scanned film vs digital

[danielleetaylor]

<p>Mauro,</p>

<p>A few things. First, a big part of the reason why the 40D looks so bad in your crop is because you had sharpening shoved up on the camera, then enlarged linearly. If you enlarge Bicubic Smoother, THEN sharpen, it looks better. No, it's not resolving the small text, but at least it's not horribly pixelated giving a false impression of relative performance. Sharpening is always the last step.<br>

<br /> Second, I still question the 40D's performance in this test, but I'm not sure yet what else might be wrong. You should have shot that particular lens at f/4 which would have improved things a little bit. Other than that nothing stands out at me right now. It's not that I expect a 40D to out perform a 35mm frame on maximum resolution at ISO 100, it's that this still looks like the lowest performance I've seen a 40D turn in. But I don't know why.</p>

<p>Third, something you should be frank about when posting this: if the full image were resized then printed at this scale, we would be talking about a 40x60 inch print. I don't think I have to point out that 60 inches is, from a technical perspective, beyond the ability of small format when the subject matter has fine detail and is larger than most people ever print anything. I printed this crop on my printer at a size equivalent to a 16x24 and at that scale in print there's just nothing to see between the film crop and my version of the 40D crop. You've done a good job of amplifying the differences to seem important, but most people would not expect to make a decent 60 inch print from a 40D or 35mm. At least not of a subject with fine detail intended for close up inspection. Film might look good at this size versus a Nyquist limited 10 megapixel APS sensor, but it's going to look bad compared to anything medium format or better. And it's not going to look better than a 35mm sensor as it does against the 40D here.</p>

<p>I can't (or can't easily) reproduce this test exactly as you did, so I don't know if I can determine if any other changes, besides aperture, would improve the 40D's showing. I do hope you correct your presentation with improved enlargements however because proper handling produces a noticeable difference.</p><div></div>

[danielleetaylor]

<p>Here it is inline. The linked version has all three.</p><div></div>

[Landrum Kelly]

<p>Thanks, Dave and Mauro. </p>

<p>Dave, I can see from your drum scan just how close you got to a 24 megapixel DSLR using 35mm film, but I appreciate your modesty in saying that " This is a good sample of 35mm really topping out around the 15-17mp for most color use." </p>

<p>Likewise in your case, Mauro, I appreciate your saying that "Ektar 35mm should resolve somewhere in between the D2X and the A900."</p>

<p>I'm impressed, all the way around.</p>

<p>Well, now, boys and girls, what does that imply about Ektar 220?!! I would say that it would be up there at about the reolution equivalence of a 60 MP digital sensor--or better. Anyone want to buy the Phase One P65 digital back, or will a used film 6x6 or 6x7 camera suffice? I get happier and happier everyday about buying up the Hassy 6x6 stuff, but I think that I will sell the H1. I think that de Bakker is right: better to stay with one system and master that system. Mamiya or Pentax would have been fine, too, of course.</p>

<p>Thank you, guys, for the solid results. Your results are not hyperbole, are not speculation but are based on hard evidence, are not the work of zealots but of true craftsmen, and show what the choices really are.</p>

<p>Thanks again. This thread has certainly been worth my while. Those who get tired of these threads can always go read something else, as far as I am concerned. There is always a new crop of potential converts (and reconverts) to film coming along who simply want to know the facts.</p>

<p>--Lannie</p>

[don_e]

<p>Sharpened, this that and the other operation and filtering, format conversions, compression, resizing, resampling...by the time these images, digitized film and digital camera images, reach this forum, none of the captured pixels exist in the image. That's not considering the circumstances of the original capture yet: platform, lighting, sensor quality, emulsion quality, lenses, aperture, shutter, iso, and the method of massaging some equivalence out of them. The absence of controls, too.</p>

<p>What conclusions can be reached through all that?</p>

<p> </p>

[Dave Luttmann]

