Film vs. Digital

[John_4185]

Back to Richard with what I hope is pertinent.

First, these so-called modern digital cameras are what I call techno-vain; they can create profound confusion in what is really a simple camera 'interface'. They like to throw up onto the screen every darned setting they are capable of so that the world is obscured with irrelevant metrics, much like a few photographers view the world - as photometrics, not imagery, not the Thing Itself.

You (and I) have used small cameras and our brains for forty and fifty years, respectively. We don't need that stuff. If you get one, then I suggest you find one you can tell to shut-up, be silent, don't display a darned thing in the viewfinder but the scene itself. If the camera will not operate properly without shutting up and stifling the techno display, then it points to profound shortcomings in the technology. If it has more than three buttons, ask yourself what the heck is going on here, what is wrong with the technology, why you need a pilot's mentality to make a picture.

Oh, one more - if you get one that has a default setting to makes a phoney a camera-like noise, don't buy it - spite the manufacturer. Once I was near a table full of camera guys and heard what was almost the classic Nikon-F (close to the Nikon-S motordrive but with mirror sound.) It came from a little digital camera. I knew that very moment that something was terribly wrong with the whole paradigm.

I work with digital on the day job and I tell you I just hate the cameras. I hate any inanimate thing that sticks irrelevant technolgy in my face. I don't even like camera gear that uses batteries. So that's my prejudice, up front and center.

Peace,

[Eric Leppanen]

West Coast Imaging is a well-respected digital lab located just outside of Yosemite. Their conclusions are as follows:

- When printing to an Epson printer (360 DPI native resolution), you need only a 240 DPI digital source file, which is uprezed to 360 DPI in Photoshop. They see no visible benefit from providing source files with greater than 240 DPI.

- When printing to a Lightjet/Chromira printer, you need only a 200 DPI digital source file. They see no visible benefit from providing source files with greater than 200 DPI.

Please reference their article: www.westcoastimaging.com/wci/page/info/photoshoptip/tip25.html

Applying the 240 DPI number, we get the following maximum output sizes (rounded to the nearest inch):

Canon 20D (2336 x 3504): 10 x 15"

Canon 1Ds2 (3328 x 4992): 14 x 21"

Phase One P25 (4080 x 5436): 17 x 23"

Phase One P45 (5412 x 7216): 23 x 30"

My experience with drum-scanned 4x5 is that I start seeing obvious artifacts of over enlargement (fuzzy pine needles on trees, etc.) starting around a 24 x 30" print size. So in theory, it would seem that the P45 is getting into the realm of 4x5-based print resolution. Similarly, to my eye 6x7 starts running out of gas beyond a 16 x 20" print size, so to hazard a generalization, I would say that:

Canon 20D is roughly equivalent to 35mm film format

Canon 1Ds2 is roughly equivalent to 645 film format

Phase One P25 is roughly equivalent to 6x7 film format

Phase One P45 is roughly equivalent to 4x5 film format

So Richard, there is your plain language comparison. Note that we're talking about resolution that actually shows up in a print, not how much resolution you seen on your CRT/LCD monitor in Photoshop.

Of course, to achieve these results requires lenses capable of fully resolving the digital format. The limitations of the Canon wide-angle zooms in this regard are well known, and the jury is out as to how many MF lenses can stand up to the P45.

[Steven Barall]

Cameras are just tools. Comparing large format film to digital is like comparing a screw driver to a hammer. Pick the right tool for a particular job.

Ultimately you need that camera system that you are actually going to use. If you look at your camera and get a feeling of dread for whatever reason, it's not the camera for you. It all has to serve your goals.

Good luck and be happy.

[David Luttmann]

Steven,

If you could you tell my wife that the handle of the screwdriver isn't a hammer.....I'd really appreciate it! ;-)

[Glenn Kroeger]

Like the blind men examining the elephant, everyone here is right to a certain extent.

As a seismologist, digital sampling theory is my field, but as a photographer, images are my passion.

An instructive experiment is to photograph a standard USAF resolution chart using a DSLR and then MF or LF film followed by a very high resolution drum scan. Examining the results show the difference between digital and analog capture of the line pairs. The DSLR image will show very clear dark lines up until the cutoff frequency of the sensor at which one gets a severly alaised rendition followed by mush. The film will show a higher resolution, where resolution is defined by the last lines that can still distinguished. But long before that, the lines become low contrast and mushy (like fuzzy swollen rice grains!) unlike the digital capture. So which is better? Do you like your diagonal lines fuzzy and swollen or sharp with stairsteps? That depends on the subject and the response of the human eye-brain system to images. It also depends ultimately on the subject matter in the image.

