Film vs Digital - Color Rendition

danielleetaylor · March 28, 2011

I would love to see these claims verified by someone with no axe to grind in either camp. I've seen prints from 8 MP cameras that out-resolve any of the tens of thousands of 35 mm slides I've taken over the years, or the hundreds that I've had scanned and printed professionally. Are my anecdotal tests inferior to yours? And how exactly does one measure MP with a microscope?

There are huge variables at play. One of the most critical yet often over looked factors is that film's resolution depends upon the contrast of the detail being resolved. The resolution of digital sensors varies much less with detail contrast. So 35mm Velvia 50 might out resolve a particular DSLR when shooting a B&W line chart, but not when shooting a real landscape.

Pixel peeping a test chart or studying a slide under a microscope also emphasizes small differences in detail which are completely missed in a real print. Resolution at MTF10 is most important when counting line pairs under pixel peeping conditions. But in a real print the resolution at MTF50 is much more important to the viewer's perception and enjoyment.

Now add in all the other massive variables: lenses, technique, subject matter, lighting, scanning equipment, film flatness, RAW converters, post processing choices, printers, etc. It's easy to see how personal results can be all over the place.

I've gotten into long discussions about this before on photo.net, complete with endlessly studied and debated samples. In my opinion 35mm Velvia 50 on an Imacon is roughly equivalent to 16-18 MP processed using ACR on high contrast detail, and falls somewhere between 12-15 MP on low contrast detail. (Note that I listed scanner and RAW converter as they are very significant.) Velvia is the best color emulsion available today by a significant margin when it comes to resolution. I would rank just about everything else considerably lower, at least in 35mm size.

danielleetaylor · March 28, 2011

Looking at Mauro's crops...

* I don't think anything gave an ideal rendition of the scenes.

* The 40D did OK with all of the scenes. Velvia 50 did OK with the sunlit flowers but completely blew the studio shots. Ektar gave a weird rendition of the sunlit flowers but did OK in the studio. (Which surprises me. I would have expected the opposite.) Portra is too cyan across the board.

* TMAX was excessively contrasty for the sunlit scene and even for the portrait (the shirt).

That leaves me with the following question which may throw fuel on the thread: what is the purpose in trying to match the digital shot to any of the film shots given that none of them gave a particularly pleasing or ideal rendering of the scenes?

Finally seeing the full set of crops just reminded me of how lucky we are to live in the digital age regardless of whether our shots start on film or digital. We have incredibly precise control over color, tone, and contrast in our images. More control in color than even expert darkroom printers had with B&W in the past. I don't know about any of you, but I have become spoiled by this because looking at any of the crops I can't help but think "this can be improved with some PS work."

Great color rendition, or great B&W rendition, does not usually come from a particular film or film in general as opposed to digital. (Though sometimes you get lucky and the scene seems to magically match the response characteristics of whatever you're using.) It comes from the work of the photographer in lighting; film choice and color filtration when film is being used; RAW conversion when digital is being used; and final post work before or during printing, which can sometimes be extensive whether it's done digitally or in the darkroom. Digital just affords more freedom and precision, even if you're starting with film as you can scan and then have access to most of the available digital tools. (I say most because RAW converters can do some things that are near impossible to duplicate with PS alone, namely in color temperature and tint.)

Maybe a better contest would be to photograph 3 interesting scenes and ask everyone to submit their most pleasing or creative renditions from the RAW or film samples.

rishij · March 28, 2011

Very good points Daniel. I agree with your resolution assessments of Velvia 50 film scanned on an Imacon. Actually for low contrast (1:1.6) material, Fuji themselves rate Velvia 50 to resolve ~5.5MP of information for a 35mm frame of film. So Fuji Velvia's resolution translated to megapixel count using Fuji's own ratings leads to a resolution of anything from 5.5MP (low contrast) to 22MP (high contrast).

Interestingly enough, there was a study done quoted in 'Basic Photographic Materials & Processes' where they sent the same piece of film shot of a resolution test chart to 6 different companies & had them assess the limiting resolution in lines/mm. There was a variance of up to 23% in reported resolution for high contrast targets & up to 40% in reported resolution of low contrast targets (p. 269, 2nd edition).

Hence, as Daniel pointed, there are many sources of variance in personal evaluation. A considerable amount of subjectivity is inevitable. Furthermore, you have to use a really good scanner (IMHO an Imacon or a LS-9000 or Minolta DSE 5400 w/ some modifications). I'm not even gonna mention drum scanning b/c it's too cost prohibitive.