<p>Lannie....just to be clear....that was an Imacon scan, not a drum scan. It is possible a drum could pull more out, but I'd say it's a matter of diminishing returns.</p>

<p>And yes, there have been many converts I've noticed even locally. I'm amazed how many people I meet locally with a new interest in film....prompting them to buy up used MF and LF gear. I've done a few basic film scanning workshops for locals here showing them how to profile film, fluid mount, etc, etc. The results are always the same. When they compare it to the DSLRs they're using, they always say something like "holy cow....I never thought film would be so much better." It's always fun!</p>

[Landrum Kelly]

<p>Thanks, Dave. I'm suitably impressed.</p>

<p>Don: "What conclusions can be reached through all that?" </p>

<p>Don, I think that, when one presents a link such as that which Dave has provided, along with 100% crops, and then backed off from high end claims about what the absolute resolution is, one can show quite a bit--enough to make believers of some of us.</p>

<p>In addition, one doesn't have to mortgage the house to make the leap into MF these days. If one doesn't like it, one has not laid out that much at the prices I am seeing on eBay these days--and, if one absolutely must decide that one made the wrong choice, one won't lose much on resale.</p>

<p>I'll keep shooting digital, of course, when the situation calls for it. I'm no fool, and I can shoot it now without a lot of trouble. I'm used to it. I just want something better, hopefully without bankrupting myself. MF sounds good to me right now.</p>

<p>--Lannie</p>

<p> </p>

[mauro_franic]

<p>Don E, "by the time these images, digitized film and digital camera images, reach this forum, none of the captured pixels exist in the image"</p>

<p>you have the raws posted for you to use.</p>

<p>The 40D shots are pin sharp to the point moire is perfectly defined. This shot is as good as it gets with the 40D. f8 is not the bottle neck of detail and you should not see any additional detail at f5.6.</p>

[matthew_jorgensen]

<p>I had the same question and I have concluded for myself that neither is better then the other just different. It all depends on what your trying to achieve. That being said here is a little video were they test film vs digital by blowing 2 of the same pictures up to something like 22 meters or 72 ft tall. Check it out its really intresting http://fwd.five.tv/videos/challenge-blow-up-part-3<br>

as always take with a grain of salt.</p>

[mauro_franic]

<p>You are welcome Lannie.</p>

<p>It is all about sharing test and examples abd debating together (based on facts).</p>

<p>Regarding the print size at which film gives you a visible advantage over the 40D, that to me is about 11x14. Some people can print 20"x30" without seeing the difference. </p>

 

[Dave Luttmann]

<blockquote>

<p>You are welcome Lannie.<br>

It is all about sharing test and examples abd debating together (based on facts).</p>

 

</blockquote>

<p>Very true indeed!!!</p>

[danielleetaylor]

<p><em>The 40D shots are pin sharp to the point moire is perfectly defined. This shot is as good as it gets with the 40D. f8 is not the bottle neck of detail and you should not see any additional detail at f5.6.</em></p>

<p>Mauro the lens you used peaks in center resolution at f/4. It has dropped by a small but, for something like your test, significant amount by f/8.</p>

<p>What is your opinion on the post processing?</p>

[francois_p._garnier]

<p>Daniel Lee Taylor quoted Normen Koren as evidence for the superiority of digital image quality. But, working as a scientist myself, I have to say Mr Koren made a lot of mistakes in his most quoted article:<br />1. All his conclusions are based on the thesis, that a 4000 dpi scan of Fuji Provia 100F represents nearly the best image quality you can achieve with film. A 4000 dpi scan is aquivalent to a resolution of only 80 Lp/mm. But fact is that this is only the resolution limit of an amateur scanner like the Coolscans (to be more precise their effective resolution is about 3600 dpi in real life, aquivalent of about 70 lp/mm). But it is definitely not the resolution limit of slide film.</p>