What I can tell you from experience... my own digital processing and printing (it is critical to do this yourself, as processing flow and sharpening procedures greatly influence the resulting print) is that I have produced 13"x19" prints from an 8MP DSLR that greatly surpass anything I was able to produce from 35mm chromes using both analog and digital enlarging processes.

Having recently worked with output from a Nikon D2X and Canon 1DsMkII, I find these files nearly, but not quite equal to drum scans from MF (6x8 cm) Velvia. They are much less noisy and thus produce better skys, but they are also lower resolution and thus produce mushier foliage. Apples and oranges as many have said.

Similar to the WCI results posted above, I find that when printed at higher than 200 ppi, the results are spectacular. I find I can go as low as 180 dpi (upsampled to 360 for printing) before things really start to fall apart (or really smear apart!). This translates to about a 16x24 for the Canon 5D or Nikon D2X and about 18x27 for the 16MP Canon 1DsMkII as the point where drum scanned film will show a clear advantage in detail but still a disadvantage in grain and noise.

I have printed some files from the new Mamiya ZD at 20x25 (200 ppi) and they are as sharp at that size as anything I have printed from 4x5... but at 30x40, the LF film will still prevail. This comparison also assumes you print to the very edge of the frame... the LF film will allow post-exposure cropping at 20x25.

I have just been looking at some raw files from the new 39MP MF backs. These files are spectacular and the processing tools (Nikon Capture, PhaseOne C1 and the latest Hasselblad/Imacon software) are starting to deal very effectively with the CA problems. Canon needs to address this issue to remain competative with wide angle lenses (they also need to make some better WA optics!)

What digital has done for me in landscape work is open up new possibilities for making images where the added speed (in framing and exposing), the added speed (ISO) and the added DOF allow me to capture images in situations of rapidly changing light that I could never capture in LF except by accident. OTOH, when I have the time to pre-plan and setup a shot, MF and LF film still hold their own.

[bglick]

Emre - will Canon / Nikon get to 24 - 28 MP? It's certainly possible in my opinion. I bet the MF back makers are wondering that same question. :-)

If Canon gives up the 1.5 : 1 aspect ratio and go closer to a square, say 1.15 : 1 and maintain the same image circle diaganol of approx 45mm, they can add 18% to the area, or 17 MP x 1.18 = 20 MP.... to gain the remaining pixles, they would have to either make them smaller, or probably just add a few mm around the border of each dimension. Example, 33 x 37mm sensor would only increase the image area diaganol from 44 to 49.5mm. Something most of the L lenses can easily absorb, this equates to 24 MP..... then, make the pixels a tiny % smaller if they wanted 28 MP... but I think 24 MP is surely in reach, IF, they decide to attack that end of the market.... I think they will, but maybe in 5 years, just a wild guess.

Eric, great post indeed.... I tried this kind of post initially, (sumplisitc, but it got challenged :-) but I think working backwards from the print rez is great cross check. However, I think you short changed film a bit, and i know WCI also doesn't address this well. Based on the DOF, the quality of the lenses and the alignment issues discussed before, 45 film can be pushed further then 24x30. If using Velvia, at the point of exact focus, the film can resolve up to 60 lp/mm....the printers being fed 240 dpi which equals / 51 = 4.7 lp/mm (with 1/3 loss in printing efficiency, this is 3 lp/mm to print, VERY SHARP), so 60 lp/mm / 4.7 = 13 x with no sharpening, or 46" x 62". Now, again, this is best case scenario for 45, low f stop, infinity focus (or shooting a flat subject, or front tilt tracked the subject plane perfectly), sharpest lenses, Velvia, perfect alignment, best scan, etc. etc. When you add some DOF, and reduce some of these variables to less then optimum, this max. print size will obviously be reduced, and possibly down to the 24x30" you mentioned. BTW, if one prints on high rez glossy papers, the paper can hold even more detail, up to approx 7 lp/mm, assuming the printer can deliver such rez.



One issue WCI never will discuss is the need to overscan to resolve desired dpi. I am not knocking Tango scanners, I have worked with many tango files, I also have read many independent test resolution test results of the Tango vs. other scanners. Although the Tango scans up to 11k dpi, (which is overkill for photographic film) you must overscan by a factor of 1.25 x to actualy resolve the desired dpi (assuming the rez is in the film), this is called scanning efficiency.