Scott Ferris, FYI, the way you get MP ratings for film is by assuming that a line pair resolved is the equivalent of 2 pixels resolved. Hence, if you take Fuji's rating of 160 lines/mm resolved by Velvia film for high contrast subjects, & you take the dimensions of the film (24mm x 36mm), you get:

[160x24] x [160x36] = 22,118,400.

Or, 22 megapixels.

If you look at my scan of one of Mauro's test shots of a target on Velvia 50 film, you can see that lines are resolved down to about 4.25 on my Imacon 848 scan (full resolution file <a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/MinoltaDSE5400vsImacon848.jpg">here</a>):

Note these images are shown at 400% for clarity.

I actually think I'm being somewhat unforgiving by saying the limiting resolution here is 4.25... in fact, viewed under a light microscope, I can clearly see the lines resolved down to 4.75. But let's just stick with 4.25 for argument's sake.

The value read out multiplied by 100 gives you the lines resolved per picture height. The height of this picture is only 1/8.6 the height of the full shot (measure it in Photoshop if you don't believe me... here's the <a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/FullShotofMauroTestChart.jpg">full shot</a>).

Hence, we have:

[4.25 x 100 x 8.6] = 3655

Since 35mm Velvia has a 3:2 aspect ratio, our megapixel equivalent is:

3655 x 3655x1.5 = 20,038,538

Or 20MP.

Let's compare that to the same target shot on a Canon 5D Mark II by dpreview (view full-resolution file <a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/Canon5DMII-ACR-LimitingResolution.jpg">here</a>):

Note that the above is shown at 200% for clarity.

Dpreview shoots these charts so the full picture takes up the full vertical field of view, so we just multiply the limiting number by 100 to get the number of pixels resolved along that axis. I'm gonna say we can resolve the lines down to 35.5. So here's our calculation:

[35.5x100] x [35.5x100]*1.5 = 18,903,750

Or 18.9MP. I'd imagine color resolution for certain colors might be worse.

So if you're asking why I think 6x7 film will clearly out-resolve any 35mm digital sensor of today, there's your answer. Because here even 35mm film out-resolves a Canon 5D Mark II. You may think the digital shot looks better, but that may be because of acutance... but that's a whole 'nother subject. Suffice it to say, though, that I'd probably agree with you that a real-world 24x36 print from a 5D Mark II would look better than the same print from meticulously scanned 35mm Velvia. Because of the low noise & the less gradual fall-off in faithful reproduction of contrast.

Now what'd be interesting would be to re-perform these tests with test charts of varying contrast. Because there I believe the 5D Mark II would come out ahead of 35mm film b/c of exactly what Daniel mentioned: digital has a sharper fall off than film (you can see it in the test charts above).

So what I'm saying is: Scott, you may not be wrong. In your experience, given your scanning methods & the contrast of your subjects, you may very well find digital to outperform film. But don't fail to recognize the potential of any one format when pushed :)

Daniel, or anyone, what's the easiest way for me to measure the actual contrast of my printed test chart under my lighting conditions? And when Fuji quotes Velvia to have a resolution of 160 lines/mm at a contrast of 1:1000, are they really talking about 10 stops difference between black & white? I'd imagine my reflective print under halogen lighting would have much less contrast than that... I imagine that if I printed a huge patch of black vs white on my printer then held it up under my halogen lighting then took spot meter readings I could get the exact contrast?

Cheers,

Rishi

P.S. Credit goes to Mauro for taking these shots on film & mailing them to me. Though Mauro may start some flame wars by using the term 'film vs. digital', at least he's always willing to provide the raw images to whomever may ask. I really applaud him for that; he's not scared to back up his claims with hard evidence. That being said, Mauro sometimes I do find you to over-exaggerate some of your claims... your initial measurements of the resolution of this film shot scanned in on your LS-9000 was helped by some ridiculous over-sharpening :) But you were in the general ballpark!

rishij · March 28, 2011

I bought my Nikon 9000 just prior to discontinuation. I'm glad that I bought it then, but it doesn't make sense at current prices.That will be the the end of film.

</blockquote>

Any electrical engineers here that want to get together & build a new scanner? I've got some software devs here in Seattle I could pull in on the project, I've done extensive research on materials to flatten film (no wet-mounting please) & some methods to enhance the resolution of scans using current lower-tech linear CCDs, as well as some ways to increase Dmax detail for slides, and some methods to compress the tonal range of negative scans to avoid the ever-so-present 'washed out' look of over-exposed negative regions... I bet if we could make a 35mm & MF scanner that is actually user-friendly, introduces no film curvature-based softness, is high-enough res to extract all the detail out of even MF, & maintains color accuracy by use of custom profiles, there would be a market for it.

