Monday 14 June 2010

Nikon, Noritsu, Epson: a comparison of scans and prints

well ... and a little more

Recently I decided to photograph parts of a wedding using my 35mm EOS and negative film. I thought about it and was having trouble deciding if I should go with 4/3 digital or 35mm film but decided on the day (on the spot) to do the main ceremony with my 35mm EOS after taking some test shots and feeling that digital was going to be challenged with the lighting.

For the impatient, the findings


In a nutshell:
  • Scans of 35mm negative perform much better than digital in the harsh light of full daylight when you expose for the shadows and let the film compress the highlights
  • for zero effort the Noritsu scans at develop time give dam good results, to the point of making other systems nearly pointless (make this definitely pointless if you're not a scanner nerd or a skilled / professional scanner operator)
  • Epson flatbed scanners (like 4870, 4990 V700) do really quite well on prints up to 8x12 if you know what your doing
  • if you want to get everything out of your negative you need a good scanner such as the LS-4000. When you do you'll be amazed what is in 35mm negatives (and wonder if you really do use that 5D MkII enough to justify spending AUD$3000 on it?)
  • I love a full frame 35mm size capture for shallow DoF
back to the details

The lighting was under gum tree shaded outdoors, at 11:00am ... so yes, stark shadows and bright bright high lights = blown whites and blacks that resemble inks from Satans bottom. I just wasn't going to risk it.

Now before anyone says fill flash, I'll remind you that on a sunny 16 sort of day (or even f11) that as many cameras are limited to 250th of a second with a flash it would mean that I'd be using f8 on the lens if I wanted to do fill flash (even with 100ISO).

f8 might be good for scenery but on 4/3 it spells sharp not diffuse backgrounds (and pretty clear ones on 35mm too).

YUK ... I may as well use a iPhone or compact digicam. So if you desire shallow depth of field then you can't use fill flash with a focal plane shutter camera ... Even 500th (on a lens shutter) will still only bring me to f5.6 so its available light or sharper backgrounds

Using available light and large aperture was of course perfect in that situation as it gives shallow depth of field, diffuse backgrounds and fast shutters to freeze action / counter camera shake

Now one of the reasons I entertained doing this wedding in Negative was that a local photo place (Photocontinental in Brisbane) uses a Noritsu scanner in their minilab. Readers of my blog will perhaps recall that I'm very impressed with the Noritsu scan quality. I like the fact that I get 3087 x 2048 Pixels ( thats about 6 MPixels) images which are certainly good enough for prints up to 8x12 inch straight from the box. No cleaning, no time spent in scanning, no hassle.

I thought this was an excellent opportunity to:
  • sus out how good PhotoContinentals quality of service is with the Noritsu
  • save my self time in scanning only the potentials for big enlargement
  • have everything on CD for everyone with no effort
This is where I hit my first hurdle, you see when I called and asked about the service I asked if they (Photocont) could do the high res scans. This is were I hit my first hurdle: the staff had no clue about their stuff or how to drive it.

After sorting that out (why is it so hard?) my next hurdle was they neglected to mention that the price for the high res scan was higher than the low res scan ... $9 vs $25 Noone can give me a valid justification of why this is, but I digress.

$25 is of course verging on the highest prices you can pay for this sort of service ... $9 is a good price and one I'm happy to add to the $4.50 for develop of the film, but $25 (plus the $4.5 for develop) pushes the friendship and makes 4 rolls of film start to feel expensive on a job. I'm now well over $100 for materials alone, which is $1.22 per image. Man, no wonder photographers are leaving film for digital.

Even if I want to use film its just put my costs up way higher and that's forgetting about the film stock costs...

My next hurdle was that the files contained no colour profile information, a phone call reveals the futility of asking as noone knows what it is I'm talking about. I assume *(perhaps falsely) that its sRGB. None the matter ... they look OK on my monitor ... a little red for my taste.

Its also worth mentioning to anyone reading this that the standard reply from PhotoContinental to any question about their quality is met with "we don't offer a professional service"

Right ...

Don't get me wrong ... the staff are often very helpful and very professional, its a management thing. If you want my opinion ... don't go to PhotoContinental for printing or development services. I mean their price structure is nearly exactly that of Prolab's and they do offer a professional service. So while PhotoContinental's staff are great, helpful and friendly, their corporate strategy has them charging professional rates but backing down on any responsibility with "we don't offer a professional service"


comparing the scans


With the images in my hand (on the CD actually) I thought I would see what I get with an Epson 4870 (my 4990 is in a box in bubble wrap from the trip over) and my Nikon LS-4000 (broke that one out of packing).

