the Epson scan, because such detail simply does not exist in the Epson scan.

I am not sure I can agree with the assertion about the differences in the ability of a particular film to resolve detail are higher with the same film stock between 35mm and MF.

Sorry! I wasn't clear on that point. If the emulsion is the same, then of course the resolving power of the film should be the same. I was referring to the captured image. A standard 80mm lens on an MF film camera is unlikely to deliver as sharp results, in terms of lines per mm on the surface of the film, as a good quality 80mm prime lens on 35mm format. However, in terms lines  per picture height, or picture width, the MF lens will win hands down.

As I understand, the 8000 ED uses only one lens to scan both MF and 35mm film. The 35mm film should make more demands on the scanner's lens. Any differences in captured detail between the two scanners should be more apparent on 35mm film.

[TonyW]

Hi Ray,

That's correct, but not quite 13 years ago. Closer to 11.

Thats just me, obviously I lost the ability to do a simple subtraction  .

I've now re-scanned this Kodachrome on the same Nikon scanner. If one is going to pixel-peep, then let's do it properly at 200%.  

Pixel peeping is good in this case as we are looking to see how much information we have managed to capture between two systems.  Judging on screen is problematical (unless screen is the final destination) due to the lack of resolution in the current crop of monitors.  If printing is our final goal then it is a bit of a crap shoot until a print or two is made as it is not possible to judge quality due to screen limitations – at least I find I am unable to make this type of judgement accurately particularly relating to the amount of sharpening for print output.    

I've chosen this particular area of the film because the lady's head scarf brings out the resolution differences in the most obvious way and to the greatest degree. The comparison crops (image 02) show the detail to be very close. In fact, the recent scan, without sharpening or noise reduction applied, appears very marginally sharper than the 11-year old scan, but also very marginally noisier.

This would indicate that I probably applied a bit of noise reduction to the older scan, so such a comparison is not rigorously scientific. However, the fact that the recent Nikon scan appears very marginally sharper (perhaps not as noticeable on the jpeg) gives me confidence that the scanner is not producing suboptimal results because of possible dust on the lens and mirror.

Certainly it is a problem to compare if we are not sure what parameters applied between scans as they would need to be the same to draw meaningful conclusions. I would not be able to remember what I did to an image a month ago never mind a period of years!    In the case of your recent scan the difference are so minor and considering the magnification factor I would be happy with either.

So if you are confident that both the Nikon and Epson are set optimally then based on this recent result there does not appear to be a any good reason to choose one over the other apart from ease and speed of producing the scan.  

[Ray]

So if you are confident that both the Nikon and Epson are set optimally then based on this recent result there does not appear to be a any good reason to choose one over the other apart from ease and speed of producing the scan.  

Hi Tony,
That's more or less what I'm thinking. However, there is a certain satisfaction in doing the best job one can with the available tools. I can't deny that I prefer the 8000 ED scan, which is sharper straight out of the box, so I'll reserve the use of that scanner for the better quality film and slides that I think I might want to print at some stage.

I'm grateful to Ed Hamrick that my original purchase of the Professional Edition of Vuescan, back in 2005, is still good for the latest upgrade of Vuescan, which contains Win 7 drivers for the Nikon 8000 ED. I don't need to fork out 450 Euros for the specific version of Silverfast which is compatible with the 8000 ED. Vuescan seems tremendous value.

A quick calculation of the print size, represented by a 200% crop on my 27" NEC monitor, which has a resolution of 2560x1440 dpi, results in a print size of 8ft x 5.3ft. In other words, to see approximately the same differences on a print that I see in the 200% crops on my monitor, I would need to view an 8ft x 5.3ft print from the same distance that I view my computer monitor. A 100% crop on my monitor would represent a print size of 4 ft x 2.7ft, which is still a bit larger than my printer can handle, unless i were to make a diptych. 

One puzzle that's still in the back of my mind is the quality difference between wet mounted scans and dry mounted scans.

Some of the comparison images I've come across on the internet seem to show differences of an order of magnitiude which are very similar to the differences between unsharpened Nikon 8000 scans and unsharpened Epson V700 scans. However, other comments suggests the differences in detail and sharpness are very small and hardly worth the extra trouble that wet mounting involves.