<p>I've achieved 130 Lp/mm with Fuji Velvia 100F, 120 Lp/mm with Fuji Sensia, and 110 Lp/mm with Fuji Provia 400X. I have used my older Zeiss 50 mm lens for the tests. A friend of mine achieved even a higher values with the new Zeiss ZF Makro-Planar 2/50. Zeiss published resolution tests of many different films. For example they have got 160-170 Lp/mm with Fuji Velvia (look at the "camera lens news" publication on the Zeiss Homepage). And Zeiss is definitely not biased to film: They sell more than 99% of their lenses for digital cameras.<br />If you use a very good prime lens, you can achieve much higher resolution with film, especially slide film and BW negative film.</p>

<p>2. He is mainly referring to the 4000 dpi scan, not to drum scans, and not to enlarging in the wet darkroom or slide projection. He is ignoring these means for best quality printing (or projecting).<br />With my Rodenstock Apo-Rodagon N I can transfer this extremely fine resolution I get with film onto paper, no problem at all. The same with slide projection with my Leica projector and Leica Super-Colorplan P2 projection lens: The resolution figures of 130 Lp/mm are clearly visible on the projection screen. Such high resolution is absolutely impossible even with high end beamers. They can only achieve about 25-30 Lp/mm on screen, and the color is much worse (I have tested that as well).<br />The physical resolution limit of a 24 megapixel 24x36 sensor is 85 Lp/mm. More is not possible, with finer structures you will only get aliasing artefacts (e.g. look at the tests at dpreview).</p>

<p>3. Like all other digital doom sayers he ignored the progress in film technology. He said there will be no more R&D in film, film will become much more expensive over the years and in some years no more film will be produced. He said that in 2002 and was completely wrong. More than 35 new films hit the market since 2002, we have significantly better films today, and with regard to inflation film is cheaper now than in 2002.</p>

<p>I don't trust any internet guru. I don't know whether someone is paying them for biased writing. Some of them make a living by hidden marketing. I make my own tests. And because of the results of these tests I prefer film.</p>

[mjferron]

<p>"So, when Michael says the best b&w 35mm film has only "8-10" megapixels of "detail", he means the detail in a scanned 35mm film that he's seen looks pretty much like whatever he thinks an 8-10 megapixel dslr would show."<br>

Exactly. And side by side comparisons with a D200 and an F100 both bolted to a sturdy tripod gave me this conclusion. 100% crops had similar detail with the digital looking much cleaner as we see in other comparisons. Now don't forget I said 35mm B&W film and not Velvia which can pull in a bit more detail then say Tmax100 and has finer grain. My findings btw are based on my experience with a Konica/Minolta scan dual 4 at 3200 dpi. Yeah it's not an Imacon but it gets right down to where grain starts fogging detail. At that point it's tough to get more detail, just a bigger file. BTW It's tough to match the tonality and feel of B&W film with digital.</p>

[Dave Luttmann]

<p>Interesting Michael. A couple of years ago I posted a test at DPReview comparing the D200 with Velvia scanned on a Minolta Scan Dual IV. We had no problem seeing that fine details like grass and trigs on the ground were much better resolved by the film. This showed up in 12x18 prints quite easily.</p>

<p>The digital files were processed in Capture One. The film scan was done at 3200ppi with 16x multisampling enabled, and manual focus utilized.</p>

[mauro_franic]

<p>Daniel, "Mauro the lens you used peaks in center resolution at f/4. It has dropped by a small but, for something like your test, significant amount by f/8.<br /> What is your opinion on the post processing?"</p>

<p>That lens at f8 far outresolves the 40d (as you can see by the very well defined moire). You can verify with the resolution chart next to the can.</p>

<p>Processing doesn't change resolution either. Upsampling cannot invent detail. Now you can upsample two eyes with a straight line and make them sharp on a 30x40 print; the nose will not be printed though.</p>