This is the image I picked.

I liked it because: the brides got lovely bright metalic bits and white in full sunlight, the groom has dark hair and he's in shadow, some parts of skin tones are in full sunlight others are in shadow, it shows the DoF of a 100mm lens at f2.8 (hah) and the image was as sharp as a razor.

Hard to ask for more to start with.

I noticed that the scans from the Noritsu had clipped high lights in the white and the blacks were a little dark and inky ... nothing dreadful, but even if I wasn't inclined to compare before I was now.

Naturally all overviews will look identical (colour balance not withstanding).

Procedure:
I scanned the negative as positive on both my LS-4000 and on the 4870. I followed my normal colour negative methods on both scanners and got quite close results. Here they are:

Epson 4870
4870-3
I scanned this at 2400dpi as I don't really feel that there is more to be had from this unit and this still works out at 3344 x 2252 pixels which is more than the Noritsu.

next is the:

Nikon LS-4000
LS4K-3b
I scanned this one at 4000dpi which is a significantly bigger hunk of data than either of the other two at 5648 x 3656 pixels.


Pixel peeping

Now, noone is going to be surprised when I say that the Nikon got the best results



followed closely by the Noritsu ...

they have a distinctive look to them which I'm starting to be able to identify. Its a kind of sharpening that looks like pointillism (if that makes sence).

and a reasonable result from the Epson (some thoughtful sharpening has been applied, but more care will fix much of this)


Now, look carefully at:
  • the sharpness of the star
  • the clarity of the parts of the necklace
  • the handling of whites (yes, there was metallic thread in the white cloth)

Ok ... so the Nikon LS-4000 shows how well you can do there, but if you're not printing big then you will certainly not benefit from the Nikon ... you may as well save the thousand or so bucks on it and just go with the Noritsu scans (if you can find some place that does not charge like a wounded bull for the scans). Either that or only pull it out for the ones you like. You'd be ahead on the time involved for scanning.

Of course one also needs to compare prints (hard to do in the internet) but for my own personal interest I thought that I'd also get some prints done.

so lets compare the prints


I got 8x12's done of the three above. I also printed two of them to 12x18 inches but thought I'd restrict myself to the Nikon (which of course does that at a print density of 300dpi) and see how well the Noritsu scales up to that size (essentially dropping the print density to 180dpi ... which should be tolerable).

One of the first things worth mentioning is the reaction of the person doing the prints when I picked them up. Naturally with three prints of the same scene she at first assumed when she pulled them off the machine that there were three copies from the same file. However (being used to checking the images) she noticed there were subtle differences and wondered "what the bloody hell is wrong with the printer". Our discussion at collection time was interesting, as she only noticed subtle colour differences.




Each of the three 8x12's look good, and in isolation each will be considered as good.

There are of course colour differences, as I did not balance each to one another. Partly this was because I wanted to see what the results of taking the image sourced from the PhotoContinental Noitsu process was when taken to another service (KMart in this case).

Close inspection revealed that the print operator (and I) thought that the noritsu and the nikon were equal in sharpness, but the whites and blacks were handled better by the Nikon. The Espon looked good and handled the whites and black well too, but was a little soft.

I felt that the screen results (my screen has been tweaked by a Spyder) reflected the results in the prints, that is to say:
  • the prints from the Noritsu system were a little strong in the red
  • the prints from my 4870 were a little soft when inspected closely and compared against the other two
  • the prints from the Nikon were great.

Looking at the 12x18's the first thing I noticed was that Noritsu looked pretty darn good at that size; no real trace of break down. This leads me to wonder what the real native printing resolution of the printer is ... is it more like 180dpi?

As expected the blown high lights in the Noritsu did indeed result in lessened perception of texture in the blouse. Here is a photo of the prints taken with my digital camera (same height on a tripod).


This of course shows how difficult it is to photograph prints! There is not this sort of colour cast in the prints and I was too lazy to obsess over lighting and copy standing the prints (as I'm not trying to show colour here anyway). Perhaps it shows what I mean well enough.

