on landscape The online magazine for landscape photographers

36 Megapixels vs 6×7 Velvia

The disposable sensor shows it's true colours

Responses74
Skip to Comments
Tim Parkin

Tim Parkin

Amateur Photographer who plays with big cameras and film when in between digital photographs.

Flickr, Facebook, Twitter



00 Header Image

Over a year ago now we carried out various tests of medium format digital camera systems and film camera systems. The results, whilst interesting, didn’t tell us a whole lot about 35mm digital camera systems. Just after the D800 and D800E came out we followed these up with a comparison of these cameras, which are still the gold standard for 35mm sensors, and medium format film. With the surge of interest in film we thought it would be good to share the results with a wider audience.

If you've come here from Petapixel - you can read a longer version of the article that appeared there on this page or you can take a look at the original Big Camera Comparison we ran by clicking here. Thanks for visiting!

How to Compare

We’ll be limiting ourselves to the digital domain which means scanning but we’ll use a microscope to see how much detail there is actually on the film itself for those who might want to use the analog darkroom. Here are the possible ways to compare.

  1. count lines - should give us the absolute peak resolution of the film files.
  2. compare results on screen - good for pixel peepers
  3. compare prints and projected pictures - actually gives useful information for the photographer

Counting Lines

We photographed a test scene with various cameras, ensuring focal lengths were very similar and adjusting distance to target where there was a small discrepancy. Where aspect ratios were different we kept the short edge angle of view consistent. As we have a landscape bias, all of the lenses were chosen to be approx 24mm full frame equivalent. (Zeiss Distagon 25mm f/2 and Mamiya 7 50mm) Here’s a photograph of our test setup.

image14

image08

The test sertup as a resolution chart, a couple of very sharp 5x4 transparencies and a few ‘real world’ items (well - real world for photography geeks anyway!)

We used the number of lines we could perceive (regardless of contrast) to calculate number of line pairs per picture height (i.e. half the vertical pixel count). From this we could work out a megapixel equivalent.

Here’s a sample of resolution scale from a few of the scans we used. These scans were from the Howtek 4500 drum scanner (colour slide) and the Screen Cezanne flatbed (black and white)

image12

And here are some of the calculated stats…

table

** beyond resolution of chart.

The IQ180 and the D800E produced the full resolution of the sensor at very high contrast and hence they managed to get the theoretical maximum resolution from their sensors.

The surprising thing here is just how big those results are from the medium and large format cameras! (That should get the trolls out in force!).

These figures are supported by our colleague Henning Serger who has been making extensive tests of real world film and sensor resolution for the past few years. His figures show colour film with up to 135 line pairs per mm, B&W film up to 150 lppmm and finally microfiche films such as Adox CMS20 at up to 260 lppmm!! In his scientific tests the object contrast of the test pattern is 1:4 (two stops). For the comparison tests a Nikon Nikkor 1,8/50 AI-S and a Zeiss Makro-Planar 2/50 ZF are used at f5,6. In this test with both lenses the D800 reached 80-85 lppmm, and the D800E 90-95 lppmm. (at a 1:64 contrast Zeiss measured colour film at up to 170lpmm, black and white at up to 180lpmm and CMS20 at 400lpmm!)

Here is a list of equivalent megapixels for each line pair figure.

90lpmm = 28mp
100lpmm = 35mp
120lpmm = 50mp
140lpmm = 68mp
260lpmm = 235mp!!

This is the data actually on the film (Henning uses a projector too - you still have to get this either enlarged or scanned in and that always loses some resolution. The maximum optical scanning resolution of any scanner I’ve tested is just under 6000dpi and my best scanner peaks at 5300dpi. This gives a maximum possible digital camera equivalent of 38mp. If your scanner peaks at 3500 to 4000dpi (as many do) your maximum equivalent is 19mp to 21mp.

Interestingly this seems to be the sorts of figures that you get bandied about in the ‘better’ forums discussions of the resolution of 35mm film.

If you’re interested in medium format cameras, Henning’s results suggest that you can get within 10% of the figures for 35mm cameras using good lenses (i.e. The Mamiya 7 or Hasselblad lenses). This gives up to 125lpmm for colour slide film and for Adox CMS you can get up to 210lpmm. If you were to work out the resolution on 6x7 film you would get the following

90lpmm = 136mp
100lpmm = 168mp
120lpmm = 242mp
140lpmm = 330mp
210lpmm = 741mp!!

Henning also tested the scanned results using an ICG 370HS and an Imacon X5 and found colour film to have up to 100lpmm and Adox CMS20 up to 130lpmm. Using a Nikon Coolscan colour slide film acheived up to 60lpmm and Adox CMS20 up to 65lpmm. Henning has also tested slide projectors and has acheived up to 125lpmm - so where 4K is about 8Mp, 35mm projected slides can achieve 50-150Mp (more information in a future article).

CAN WE TRUST THE NUMBERS?

Well it turns out that numbers don’t tell the whole story (when do they ever!). Despite being verified by myself and others, the fact that the digital camera give their finest detail at very high contrast and the film cameras give their finest detail at about <10% contrast means that comparisons using numbers don’t accurately affect the perceived image quality.

The graph of the contrast of line pairs against resolution for film vs digital will look something like this..

image16

This is just an illustrative graph to show the idea that digital doesn’t lose contrast in the way that film does. In addition digital can cope with a lot more sharpening because the files are so clean whereas film loses contrast with increased resolution and sharpening is quite often limited by grain. However when digital does lose contrast it drops off dramatically while film is still giving some detail, albeit at lower contrast.

So although on paper it looks like 35mm film should keep up with modern 35mm digital and medium format film should blow 35mm digital away, the actual real world results are somewhat less amazing but still pretty impressive.