I would argue that one of the reasons these $1k to $2k scanners failed in the market (besides the advent of digital) is because they were so user unfriendly & yielded so vastly different results, ranging from poor to great, in the hands of different users (just search photo.net for all the threads over the years!).

In the digital age, though, there's even *more* of a need to digitize those stocks of film we have lying around. And it's just sad that for quality we have to buy a $13k Imacon scanner, or rent time on it for $60/hour, or get drums scans at $60 a scan.

The success of such a project would, of course, depend on the price point of the scanner. But if Plustek can sell their POS scanners for $500... I know that I or a number of us on these threads could design a much, much better scanner.

Just a thought/dream I've been toying with for a while but finding interested/qualified/motivated folks & the capital is hard when you're pitching a product meant to serve a market of diminishing returns :) I still think there's a need for it though simply based on the forum posts that pop up here on photo.net every week re: 'the best way to digitize my film'... right now, there hardly is a way at a reasonable cost/effort (unless you've got yourself a LS-9000 with glass holder). I revamped the optics in my Minolta DSE 5400 & built an entirely new holder for it & now finally get scans I could compare to one of the newer Imacons (which are just ridiculously priced). But that was a lot of work/innovation that your average Joe does not want to spend time doing.

Film users: would you be interested in such a product?

Cheers,

Rishi

rishij · March 28, 2011

I should probably say that Plustek had the right idea in their 7600Ai model, just poor execution.

For example, good ideas: including an IT8 target & offering multi-exposure for enhanced dynamic range (though, ironically, they <a href="http://plustek.com/usa/products/opticfilm-series/introduction.html">say</a> this helps most with negatives... and I quote: "A recent <a href="http://www.luminous-landscape.com/images-105/Plustek-7600.pdf">review</a> indicates that Plustek film scanner’s performance to show the shadow details is remarkable, especially in scanning color negatives."... that's not at all what Mark Segal was trying to say if you read the actual review... Dmax is the highlights in negatives, LOL, and it's Velvia's Dmax that is the beast that needs to be tamed).

So, I was saying: they had the right idea, but poor execution. The multi-exposure barely helps, fixed-focus and lack of film-flattening methods ensures no edge-to-edge sharpness (imagine buying a dSLR where 10-20% of your edges were all out of focus!), & the light source doesn't help with film's inherent defects (enter the work of Erik de G!). Not to mention the (real, not quoted) resolution is sub-Nikon's resolution, and so resolution-wise we have:

Plustek < Nikon << Minolta DSE 5400 << Imacon

And let's not even get into the shadow detail one can see within Dmax regions of Velvia under a light microscope...

I don't see why we should be stepping backward in technology as we move forward in time...

Rishi

steve m smith · March 28, 2011

In the digital age, though, there's even *more* of a need to digitize those stocks of film we have lying around.

</blockquote>

There might be more of a desire to scan them but is there really a need?

rishij · March 28, 2011

Put another way, here's a glaring example of why people may not see the potential of film. Let's compare the resolution of a 35mm scan using a Nikon LS-5000 vs Imacon 848 (same piece of film):

<a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/NikonLS-5000vsImacon848.jpg">Link to Full-Size Image</a> (Images shown at 400% for clarity)

I've already shown above that the Imacon scan shows ~20MP of information from this 35mm frame of Velvia 50.

The Nikon scan seems to resolve down to about 3.0 by my eye (it resolves a little more along the stepper axis, but I think it's more fair to take the worse of the results). Doing the same math as before:

[3x100x8.6] x [3x100x8.6]*1.5 = 9,984,600

Or about 10MP, half the resolution of the Imacon scan (which still doesn't resolve everything on the film when compared to a light microscope).

That's a pretty large difference (compare a 10MP dSLR vs a 20MP dSLR...). And that shows the underperformance of a LS-5000, regarded as one of the best 35mm scanners amongst some folk. That Nikon scan was also the best of a number of focus attempts, using film-flattening techniques. Imagine the results with even lesser scanners or worse technique.

-Rishi

P.S. Steve: ok, sure, 'desire' not 'need'.

Dave Luttmann · March 28, 2011

For what it's worth, it appears a new 35mm / MF scanner will be available in June.

<a href="https://reflecta.de/en/products/detail/~id.425/reflecta-MidformatScan-MF5000--from-june-2011-available.html">https://reflecta.de/en/products/detail/~id.425/reflecta-MidformatScan-MF5000--from-june-2011-available.html</a>

Hopefully, this will at least be slightly comparable to the Nikon. And thank you Rishi for possting that chart showing exactly what many of us have found to be the case.

swilson · March 28, 2011

I doubt it will come close to a Nikon 9000, but maybe better then a flatbed.