Forgetting (if you can) the issues of colour happening in photographing a print in my living room using copy stand and a fluro light, you can see from this close up that detail is pretty darn good on the 12 x 18 inch prints

Here is a much closer photograph of the prints:

the Noritsu


and the Nikon LS-4K

Essentially either of these 12x18 prints is sharp enough to actually reveal more detail (such as in the blouse and in the green pendant) than was in the 8x12 and be satisfactory to hold 20cm from your nose and inspect it.

Examining the prints shows that the Nikon sourced file looked great in the shadows (the grooms hair was detailed and held dark better than it appears on the monitor) and tonal range was great. The Noritsu files showed the heavy handed treatment of shadows and clipping of the high lights, exactly what you'd choose negative for.

Don't get me wrong, its not bad, the Noritsu is enough to satisfy 90% of needs and perhaps 90% of professional portrait or wedding photographers needs. If only they could tweak that output a little better it would get stellar results from it with the 35mm negative.

I feel that the operators of the Noritsu should (for their $25) put that little more effort into tweaking the high and low ends of the data sourced from the negative and then apply a little curve to that to push down the noise while keeping the signal (anyone remember dbx noise reduction?). After all ... its in the subtleties that we find the difference between hand made wooden furniture and IKEA.

Summary


This exercise has shown me that the Noritsu scanning system gives results which require a very high end desktop or professional scanning system to beat. I have found that digitisation systems allow me to get images from my 35mm film which are quite satisfactory and it remains to be seen if a digital camera such as a 5D will produce better results in the same challenging lighting situation. I have found before that digital makes it easy to get super results when the light is within the range of the capture, but with hard sunny outdoor light digital does not do as well as negative (especially black and white).

The Noritsu system allows film photographers to digitize their images and gain access to the range of services (like printing) which is available to the photographer. Using this systems allows Photographers to leaverage their 35mm outfits (very competitive prices) and get good results, I believe better in some situations than digital ... certainly better bang for buck.

But it has shown me again that it is all let down by the distribution chain. Photocontinental price their scanning service at nearly exactly the same price as the high end professional services such as Prolab do. Yet despite this pricing do not have a clue about their process, do not let you know the colour profile and appear not to know what this means. I have yet to print their file with them to see if the colours look better "in house", perhaps they have tuned their system scanner and printer by eye?

People in the industry have no idea on how to communicate what their service is and what they offer. When you ask about the scan you get useless answers like we scan at 300ppi ... really ... you start with a 1.5 inch negative and give me 3087 pixels from that ... so how is that 300 points per inch?

Something I have not discussed is the scratches on my negatives (nothing major, but without the ICE on the LS-4K fixing after scanning will be a pain) which if questioned would be answered with "we don't offer a professional service"

Again I'm left feeling that *(as in 2001) 35mm cameras with colour negative film will give results which are equaled by only the best full frame (and larger) digital cameras but at a cost which involves a much higher up front investment than using your 35mm does.

If you used your pocket digital for the snaps and your 35mm film camera for the weddings and sporting events how often will you really use it? If you are buying a $3000 DSLR kit will you be ahead on costs after 4 years? I'm betting most will still drag along that pocket camera too.

However the entire film concept has been sadly let down (still) by the industry which hopes to profit from it the most ... if the pricing of stuff like Noritsu scanning was $9 then at $14.50 (scan + develop only) per roll of 36 it would be very attractive to be using negative and getting scans like this done. Certainly it would fulfill 90% of people's needs for the high end ... heck even 90% of wedding photographers.

Given that fewer and fewer clients of wedding photographers see or appreciate the difference, I feel that to use film today you have to be aware of what makes it better, able to get at this yourself and have a client that appreciates the difference.

15 comments:

Lens Bubble said...

Here in Toronto and around, film processed and scanned at local film development shops are only done at 180dpi and has a resolution of somewhere around 1280x768 (I could be wrong on the exact res but close enough). At this resolution, it's enough to print an 8x10 with reasonable detail. Just don't know if you can request for a higher res scan. I am sure it can be done at a pro shop, but places like Costco might not.

Lens Bubble said...

Also, developing and scanning a 36 frame roll at Costco is around $5CAD.

Joshie boy said...

Interesting that you mention prolab for offering a professional service. I enquired about what developers they use for b&w - for which a "technician" had to be called. They use HC-110 for the lot. Hardly pro service in that regard. No complaints on the other stuff they offer though, aside from the price that is!

obakesan said...

yes, well ... professional by self declaration and pricing. These are the guys who rooted up 4 rolls of HIE 35mm for me and tried to suggest that the pinholes in my negs were from something other than bubbles created by their inadequate agitation during development.

no love there from me that's for sure!

obakesan said...