COMPARING ON SCREEN

So let’s take a look at the D800E compared with Mamiya 7 using Velvia 50 film and Adox CMS 20 to see just how these figures translate into photographs. First of all here's the test area again (slightly different than above).

00 Overview MFvDIGI

image11

Well I think we can see that the Mamiya 7 in colour is just exceeding the D800E in terms of detail and the CMS20 is substantially better. Normal fine grained black and white film would be somewhere in between (i.e. Delta 100 or T-Max)

Let’s see how well the Mamiya 7 results compare with the IQ180 just to see where they fit in the megapixel race…

image00

This shows that although comparing fine detail shows a fairly close match between the two, in actual fact the IQ180 looks a lot cleaner and sharper. This is a result of the high contrast edges in digital. The Velvia 50 shot is suffering a bit because of underexposure - the dynamic range of the IQ180 blows the transparency film away (the photographs were exposed for the highlights on the lightbox). However the detail on the Velvia shot isn’t doing too badly - small details like the markings on the lens look good but overall the clean, high contrast detail of the IQ180 makes the photograph look clearer.

Here’s another comparison showing the difference between an IQ180 and the Mamiya 7 but this time using Adox CMS 20 film. The Mamiya result blows away the IQ180 in this case and is still a fairly close match even at f/22. This goes to show just how good old lenses actually are (especially the Mamiya rangefinder ones) and on a sideline proves that f/22 isn’t the hellhole that many photographers believe it is (for more about that refer to Roger Cicala’s article about diffraction - http://www.lensrentals.com/blog/2013/03/overcoming-my-fentekaphobia)

iq180-vs-mamiya7-cms20-verticals

Well I think the thing we learned here is that the type of film makes a substantial difference. Adox CMS 20 has it’s own developer but can be stand developed in Rodinal - not something you can send to your local lab. However - one thing it does prove is that medium format lenses are quite capable and have a lot of leeway left for sensors to increase in resolution substantially.

WHAT ABOUT LARGE FORMAT?

We looked at large format (4x5 & 8x10) shots in our previous tests so here are a final couple of comparisons between 4x5 and the IQ180. This first is a section from a transparency that was placed on a lightbox.

05 studio-close-areas

06 slide-samples

From left to right this is the IQ180, 4x5 Velvia and 8x10 Velvia.

Finally, here’s the real world landscape photograph that we took. We’ll show a final couple of comparisons in the very bottom area marked in red and in the area on the left hand side of the screen.

07 real-world-topping-showing-areas

First lets look at a comparison of the small village on the left hand side of the shot in the red square.

08 topping-housesleft-640-iq180-vs-4x5

As you can see it’s a close call. Some things look more refined on the 4x5 Velvia (The garden chair, the car grille) and some things look a lot clearer on the IQ180 (the garage roof and walls). In a print of this section things were ranked pretty closely.

big-camera-comparison-prints

Why are the results from 4x5 not performing as well as indicated by the slide comparison? Well in the real world we had to stop down from our optimum aperture of f/11⅔ to about f/22. This reduced the max resolution of the 4x5 shots. The IQ180 needed stopping down too but that just reduced the contrast at the sensors maximum resolution and with a bit of sharpening it didn’t really do much damage.

WHAT ABOUT COLOUR?

One of the interesting things that cropped up was the way that the digital sensors deal with colour in terms of tonality and resolution. Because only one in four pixels are blue or red, quite often colour resolution is reduced in comparison with luminosity resolution. These look fine at small enlargements but when images are shown larger these artefacts can show through.

Here’s a great example of the problems with having only a few red pixels from an outing a few years ago, comparing a 5Dmk2 with 4x5 Velvia. Obviously the 4x5 Velvia has more resolution but we downsampled it to match the 5Dmk2 and got the following result (show at 200%)

09 berries

On the left we can see all of the berries on the tree but on the right (the 5Dmk2 image) a lot of the berries have disappeared. The only berries shown are where there were larger groups of them (i.e. a single pixel size berry is unlikely to match up with a single red pixel whereas a group of berries covering 2x2 sensor pixels will definitely hit a red filtered one)

Also colour film still seems to differentiate colour differently to digital (depending on the camera). Here’s a sample of our test image from the cyan square at the bottom of the view next to the pond.

10 topping-moss-iq180-vs-4x5Portra400

On top we have a 4x5 Portra 400 scan and on the bottom the IQ180. I think most people will agree that the 4x5 Portra 400 version has much clearer colour differentiation.

SO WHAT ABOUT 35mm FILM?

Well we didn’t include 35mm film in the test - it was originally aimed at comparing 10x8 film with the IQ180. However I had a recent scanning job for a colleague of mine who has taken 35mm film cameras to some of the highest mountains (Alan Hinkes - the UKs first mountaineer to climb the world’s 8000m peaks) in the world and he has allowed me to show a couple of these here include 100% crops. These photographs were taken using a Ricoh GR1 on Fuji or Kodak transparency film. These images make quite respectable 30” by 15” prints. (the first is Fujichrome and the second Kodachrome 64)

11 hinkes-4

12 hinkes-4-crop100

Here’s another

image07

image15

WHAT ABOUT PRINTS?

We made prints of our test charts at various sizes and also of the photograph taken outside in windy conditions and asked people to say which was better quality in terms of detail and sharpness. We asked a mix of photographers and non-photographers.

Interestingly quite often the digital camera photographs would perform better at smaller image sizes but once they reached about 200dpi then the film photographs were often chosen instead (this is true for some of the comparisons between the IQ180 and 4x5 and D800E vs Mamiya 7). From the comments made this was often to do with the ‘plasticky’ nature of digital images once they are enlarged beyond a certain size.