I see its optical resolution is 3200 dpi, which tells me it is not going for the high end. But if it is sharp at 3200 dpi then it could do fairly well with MF.

Dave Luttmann · March 28, 2011

We'll have to wait and see on this one Scott. I used to get incredible results from my old Imacon 343 at 3200ppi. I agree that if the optics are sharp, 3200 should be enough for my 6x7 frames. That's a 29x36 at 240ppi on print.....or 13x19 from 35mm film.

rishij · March 28, 2011

Interesting. Thanks for the link Dave. But, honestly, Nikon optics were supposed to be sharp & still couldn't resolve half the detail in the film at 4000dpi. The Imacons resolve so much b/c they have good optics & a huge enlargement factor (& correspondingly a large CCD). Looking at the form factor, I doubt they have much enlargement going on & therefore 3200dpi doesn't inspire much confidence... Neither does 3.6 Dmax.

Furthermore, I don't see one innovative thing about the scanner listed on that site. But at least it's nice to know apparently there's enough of a market that people are still coming out with these things.

I eagerly await the tests on this unit...

Rishi

scott_ferris · March 28, 2011

Rishi,

I didn't ask "Scott Ferris, FYI, the way you get MP ratings for film is......" . Dan South did say "And how exactly does one measure MP with a microscope?"

Now if you look at my posting history you will find that I have always been against home scanning film captures, the probability of image quality lose is too great. That is one of the many problems with using film nowadays, realising its potential in a digital world. Using your Nikon scanners, that get more expensive on eBay than they were new by a long way, you are losing any theoretical advantage the film might have in the first place. For prints up to 20" I have said many times wet prints from film are the best thing to do, otherwise you are better off shooting native digital. For prints above 24" the problems of enlarger lenses and paper handling start to become apparent, unless you have a very well equipped darkroom and a lot of skill wet prints over 24" are disproportionately difficult and expensive, as are enlargers for 6x7 and 6x9 negs. On the other hand digital prints of any size are comparatively easy, cheap and consistent.

So your theoretical film resolution advantage can only be realised if you wet print or you use a $13,000 scanner. In this day and age of digital printing even film captures that doesn't seem like much of an advantage. It is like saying my car has a top speed of 200mph but is governed to 155mph. Your car has a top speed of 155. For considerably less than the Imacon scanner you could buy a new Pentax 645D, or have your choice of many used Hasselblad digital systems.

Despite all that, from my experiences of pro printers printing my images, my 20"x30" 135 digital prints outperform the same ones done wet with Velvia 50, maybe it is the pro printers I have used, though they have the highest reputations, but they just look better. My best selling print, by the way, is a 20x30 image I captured on 135 format Velvia 50.

If there was a scanner market, the companies that are geared up to exploit it would. They have the hardware and software in other products, it would not be hard, they just don't see the market.

But my real interest in the thread, the reason I am so bloody minded has been asked very eloquently by Daniel.

"what is the purpose in trying to match the digital shot to any of the film shots given that none of them gave a particularly pleasing or ideal rendering of the scenes?"

That is my biggest issue with film, if you are going to digitise it anyway then save a step and capture it digitally. And, few would argue that digital post processing gives such superior control it has far superseded anything that was possible in the darkroom and adds so much quality to the final result it is, effectively, necessary.

dan_south · March 28, 2011

DLT, good points about the many variables. One important note is that the lenses that I am using today to shoot

digital full frame are much sharper than the lenses that I had at my disposal when shooting most of my 35 mm slides.

Lens quality cannot be overlooked in such comparisons.

mauro_franic · March 28, 2011

Rishi, I added the scan of the same film with my Coolscan 9000.<div>

</div>

mauro_franic · March 28, 2011

Trying again. The scan I added is from my 9000 using the glass holder. <div>

00YTJf-343081684.jpg.2f2989c04c637d8a2c44da8d2795c003.jpg

</div>

rishij · March 28, 2011

Sorry Scott, my bad. Hopefully Dan South finds the explanation satisfactory.

Also, I pretty much agree on most of your points.

"If there was a scanner market, the companies that are geared up to exploit it would. They have the hardware and software in other products, it would not be hard, they just don't see the market."

</blockquote>

I would argue though that Nikon, Minolta & Canon all dropped the ball on their film scanners by not providing Imacon-like quality at a reasonable price-point (& even then, only the newer Imacons provide acceptable scans IMHO opinion b/c up to the 848, the harsh light source accentuated pepper grain far too much to get good enlargements). The LS-9000 came the closest with their innovative diffuser & glass holders (which IMHO only worked to modest effect). But didn't cut it in terms of resolution (you can see this in the simple fact that the stepper motor axis has higher resolution than the CCD axis... less a problem on the Minolta DSE 5400 or the Imacon). Better quality/reasonably priced scanners would've given film a little longer life.