Joshie ... I can point you in the direction of some other Pug fans in QLD if you are interested

Noons said...

"But it has shown me again that it is all let down by the distribution chain".

Ah, OK: nothing has changed much since the days of film only, eh? ;)

Recently I pulled out a big A3 optical print from one of the 80's slides. It's been sitting in an album, no colour deterioration. I've now scanned the slide with the 9000 and printed it at the best my HP8700 can do, in A3.

It is amazing how much better the scan/print is! Not in colour range, or contrast. But in resolution. The optical print was obviously done with a non-flat film holder, in that the corners are clearly OOF. While the digital has corners nearly as sharp as the centre!

What's worse: if I hadn't done the comparison myself, I would never have clued on to how bad the optical print was! A-B comparisons are always more detailed.

Very interesting what you found out with the Noritsu. Indeed they are good value, when handled by folks who know what they're doing.

Problem continues to be that in this day and age of 8MP mobile phone cameras, no one cares about quality...

obakesan said...

Good Morning Noons

>Ah, OK: nothing has changed much since the days of film only, eh? ;)

well yes and no, back then custom colour prints were done by hand using enlargers, noone needed to know about anything more than magnification and fiddling with filtration. I've got an 8x12 of a custom print here from 35mm which still looks better than anything I've had done from scanning the same sheet of film ... and he worked with an interneg.


The technical additions of all the stuff like dpi and colour space seem to have never penetrated the heads of the professional printing areas. That makes it much harder for scans in one place and prints in another.

Just like in the case of fully optical printing getting colour and contrast right is still a matter for the operator. The difference is that now there is another stage between the printing technician and the print ... a computer (additional technical issues) and the proper following of the colour management chain.

The printing technician becomes the scanning operator and now needs to work more abstractly and trust the tuning of that system (rather than rely on looking at the finished product. If that technician is you (as it more likely is) and you are not the operator of the printing system then much greater reliance needs to be placed in the the colour management chain.

The sales people need to have some clue as to what they're doing, especially when they're offering the scanning service.

Personally I'm in no doubt that scans of film and digital prints produce sharper results in almost all situations but I continue to ask the question of why is this relatively simple stuff (colour space / DPI) so bloody hard to get into the heads of professionals.

Christopher Layne said...

Personally I do not believe the majority of operators to really be professionals, honestly. Basic usage? Sure. Can't understand a color-space issue? Not a professional.

With regards to the optical A3 print - it should be mentioned that the scan/print option is not inherently sharper over an optical enlargement one bit. Quality enlarger lenses are quite high resolution and will easily handle anything thrown at them (4x5 and larger without issue) - but they must be used effectively and with knowledge in hand (back to that "professional" issue again).

Presuming your optical enlargement was done by a lab and based on the fact that speed is usually their number 1 goal, they're usually enlarging in a semi-automated fashion using lenses wide-open (as it prints fastest) then flying through the RA-4 process afterward and that's the print. Of course it should be mentioned that some things like local adjustments, contrast, etc. can be easier to do off a scan than under the enlarger (RA-4 specifically). This is another reason labs stopped using it.

It doesn't take much more than basic knowledge to know that a "custom" print will always look better than this as the impetus of the printer is not on speed, but instead quality. For that quality comes price, and of course: time.

In short, optical enlargements are slow for labs - but best for quality when the neg is decent. It's sad to see so very little people doing real RA-4 optical enlarging. However, there still are people around doing it - so it's not totally dead by any measure.

Personally I mainly do black and white with nearly all negatives being traditionally enlarged and prints scanned with a flatbed scanner. Occasionally I scan some C-41 with my LS-5000. However my time hassle at the computer is much less interesting than when I'm physically creating prints in the darkroom. Even my test prints or mistakes are useful for later learning from or testing toners. But honestly, getting away from the computer has been the most liberating part.

obakesan said...

Hi Christopher

I agree with your points about businesses and time constraints. I find that this also holds true in many areas of custom services including manufacture. Speaking of colour, the prints we used to get commonly from the mini-labs are indeed not hard to better and I agree with all your points there too. The advantage for us to scan and FTP to a printer service is that we can access better equipment than we could have otherwise had access to and we don't need to worry about queuing for the access or transport.