In general people ranked the 4x5 and IQ180 fairly closely, the 4x5 coming out top just a little more often.

They chose the Mamiya 7 files over the D800E files most of the time (although a small but significant number chose the D800E over the Mamiya 7 files - it seemed there was a large factor of flavour preference).

Effectively we saw the following hierarchy.

  • 10x8 (a significant winner still)
  • 4x5 Black and White
  • 4x5 Slide and Colour Neg
  • Mamiya 7 Adox CMS20
  • IQ180 / Mamiya 7 Delta/T-Max 100
  • Mamiya 7 Slide
  • Mamiya 7 Colour Neg
  • D800E

.. AND PROJECTION?

Well digital projection is definitely getting there with 4K and we’ve just run a conference where we used one of Canon’s high end HD (1080p) projectors on a 70ft high screen. The results were very nice indeed but we’re fascinated by Henning Serger’s results with film projection (both 35mm and medium format) so much so that we will be converting a couple of our presentations in next years conference into medium format slides using LVT film writers.

CONCLUSION

Comparing film and digital is a difficult task as they are not designed to record light in the same ways. However I hope this has shown that with the right scans medium format film can compete quite well with digital and even 35mm film can produce decent size prints (enough for book publishing definitely - see Alan Hinkes “8000m”, nearly all shot on a relatively cheap 35mm film camera (Book currently available from Amazon).

Some comments on the results though

- Medium format lenses can resolve impressive amounts of detail. More than enough to cope with digital sensors to increase in resolution at least four fold more more.

- 35mm lenses, even legacy ones (Henning Serger uses an old manual Nikon 50mm in addition to the modern Zeiss lens) still outresolve all digital sensors.

- 4x5 film competes very well with the best that medium format digital has to offer

- in projection, film delivers an outstanding, unsurpassed quality at extremely low costs (both 35mm and 120)

- optical printing of film is also an excellent quality option at very low costs.

SCANNING COSTS

Fortunately I run a drum scanning business so we had unlimited access to high end flatbeds and drum scanners. We scanned the film using various scanners from a cheap Howtek 4500 drum scanner, a Screen Cezanne Elite Pro flatbed, an Aztek Premiere, an ICG 380 and a Heidelberg Primescan D8200. Microscope measurements were also taken using a 80x stereo microscope.

People will obviously say “well that’s fine if you have a drum scanner! What about your average photographer”. Well the beauty of film is that you can send your film off for very high resolution scans at any point in time. For most purposes though something like a Nikon 8000 or a Minolta Dimage Multi-Pro will produce images that will match the best of 35mm digital (with the right camera and lens). An Epson V750 gets very good results with 4x5 and 8x10 (especially with negatives).

Even drum scanners are now affordable for the dedicated film photographer. I purchased my Heidelberg from Karl Hudson who refurbishes, delivers and gives a warranty for his scanners for less than the price of a top end DSLR.

Great enlargers can also be had at amazing prices and projectors that blow away 4K can be had for peanuts.

ISN’T THIS JUST A PRO FILM RANT?

Actually no, well maybe a little. For myself I happily use film and digital and am just about to invest in a Sony mirrorless camera that I will use alongside my 4x5 and 10x8 cameras. The discussion shouldn’t be about film OR digital but film AND digital. They both have so much to offer..



  • Adox CMS20 is a 3 to 6 asa film. Stoped to f16 you need a tripoid in bright sunlight to hold your mamiya. Very usefull indeed ! And who could imagine that a 8×10” has more resolution than a 40x50mm sensor? Seems that your scanning busines is realy down…

    • You’re not going to get the best quality out of digital or film hand holding. Hence why many photographers use sensor plus on the Phase for hand held work (or big strobes).

      Are you asking if my scanning business is “down” as in “with the kids”?

      • Pietro

        Dear Tim, many thanks for this contribution. I think this (Together with the 8×10 film / digital comparison), is an excellent piece of information, also to help a photographer decide which equipment to use for one’s specific work.

        I also find essential, even if not scientific, the print perception experiments, together with alongside comments.

        One thing I did not quite understand though, is why results from color negative film weren’t included. However, you mention one undefined color negative in the general “hierarchy list”. May I know which one is it? I am asking as I always use my Mamiya 7 with the 160 Portra, and would be curious to know, your comments on how this compares to the 100 Provia of the tests. It would have been great also to see how the 51MP Cmos sensor would have compared! In any case your work is excellent.

        Thanks, Pietro

        • Tim Parkin

          Hi Pietro – colour neg film was included in this test (the 4×5 portra scan of the grasses and moss vs the IQ180). In the original test we also tried Fuji Pro160S in 4×5 and 10×8. The resolution was just slightly lower than the Provia but not far off. I would guess that the 50mp CMOS and the Mamiya 7 with Portra 160 would be pretty close with a good scan. Thanks for the feedback!

          • Pietro

            Thank you Tim! I was meant to say: color negative on the 6×7 format (I saw the Portra used on the 4×5 for the comparison).

            • Tim Parkin

              Ah OK – we did try it on the Mamiya 7 but I had problems scanning it. Since then I’ve made a couple more scans on a better scanner (Screen Cezanne Elite Pro). See here for a comparison between the IQ180 and Mamiya 7 Portra 160 – http://static.timparkin.co.uk/static/tmp/portra400_cms_06mic_8000-vs-4000.jpg

              • Pietro

                Thank you Tim! very interesting to see it “live” on the image crops. Did you also shoot a finer grain film (like the 160) or just the 400?

    • John Wayne

      Uhm, no. It’s ISO 20 in overcast conditions and ISO 12 in sunny conditions. Not sure how you made up that 3-6 claim. Also, you’re clearly not a Mamiya 6/7 user. If you were, you’d know that the 75/80/50/65 lenses are extremely sharp, even wide open.