And Scott, yeah I agree that's a good question. The really interesting question to me is whether or not film actually captures more color information than digital (given its broader absorption peaks compared to transmissive spectra of bayer pattern filters).

Has this ever been accurately assessed/answered? I can think of a couple ways to try the experiment...

Rishi

rishij · March 28, 2011

Yeah but Mauro you did some extensive processing to that scan to get it to look like that... which is fine, it's just that most images you can't treat that harshly. That's why I was conservative and gave the Nikon LS-5000 only a resolving capability of '3' on that chart. Although I'll admit I think the LS-9000 has higher resolution than the LS-4000/5000, perhaps b/c of better optics or a larger enlargement factor/CCD? I don't know.

To be clear, I had to perform an unsharp mask of +168, employ NeatImage to reduce noise, then boost contrast 60% to get an image that looks like your LS-9000 scan (qualitatively), shown below:

<a href="http://rishisanyalphotography.com/ForumPostFiles/photo.net/DifferentialScanningTechnique_Velvia50-3.jpg">Full-Resolution File</a>

With that kind of enhancement, my Minolta DSE 5400 resolves as much as the Imacon. I don't buy it simply b/c you can't do this kind of enhancement to normal images.

Also, you have considerably more resolution along your stepper motor axis than your CCD axis (I know you know this)... this is less of an issue with the Minolta, and almost a non-issue with the Imacon (which makes sense, since the latter have increasingly higher resolution linear CCDs).

-Rishi

rishij · March 28, 2011

Seriously Mauro, your last post is really misleading. My Imacon scan was not sharpened/enhanced like your LS-9000 scan... it's just not a fair comparison.

mauro_franic · March 28, 2011

Rishi, you are correct on the direction too. My Coolscan resolves 3900 dpi along with the motor and 3650 across the sensor. A sensor upgrade could have realized some gains.

To your point of building dedicated scanners, I can help the project with engineering and financing but the effort may be better spent financing/lobbying with Nikon for additional production. I thought about approaching them but I can never get the time.

mauro_franic · March 28, 2011

Sorry, I always apply sharpening to measure resolution. It makes it easier for me to evaluate.

Both the Imacon and the Minolta resolve the film almost completely. The Coolscan leaves detail unresolved.

rishij · March 28, 2011

We should think about it. There are ways around a sensor upgrade than can still yield an increase in resolution. Perhaps trying a proof of principle & then approaching a company would be more reasonable. At any rate, I think those of us passionate about film owe it to ourselves (& others) to try. Thanks for the (preliminary) support. I'll try & get back to some experimenting & shopping around of some of my ideas in my 'free time'. Are you an engineer by trade?

mauro_franic · March 28, 2011

I have a bachelors in Engineering, a Master in Engineering and an MBA in Finance.

(Industrial)

scott_ferris · March 28, 2011

And that misleading representation is the reason Mauro and I have had such heated threads and the reason people think I am a digital zealot and pick on him, he too often presents excellent work in stupidly, and unnecessarily, misleading ways.

I have seen him do just as drastic disservices to native digital captures from both his 40D and a 5D MkII. I feel it is a shame and try to point the inconsistencies out.

I did a lot of work on film scanning and digital workflows in 2002-2004, I was investing in AF gear for the first time and wanted to know if I should get a 1Ds or two 1VHS's, I got the two film cameras. At the time very high quality wet prints were still easy to get and the time commitment to anything digital was ridiculous. But time moved on, with the goalposts, and one of those $2,000 film cameras ended up with less than 100 rolls through it. You are 100% right in saying the scanner manufacturers dropped a ball, and that did have a big effect on film use, if scanner technology and software had even approached digital camera technology I am pretty sure I'd still be shooting those 1V's instead of using them as bookends. Maybe that was Canon and NIkons plan......

mauro_franic · March 28, 2011

Dan, your point is always true form me. Some of my film 15 years ago was shot with a consumer level zoom (sometimes even kit lenses). Over time my lens collection grew and so did the contrast and detail of many shots.

scott_ferris · March 28, 2011

And I have a large stomach and had lentil patties for dinner (non industrial)!

Sign In

Film vs Digital - Color Rendition

Recommended Posts

Link to comment

Share on other sites

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

Posted Images

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Create an account or sign in to comment

Create an account

Sign in