<p>Flat areas without detail will look nice but detail will not be there. Some apparent detail may even be created as a guess by the upsampling algorithm hence linear upsampling is better for comparison than bicubic. For example, if you look at the can upsampled by Lannie, it appears you can read the "Favorite Recipies" label but the letters have been deformed into somethin else that was not in the scene.</p>

<p>If you compare the two "Favorite Receipies" labels, that difference is massive and not the result of just a 10% difference in resolution.</p>

<p> </p>

[mauro_franic]

<p>Michael, "B&W film and not Velvia which can pull in a bit more detail then say Tmax100 and has finer grain"<br>

My 35mm tmx have 40+MP of tru resolution. About 10 times of my 40D. I have no clue of what you are talking about. Outresolving TMAX with a 3200dpi scanner? Gee wee.</p>

[mauro_franic]

<p>We are all pulling hairs now.<br>

No one, or very few, disagree that the 35mm film can provide better resolution and better dynamic range (negative film) than a top DSLR in 2009.<br>

Arguing by how much is a silly exercise unless you perform your own tests to know/convince yourself.</p>

[Dave Luttmann]

<p>Mauro, you'll find the most vocal naysayers here have never done any testing at all. Those of us who have and posted our results over the years know better. Let the rest find out for themselves as 1000 samples from others proving them wrong will not convince them.</p>

<p>Regards,</p>

[mjferron]

<p>Mauro I'm not debating film vs digital. I like my B&W film and these orange/apple comparisons are a waste of time. You can scan that B&W film till you have a 100mp image but you are not going to create detail that is not there. My darkroom prints are no more detailed than my 3200 scans though i prefer my silver prints to digital.<br>

Dave I do find film does better when you do include things like grass and leaves. The randomness of silver film grains seem more suited to the irregular shapes while digital holds it's own just fine when shooting subjects like cityscapes where there's a lot of straight lines.</p>

[Dave Luttmann]

<p>You bring up a valid point Michael. The reason I prefer film for landscapes is that I find it renders ultrafine details like leaves and grass far better than digital can. I've seen enough landscapes shot digitally (and done many myself) over the years that people were bragging up the quality...only to fine they looked artificial and lacking in texture.</p>

<p>I use digital capture for a fair bit of street work as with handheld shooting, high rez isn't that important. After I run the file through Silver Efex, the extra detail would fade away anyway. Of course, nothing beats the huge range of B&W films for that work though. I can tell you one thing though....the 3200ppi scans from the Minota do not pull all the detail from TMax when coupled with good glass. I see more detail from the Nikon 9000 @ 4000ppi and from the Imacon 949 @ 6300ppi. A 3200ppi scan cannot even resolve 70lp/mm....and Tmax is quite capable of going beyond 100.</p>

[mjferron]

<p>Dave I'll have to have a few of my better frames scanned with some quality equipment and see how that works out. I know with medium format B&W things like grasses just screamed for attention. I never saw that even with the 5D I owned for a short time. And that film was scanned on a lowly V700. I think one of the reasons M.R's 1ds vs pentax 67comparisons over on LL looked so close in detail was the subject was a distant cityscape with lines that favored digital.</p>

[Dave Luttmann]

<p>True Michael. The funny things is that even on the building sample where he said the digital beat the 6x7....you can clearly see more detail from the film than the DSLR. The funny part is he makes an excuse that it appears that way because the digital file needed more uprezzing than the film file.....well duh!!!</p>

<p>I agree that MF is the way to go....unless you're crazy like me and cart around the 4x5 gear as well.</p>

[todd_carson]

<p>Shall we stop looking at carefully constructed, manipulated, and protected tests, and look at some real world samples? So that I am not accused of poor scans or fixing samples I'm going to present 4 crops from Les Sarile's high resolution scans, and 4 crops from random DPReview sample shots.</p>

<p>Set 1</p><div></div>

[todd_carson]

<p>Set 2</p>

<div></div>

[todd_carson]

<p>Set 3</p><div></div>