  • km

    Nice comparison. I use two Fuji 6×9’s, one a 65m and the other 90mm. Both do a wonderful job but I do have problems with keeping the film flat for scanning with a Nikon 9000. I usually just go with the glass carrier; however i can still come close when comparing them to my Sony A7R. The colors with film are just spectacular, although sometimes hard to bring out when scanning. Been thinking of a hasselblad X1. I”m wondering if it’s worth 13k? I’m starting to go back to film again,(got into photography in the 70s) so that might be an option. I would like to get a scanner with full tech support, as opposed to trouble shooting everything myself, as well as repairs(!)

    Thanks for posting and showing how wonderful film is!!

  • Jerome

    I think I have already seen these images in an older comparison.

    • Hi Jerome – No you haven’t seen any of the tests here. The D800, IQ180 and Mamiya 7 photographs from the main body of the article were taken after that first test and have only appeared in a forum and the LF comparison shown, whilst it’s the same test as in the Big Camera Comparison has been rescanned.

  • Arup Roy Chowdhury

    Would it be possible to compare Velvia with Sigma Foveon sensors in future, notably the Merrill series or newly released Quattro, would be interesting for sure.

  • Danny Chau

    Very much enjoyed this comparison test, what intrigues me to know is how much improvement if the Nikon D800E is coupled with the best lens such as the Zeiss Otus? On top of this would be the use of HDR or multishot which also improves resolution/dynamic range as it uses neighbour pixels which used on super resolution processes. Any chance to see this in your future articles?

    Although the film out resolves digital in this test, but the format is not exactly the same size, we are now comparing 35mm to 6x7cm in order to have some sense in the comparison. While digital technology is constantly changing and relative to age of silver gelatine technology, digital is still at it’s teenage stage. Even at the present level of development, digital has already out shine conventional film and has allowed photography to access to all people and not only for the few as in the film days. What makes digital win the argument is the weigh, the speed and for people with a more knowledge in digital manipulation, one can shot and stitch panorama photography to offer incredible resolution and quality with a small package which was not possible with film.

    Currently I’m using the Lumix GM5 with fixed lenses and the quality from this tiny camera out perform the Nikon F2 with Tri-X pan from when I was a teenager, I have made 60″ x 40″ print from these little sensors and it will blow many minds of how good they can be. I guess we are in the best time ever as we can choose and play with whatever technology that are available to us and this means more fun for all.

    • I’m not sure it can improve a huge amount as there was already very high contrast between adjacent pixels. The Zeiss Distagon 25 f/2 resolves 3740 lines per picture height on the 5D3 and the Zeiss Makro Planar 50mm resolves 3760. I’m interested in looking at what superresolution can do (if only a curiosity rather than a belief that it’s really useful for landscape photography).

      • eyeofskye

        I can’t offer any results but I recommend you have a go at superresolution. I experimented with PhotoAcute at a time when I was using a Nikon D2X (12Mp) and reckoned that I could achieve the equivalent of 18Mp using about 10 shots in continuous shooting. I’d hardly call it a convenient technique but with the right combination of iso and shutter speeds it gave good results.
        I expect if you used it routinely it might become less tedious.

  • jens g.r. benthien

    Tim, thanks a ton (again) for your comparison!

    I’m still working with film: Fuji 6×9, Plaubel 69w proshift and an Arca Swiss 6×9 with 6×9 roll film back.

    Resolution from scans with a calibrated and profiled Nikon LS 9000 is superb, and if this is not enough, there are drum scan services available.

    Let’s face it: The two Fujis (GW and GSW) are the sharpest, delivering most detail. The Plaubel is a bit softer, but has an enormous angle. And the Arca – well, it lets me use **any** lens with tilt, shift, rise, fall. This is a feature no other system on the digital side can do.

    In short: because prices for photography drop, it’s better to work with film. No customer would pay the required fees for a digital medium format system in this declining economy. Because I’m in the architecture and industry field, nothing can compete with medium format negatives or slides.

    The Nikon D810 is great, but has too many limits for real professional work in my business. And a PhaseOne back with an Arca or Linhof is economical nonsense as long as the investment is as high as it is currently.

  • Jim

    This appears to be another comparison of film size and sensor size that doesn’t make a lot of sense in real world use in many ways.

    Contact printing B&W I get comments about ‘how can I get the same quality as your 8×10 with my smaller format negatives’? Easy, just do contact prints. With smaller formats and the higher quality lenses your contact prints will be even better – just smaller.

    If the images look the way you want you use whatever works for you.

    Now if we could only get ISO 25 or even TechPan range ISO settings with digital without having to resort to Neutral Density filters I would be happy.

  • Clive

    Thanks for the report on your comparison and findings. They are about what I expect, and you save me the trouble. I have only an Epson scanner which gives me results which are about as good as an old enlarger.

    I use both film and digital cameras. I find that to get “contact print” quality, I can magnify film up to about 6x with modern systems. Whether enlarged or scanned doesn’t make much difference. A good digital camera image enlarges to about 8x. After that, it is a compromise of size against quality. I like small, high quality prints.
    The advantage of using a small (35mm camera) is the increased depth of field.
    The advantage of using a bigger camera is that the image enlarges to a bigger print.
    Camera movements sometimes help, in any format.
    For the best image quality, I use a 4×5 BetterLight scanning back with its tri-linear CCD sensor (no interpolation, just clean RGB layers). But, it comes with limits. Wind, moving clouds and moving objects are troublesome. The system also needs batteries, wires and a computer. Keeping everything charged up is a bore.
    For the reliability and convenience of a film camera with perhaps only one small battery, I much prefer using film. For larger formats, one battery in the light meter, is enough! A few film holders, or a roll of film, are not much trouble.
    Digital cameras are convenient for snapshots. Good digital cameras are very expensive, relative to film cameras. Storing digital images is a big ‘ol pain.
    Current films are very good.