On the subject of black and white I've spent some time with very experienced printers working on prints from 4x5 negs in the traditional ways (even pyrogallol development of negs) and found that the Tri-X 4x5 neg (took many of same subject) that was scanned by me and printed on inkjet (carbon on cotton rag) looked every bit as good as the best from enlarger (and quite a bit better than many of the failures).

That is of course after the benefit of much learning and analysis on the computing side ... so both are essentially made by a skilled technician with care time and effort.

thanks for your comments :-)

Photo–Smith said...

Let me see if I can help with some misconceptions:

"People in the industry have no idea on how to communicate what their service is and what they offer. When you ask about the scan you get useless answers like we scan at 300ppi ... really ... you start with a 1.5 inch negative and give me 3087 pixels from that ... so how is that 300 points per inch?"

What the Lab tech is trying to tell you is that his machine prints at 300dpi, that is the native resolution of the laser writer.
So when the operator scans he selects the final print size say 6x4" the machine then scans at the required ppi to match that size which will be approx 1800x1200
If you ask for larger print sizes say 8x12" the operator will go into a different mode to select the 8' wide paper, the resulting scan will be 3600x2400 and will take a lot longer to scan.
This is why higher resolution scans cost more money, it means the operator will have to select a larger scan size and this will give longer scan times.

If you have your originals scanned at a lower res and later ask for 12' prints from the CD the machine will up-res in its RIP giving softer looking prints.

Second point
No machine Printer anywhere has a gamut that is the same as sRGB they will be similar but never the same, in fact different paper surfaces will have different gamuts.
A decent prolab will supply you profiles for each paper surface that you can use for soft proofing in Photoshop.
If not you can download profiles from other labs for the machine your lab uses from here:
http://www.drycreekphoto.com/

I can provide you with gamuts of our Agfa Dlab but that probably wont help.
But if you look you can see a comparison in Apple colorsync (the sRGB is the grey net the Agfa the colour)
http://www.pbase.com/mark_antony/image/126870379

Se how different they are?

obakesan said...

Hi

>Let me see if I can help with some misconceptions:


ok ... go for it.

>What the Lab tech is trying to tell you is that his machine prints at 300dpi, that is the native resolution of the laser writer.

then why didn't he say that? Especially when I asked about "scanning"

why also was he (and his supervisor and the technician) unable to comprehend the concept you just offered, particularly when I suggested that this was what they were probably meaning?


>If you have your originals scanned at a lower res and later ask for 12'
>prints from the CD the machine will up-res in its RIP giving softer
>looking prints.

of course, but did you notice when I upresed in photoshop it didn't turn out half bad?


>Second point No machine Printer anywhere has a gamut that is the same as
>sRGB they will be similar but never the same, in fact different paper

sure, where did I say they did?

>surfaces will have different gamuts. A decent prolab will supply you
>profiles for each paper surface that you can use for soft proofing in
>Photoshop. If not you can download profiles from other labs for the
>machine your lab uses from here: http://www.drycreekphoto.com/
>

which does nothing to explain why the lab had no idea what a profile was.


>Se how different they are?

of course, but I don't see how you've cleared up any misconceptions so much as perhaps had your nerve touched as if I was somehow criticizing you in particular

moosekaka said...

hi, i am confused do you have the nikon coolscan 4000 or the coolscan iv? you had a review somewhere where you compared a nikon coolscan iv to an epson 4990 i think.

interested because i just bought a used coolscan iv from ebay.

obakesan said...

Hi

I had all three, but now I only have the LS-4000 ... my personal feeling about the LS-IV was that it is that most of the time it wa barely sharper than my Epson. Focus was not as precise as the 4000 got me.

In this article
http://cjeastwd.blogspot.com/2009/03/nikon-iv-ed-vs-epson-flatbed.html

I was using a LS-IV ED, but in this comparison (this post itself) I'm using the 4000. I sold the IV-ED to finance the 4000 because the 4000 was better value for my needs.

Anonymous said...

What is your opinion of Fuji Frontier film scans? All minilabs where I live use those machines.

Would a consumer scanner for home use produce worse or better results?

(They scan 36 exposures at a resolution of 1800x1200 for a price of only $2 USD, which is good enough for quality 4R prints.)

obakesan said...

Paul

so far I have not had a Frontier scan done, but when my dad got some done they were scanned at something piddly like 1800 x 1200 pixels ... this was only sufficient for 6x4 sized prints.

I've been talking to the local lab here who has a frontier, but getting them to operate it at high res is tricky. Its an interesting point and I think it'd be a good comparison for a blog post :-)