  • donghui

    It was very useful article to me, May i ask few questions… through the facebook or Email.

    I’m currently working with 8×10 and i really don’t have a skillful technique to scan my films with v750. (i have anti newton ring glass).

    my scanned film quality is really bad.

    Even though i’m a college photography student, i dont’ have any mentor to ask.

    thanks!

    • Ask away!

      • donghui

        whenever I’m scanning, i put coins on each edge and attach the film with anti newtonring glass.
        but i can only scan up to 1800dpi tiff file.

        is there any tutorial or other way that i could scan 8×10 film with ?750?

        i wish there is a video tutorial…

    • Danny Chau

      I suggest you to try Vuescan software for your V750, here the link with 30 days free trial

      http://www.hamrick.com

      I managed to have plenty 5×7 and 10×8 black and white negative to a very high standard, especially if you select multi pass scan and will offer higher dynamic/density than single pass scan. On top of this, this software will run all your other scanner so as long it is supported for one fee and support/updates for lifetime.

  • Ian Christie

    A nice article. I agree with another comment that it’d be interesting to compare Velvia with Foveon results. I use a Sigma Dp2 Merrill alongside a Mamiya 6. The Merrill easily beats 35mm film in my view, and is not that far behind the Mamiya when I use 400 ISO film.

    Thanks and all the best for Christmas and 2015.

  • Willem

    B&W scans on the Screen Cezanne, which does B&W on 8bits rather than 16bits for colour by its software limitation. Would 16bits have made a difference?

    • I’ve scanned the black and white film on a few scanners, some of which are 16 bit, and it didn’t really make much difference. The Screen Cezanne results were just about the sharpest results.

  • Albert Macfarlane

    I bow before your technical analysis, but your last comparison of prints using as an end result “which has the better quality in terms of detail and sharpness” is frankly embarrassing. You describe what is essentially a social science experiment in perception. However you give no details on how the subjects were chosen (naive v expert), no details on how many subjects were used, no details of how many prints were used, and no description if the prints were ranked by the observers or whether it was a simple choice (“which is best ?”). You also give no numbers and no statistical analysis (“what was the probability that the results differ from those expected by chance alone ?”).
    But what makes the analysis of prints quite worthless is the statement that you made the prints yourself. For a trial with some kind of claim to veracity, each print should have been made by an expert (or accomplished) printer, asked to produce a print with the finest detail and sharpness in the size requested. The origin of each print (camera and film v digital) should have been coded and the experiment run by someone without knowledge of the code. After assembling data, the code would be broken and an analysis produced. This complex arrangement is required because of potential bias introduced by the printmaker.

    • Hi Albert – many thanks for the comment. The print comparison was in no way a scientific comparison. The small ‘test’ I did was more to see if what I could see was in any way reflected in other peoples opinions. The lack of scientific rigour was more to do with time available because the comparisons so far has probably taken at least a solid week of work (if not more) and I couldn’t afford any more.

      As for making the prints myself, you may be entitled to say that I’m biased but as a photographer who uses digital and film on a regular basis I would hope the results would be at least as credible as myself (take that as you will). Because the test lacked any scientific rigour I was in no way going to present the results in a scientific light (I’ve a PhD that including a lot of experimental engineering so I know what a proper analysis looks like). To do so would definitely have been misleading.

      The test was merely a “I wonder what other people will make of these?” comparison. If I had expressed any bias I think I would have painted 5×4 in a much better light. As it was, the innate ‘sharpness’ of the digital files at certain resolutions and the lack of grain (allowing sharpening). Gave a perceptual sharpness to the files that meant the digital quite often beat the film even though the film contained more detail.

      The result in of this in many ways also highlights that perceived sharpness often has little to do with detail – which is why many people have ranked even very small megapixel cameras as having more detail than medium format film. On top of this the sharpening algorithms that need to be applied to film to get them to match digital are not something the average photographer would probably play with.

      And using other printers is a difficult one – everyone has different tastes for sharpness and grain and hence one persons ‘sharp’ would be quite soft in comparison to another’s.

      The work involved in doing a test as you say (and doing it at multiple different sizes) is quite outside the scope of something we can justify.

      I don’t agree that the results are worthless though. They may be worthless to you but each person can take away whatever they want from them. They were definitely worthwhile to me and to quite a few people I have talked to.

      The print tests I made were just as described. An ad-hoc comparison by one person of the way that the different files behave in print. No more, no less.

      • Albert Macfarlane

        Thank you for your thoughtful reply. And congratulations on your PhD. You will therefore be familiar from this quotation from Richard Feynman in a lecture on Cargo Cult Science
        “The first principle (in experimentation) is that you must not fool yourself – and you are the easiest person to fool.”

        • It was a long time ago and nothing related to photography or optics (“drill head pump motor diagnostics using mixed finite element electromagnetic and circuit modelling algorithms” – as boring as it sounds). And yes – you can convince yourself of anything which is why I at least made an effort to show other people the results and their comments were interesting which is why I repeated them. Not more, no less. ;-)

      • Carly

        All information considered it is thought that readers would have the intellect to comprehend what was written under the conditions mentioned and be able to conclude their own findings : )

  • Gordon Burns

    Very interesting article – many thanks for posting it. I have a D800 which I use primarily for photographing artworks (typically oil paintings) in my large format fine art printing business. I use a Nikkor 85mm AF-S f1.8mm lens and often will use the ‘super-resolution’ method to achieve the highest possible resolution I can get. I try to produce a digital file at 360dpi the is the same size as the origin artwork where possible. Often file sizes can be quite substantial as a result. I found the results with the D800 and this lens superb, but your article opens up some interesting thing for me to consider. I doubt I could go the route of using large format film – extended production time factors would be prohibitive. But I wonder what advice you and your readers may have about the best possible lens to use to get the best possible results in terns of resolved detail and dynamic range?
    I print to a Epson 9900 large format printer in a fully colour managed work flow, and do get superb prints (predominantly on matt fine art papers).

    • I think a shift stitch on a good 80mm medium format lens would work very well indeed. Joe has used a mirex tilt shift adapter from Nikon to hasselblad and it gives resolution of about 60mp equivalent and a very nicely resolved and flat field rendering.

    • Danny Chau

      Hi Gordon, I too use a D800E which replaced my trusted BetterLight Super 8k2 scan back for all my artwork capture. The three lens I use is 60mm Macro Nikon, Sigma 50mm Art f1.4 (for larger art pieces at close distant) and Zeiss 135mm f2 (minimum aperture no lower than f11/f13 for best quality), I use large soft box flash multi shot and stitch it together in Photoshop (make sure you process it in RAW with the auto lens profile checked, as PS will automatically iron out the lens distortion and Vignetting), I dare to say the quality I’m getting now is better than the Scan back with Rodenstock digitar lenses. The most important thing in artwork capture is colour management, I use a Xrite SG chart with Qualux Q DC for camera/lens/light custom profile (I made my own reference with Xrite EyeOne Pro for maximum accuracy and it’s far better than Profile Maker 5).

      Qualux link: http://www.qualux.com/eng/t_product_q_dc_eng.php

      The D800E/D810 gives a sharper result than D800 because the removal of the anti alias filter.

      • Thanks Danny. I have been doing this professionally for over 10 years, and of course have a fully colour managed workflow with custom profiled papers and screens. I agree in theory the D800E could possibly show more detail but I think in practice the difference will be hardly noticeable on artwork unless there was a strong linear pattern perhaps. I use Cool Lights with soft boxes rather than flash, and I use polarizing gels over the li9ghts and a circular polarizer on the lens to remove ALL unwanted surface reflections. With stitching multiple images of the paintings where necessary I can capture even the largest of paintings with stunning detail and clarity. The lens I use is great – but read in the earlier comments that older medium format camera lenses could be better maybe. Just wondered if there was a better lens I should be trying. Do you think the new Sigma Art lens would be any better?

        • Danny Chau

          I used to use cool light too when I was using the Scan bank and flash is much simpler and quicker now. I have tried many medium and large format comparison myself and I do believe most of the R&D now goes to 35mm and if you look at tests done at DXO site, 35mm format/lens combination 35mm wins hands down. These new range of Sigma Art lenses are as good as the Zeiss but for 1/3 or 1/4 of the price of Zeiss, highly recommended, they are very sharp for copy work.

          Lens design have improved tremendously in the last couple of years and it’s worth investing to get maximum resolution for reproduction work.

          • Thanks – I did read some excellent reviews on the new Sigma Art lens, but also some sorry reviews where it showed the quality control on these lenses was not wonderful. Some were excellent and some were no good. However worth a look and I will see my local supplier for a closer look.

          • It would be tempting to say that using one of these with a a6000 and a Hasselbled 80mm would produce results that would exceed that of an IQ180 but I haven’t tested this.. I would imagine the fine colour texture rendition would be better too because of oversampling. Thoughts?

            • Danny Chau

              I’m certain that when I use multi shot with high quality 35mm plus best optics will definitely give the medium format a run for it’s money. I have test between a Pentax 645D with 55mm @f8 against Sony A7R with Sony Zeiss 55mm f1.8 @f8 and the result was the Sony has better sharpness.

              Dynamic range and texture rendition is an interesting topic as we could easily add this during our processing, increasing contrast on local contrast will give the feeling of increased depth and sharpness.

      • paulraphael

        Testing by Lumo Labs shows that with optimal processing, there’s really no quality difference with d800 and d800e. The one with the AA filter just needs (and can handle) slightly more sharpening. This is because Nikon did a precise and conservative job specifying that AA filter. It’s very weak. Which is why it’s actually possible to get aliasing with the d800. It’s just less pronounced. http://www.falklumo.com/lumolabs/articles/D800AA/D800AAFilter.html

  • Geoff Goldberg

    Very interesting test, and well done. Its a slippery slope, with lots of variables to be taken into account. What I like about what you did is that you crossed many boundaries, looking for a real-world, practical comparison, and then explained why you did that. Too often, testers get caught up in the methodology for its own sake (or restrictions) and lose sight of why they were doing it at all. You didn’t, so kudos.
    I could use a bit more definition on the different film types, and why each varies so much – but that is a small item. Also, someone once said that film up to about 10x enlargement holds its tonality, and then something else happens. That may be a good rule of thumb, printed directly or scanned. Keep it up, do more!

    • Tim Parkin

      I’m going to be writing a few more articles about film in 2016 so keep your eye out!

  • Mike Herring

    I still love using my 4×5 view camera. I hope I will still be able to purchase film for it in the future.

  • Dave

    I believe the comparison is a flawed. You use a medium format film Fuji Velvia ISO 50 vs Nikon D810 (35mm ISO 64).

    Let reshoot using Nikon F5 with 35mm ISO 100 vs Nikon D810 35mm ISO 100 using the same lens (50mm at F8), and let see the resolution.

    Using a medium format or 4×5 then compare with a 35mm digital is like compare a truck with a family car.

    • Tim Parkin

      It all depends on your criteria for choosing things to compare. For most people the criteria will be 1) cost and 2) size 3) weight. Given these, a small medium format rig (like the Mamiya 7) or a simple 5×4 kit can cost and weigh about the same as a D810 DSLR. I can’t imagine anyone choosing between film and digital saying “They must both be 35mm!” so that comparison doesn’t really help anyone making real world decisions. However for people who feel the need to make a comparison where both shots are taken on 35mm, you could look here http://www.imx.nl/photo/leica/camera/Leica-camera/styled-34/

  • DaveK

    I think these comparisons are great. I now shoot digital because I can’t get 4×5 film developed anywhere in the Pacific Northwest.

    I think this comparison misses the point. When digital first came on the scene everyone wanted to know if it was as good as film. Resolution was the comparison.

    Today I think the question to be asked is whether you can create pixels from film that look as clean / “good” as those from a D810. For example, can you scan a 6×9 (similar aspect ratio) transparency and create 36 megapixels that are as clean – noise free as those from the D810? They obviously won’t have the dynamic range.

    You would have to scan the film at about 2200 ppi. Nothing I’ve ever scanned at 2200 ppi looks as noise-free as my D810.

    • Tim Parkin

      They can’t make it as noise free as digital but noise free enough that you can’t tell in a print is a different matter. As for dynamic range, film still beats digital (if you’re happy to use colour neg). If you scanned 5×4 or 10×8 colour neg at 2200ppi it would look as smooth as your D810 (at print sizes where the D810 image wouldn’t fall apart because of detail). But digital definitely looks smoother than film in areas where there is smooth tones. The curveball here is that many master printers like to add a little grain to digital images.

      • paulraphael

        I’ve started doing adding noise to 60″ prints from a d800. The reason is that the images are so low in noise, areas without texture can look like a plastic, barbie-doll version of the world. Film avoids this problem in big enlargements because the grain tricks the eye into seeing texture. The nice thing with adding digital noise is that you can control it precisely. I don’t try to make it look like film grain; I just make it barely perceptible at close viewing distances. Using monochromatic noise helps.

  • Carly

    Agree it always has been film and digital not one or the other, they can compliment each other, especially in practical use. Still enjoy 500CM with film and digital backs.

  • deus_ex_mamiya

    This is a good, thorough test. What it shows conclusively – but does not state, oddly – is that on an equal-area basis (“per square mm of film or sensor”), digital easily beats film. In fact, “digital trounces film” might be a more accurate description.

    Where film comes out on top in resolution of detail, is purely by using much larger formats/areas than the digital sensors. In that regime – and assuming that slow film can be used – it is competitive or even superlative, both on quality and price (when shot in low volumes).

    So my only criticism of the article is that this key distinction really has to be emphasised:
    * Digital >> film, for equal capture areas
    * Film >= digital, when its capture area is much larger, and is well scanned

    • Tim Parkin

      Absolutely! Although the curve ball is Adox CMS20 black and white which can manage over a 100Mp resolution on a 35mm frame. Very difficult to digitise though (you have to make an enlargement onto 5×4 or 10×8 film and then scan that). Thanks for the summary :-)

  • Nigel Puttick

    Interesting comparison of resolution, which raises the question of dynamic range especially with fine low contrast detail. Could you carry out a comparative test of DR in a similar manner? I suspect the differences at the sensor/emulsion stage would be clearly evident, but hard to demonstrate in a print owing to the restricted DR of any print, whether darkroom or inkjet.

  • Nigel Puttick

    Interesting comparison of resolution, which raises the question of dynamic range especially with fine low contrast detail. Could you carry out a comparative test of DR in a similar manner? I suspect the differences at the sensor/emulsion stage would be clearly evident, but hard to demonstrate in a print owing to the restricted DR of any print, whether darkroom or inkjet.

    • Hi Nigel – not totally sure what you mean as low contrast implies low dynamic range? However I think you mean ‘is film or digital better at low contrast detail’ and I would say that digital is probably better although it depends on the size of the capture medium.

  • paulraphael

    I’ve moved mostly to digital from large format film for a few of reasons. One is that for film to beat digital, you must have very high quality scans, which are expensive and slow to produce. My own results show that the secondary optical system in darkroom enlargement introduces such a quality handicap that it lets my d800 handily beat my 4×5. I believe (as your tests show) that the 4×5 would crush the d800 with a high-res drum scan, but this workflow just isn’t possible for me.

    The other issue is signal-to-noise performance. With most films, once MTF drops to about 5%, the detail is basically gone—it’s too close to the noise floor. But with the better sensors operating at a low ISO, the noise floor is so low that 5% MTF can be judiciously sharpened into strong, natural looking detail. This evens the playing field beyond what the numbers and high-contrast test patterns might suggest.

  • Alien

    Well I don’t know I find the whole thing confusing…I have a D810 and I’m sure I will be happy with it!

  • Alien

    So film and scanner still best…hmmm
    I did see 2 images one taken with a Hassblad 50MP compared with 36MP D810 and I couldn’t pick the difference
    I have 20/20 vission….
    It seems it depends on “how big ” your images will be …as to whether it’s worth going film…Capture one software seems to give great results …on digital images.
    It also depends how much time you are prepared to spend…
    So if blowing up really big images go film;
    ( and scan)… If going to say A2 high end digital should be sufficient… Hmmm

  • Gencho Petkov

    According to standard definitions, the limit of measurement is defined not less than 10% contrast. At lower parameters not measured resolution, because the level of the noise deviation is higher than the values in the table/chart information.
    See table Contrast / LW/PH

    Sample for Sigma 20mm f/1.4 DG HSM | A + Canon 5DS r @ F/2.8 (2 stop) @ 50% contrast

    5403LW/PH = 120 lpmm

    http://www.photozone.de/canon_eos_ff/973-sigma20f14art?start=1

    • That should probably read “According to one of the standard definitions.. ” there are many standards. MTF50, MTF9, MTF5, etc. (MTF10 is an uncommon one – where did you get that from?)

      • Gencho Petkov

        I do not want to fall into meaningless disputes are this is applicable standards from the film era. Just because someone does not suit him and invent their own different percentage deviation = 1%, 3%, 5%, 9% … not my problem.

        Norman Root – a professor of photography developed Imatest

        http://www.normankoren.com/Tutorials/MTFcurve_Velvia_Canon2870.gif

        • Norman is a acquaintance of mine (his surname is Koren, not root). He gifted me a copy of Imatest Master as part of the analysis of these results (and other work). And… Oh look! A page in Imatest master where you can define your own %ge!

  • Dennis W

    Thank you for the article, I thoroughly enjoyed reading it and agree with your conclusions. The only thing that I disagree with is your statement that all 35mm lenses our resolve digital sensors. If that were the case then we would not see an appreciable gain in resolution from lens to lens. I agree that you can get some older lenses that are incredibly sharp, but I can’t believe that the Nikon 50 mm is as sharp as the Zeiss 50 mm. My personal testing tells me otherwise.

    • I would have to agree with you, though most lenses are fairly equal at f/5.6, wider than that is what separates the men from the boys. There’s a reason that Nikon put out a list of lenses that would “perform best” on the D800 when it was released, a lot of their lenses were out-resolved by the new sensor.

    • Hi Dennis – sorry for the late reply. All lenses are rated at a certain contrast level, typically 50% contrast between pixels. This resolution limit is what you’ll typically see in magazines etc. However, if you place the limit at 5% contrast then the resolution is much, much higher. What you see when you use a better lens is typically an increase in contrast rather than an increase in resolution. The corners of lenses are a different matter …

  • This is a great article, thank you for going to all the trouble of this test!

    For me, it’s the colour resolution and subtle gradations of film that keeps me shooting it for important projects. Even a 12MP camera can produce large prints, but to recreate subtle shifts in colour with digital, you either need to have a really high pixel density and basically “oversample” your image (print smaller than what your maximum would be at 300dpi) and preferably a camera that produces 16bit files. Neither method is a good as medium or large format film, in my opinion.

    The Sigma Foveon cameras are interesting in that they can reproduce more subtle colour variances than bayer cameras, but it’s difficult to get pleasing skin tone out of them.

  • Alexander Gansky

    Which scanner are you user to scan all of films?

    • Heidelberg Primescan 8400 and also a Screen Cezanne Elite Pro.

  • Low SK

    Try using something like Docter Optic, Apo-Germinar 240mm f/9 for your 4×5.

  • I love to use film for my own use, I’m not bothered about which captures the most pimples or blades of grass, digital or Analoug. But scanned colour negative like Portra is just beautiful to look at and that counts for more than any charts or graphs can say.

    • You and I know that but some people want to know the other capabilities too. Actually I need to know them because I wouldn’t want to use 35mm film to capture the sort of images I want and I only know that through my own testing.

      • I also think the more digital progresses the more the character of film comes out. Film also has strict limits which are good, only digital produces truly awful HDR which is becoming the norm. Its this desire to see every detail that has shadow and highlights trying to compete with midtones

  • Scott Hays

    It was nice to see a complete (as complete as I have found) comparison on the subject.

    One of the reasons I came looking was after seeing Fuji’s 100mp camera at the cost of around $48K. Now, I primarily shoot film and have for about 40 years now. I do shoot digital, but having shot film that long I am still capable of seeing the difference.

    When Canon first came out with their Mark I, and it stayed the powerhouse of the industry for years, people kept asking why when it continued to have a lower mp rating than other cameras. Well; besides the fact the thing was indestructible, what people weren’t looking at was that the actual megabytes it put out were much higher than cameras with a higher mp. Meaning, it doesn’t matter how large your mp is if your mb is low. Now: if you are never going to make a print, or you are printing small, your mb’s don’t really matter.

    But here we are again talking mp. I still print primarily in the darkroom for my b&w. That is what b&w film is designed to do. That is where you get your true look of depth in your print. However, with that said: when I do scan my 4×5 or 120 and work in Photoshop, I don’t think there has ever been a time where I have been concerned with my overall mp size. However, when I am creating my overall print size, etc… the thing I am looking at is my mb size when I look at my image size box. Now, this is going to be a stretch and just an example, so bear with me. If I make my print size a 20×24 and for some odd reason I end up with a mb size of 6.5; I automatically know that print isn’t going to look worth crap. Don’t care how many mp I started out with. However if I start out with a mb size on that print of 85 mb, I’m pretty confident that I am going to be working with a decent print straight out of the bag.

    The mp war has been the greatest marketing tool ever developed. I think the mp/mb thing have caught up to each other by now. However, if you have a camera that says 21 mp, and you aren’t shooting RAW or at least the highest level of JPG, you are NOT getting that 21mp. No where close most of the time, which automatically reduces that mb for your print size.

    I scan in a 4×5 at say 400 dpi and I am consistently coming in at 80 mp befor I have to do anything else. And I’m not using a top of the line scanner.

    Hopefully more people will start looking at the mb on their screens in Photoshop if they are scanning and be less concerned with mp. There is a HUGE difference in quality.

On Landscape is part of Landscape Media Limited , a company registered in England and Wales . Registered Number: 07120795. Registered Office: 1, Clarke Hall Farm, Aberford Road, WF1 4AL