The Tango test results yeild about .80 efficiency, in other words, when scanning a 100 lp/mm target, at 100 lp/mm (5100 dpi) you will only resolve 80 lp/mm in the digital file. However, if you scan at 100 lp/mm * 1.25 = 125 lp/mm, the digital file will resolve 100 lp/mm. As you can imagine, this is significant if you trying to grab all the films detail. You can run test scans of targets to determine this for any scanner. The 100 lp/mm example was only used to simplify the math, color film is much less of course. The best scanners tested, most from the next generation after the Tango, have efficiencies up to 96%. Simply better lenses, electronics, software, etc. Of course consumer type flat beds can have efficiencies for film down to 35% when scanning near the resolvable film detail levels. But be careful mentioning this to WCI, they are extremely defensive over their Tangos.

> Cameras are just tools. Comparing large format film to digital is like comparing a screw driver to a hammer.

I get your drift Steve, but, a hammer will not perform the task of a screwdriver and vica versa. However, digital and film are both methods of capturing image data, with the exact same light levels striking them.... so they are two different tools that perform the exact same task, albeit with some pros and cons of each, which is what these discussions are always about. Considering the price of conversion, or even adding one vs. the other is so steep, I think these discussions are healthy.

To the original poster, Richard, I am sorry for digressing into geek land a bit, but as with some of the recent "hot" threads, this thread is a perfect example of just how hard it can be to oversimplify a response about a very complex subject, without loading up with disclaimers, caveats, etc. Oversimplification can often lead to bad information, bad purchase decisions, too high / low expectations, etc.

Glen, I just noticed your post when I finished writing this.... and I appreciate your input....as you raise yet another issue.... resolution on test targets is a fuzzy and subjective issue....the sharpness-to-blur curves are clearly not the same in film / digital, so trade off exists in which subject matter is friendlier to the types of blur, as you pointed out. However, this should not suggest the comparison is too hard to evaluate, and therefore 8 MP resolves as much as 810 film.... it's a fine detail, not a deal breaker in the comparative methodology.

Since you have lots of 4x5 experience, I am curious of a more thorough opinion of the 39 MP backs vs. 4x5 Velvai, which is what I think you shot? Or if your just starting analyzing such, keep us posted! If this thread is on-course, I would suspect this back can beat 4x5 film when there is signficant DOF (difraction limited f stops for 4x5) in the scene, but when shooting a flat subject or using tilt to track the subject plane perfectly, the 45 will still out perform the digital by small margin. The suspense is tooooo much :-) Again, thanks for the contributions!

[Glenn Kroeger]

WG:

I don't have the data yet to compare 39MP to 4x5 Velvia... but based on looking at the 39MP files, I suspect they are neck and neck... as for that, 22MP backs are close. Lenses for the digital back are an issue, but the new Schneider Digitars may be up to the task. Film flatness and grain are issues for the 4x5. I have to admit that after looking at and printing digital files, I have realized that much of the quality I have always ascribed to LF images has been due to low grain... and digital can deliver that quality easily. I have come to dislike grain more and more so that I am now bothered by the grain in my MF images. The biggest problem with 39MP is $30K!

DOF is clearly an advantage of smaller sensors. Many folks have found the Nikon D2X (12MP) equal to or better than the Canon 1DsMkII (16MP) for landscape work because of the greater DOF on the smaller sensor.

As for DSLRs exceeding 20MP, it will undoubtable happen this year. If Canon simply scales the pixel density of the 20D (which has very low noise and good high ISO peformance) to full frame, they will be at 21MP.... and the 20D is 18 month old technology by now.

Finally, when comparing theoretically, I think that we place too much emphasis on resolution. First, it is poorly defined... even the Rayleigh criterion is dubious for non point source images. Secondly, the human eye-brain system clearly prefers crisp edges to high resolution soft edges. I have looked at alot of test charts where line pairs are technically resolved, but still look like crap. I will take a lower resolution with high accutance and low noise any day... this is why digital looks so good with apparently low resolutions. I have seen 20x30" prints made from 8MP files that are stunning... but they are of images that don't rely on fine detail and are viewed at resonable distances (image diagonal).

Vegetation in landscapes is one area where high resolution really influences perceived sharpness... and it remains a bane of digital sensors.

Glenn

[John_4185]

Like the blind men examining the elephant, everyone here is right to a certain extent.

And I suppose only you have the Truth. So let's see what you prescribe.

An instructive experiment is to photograph a standard USAF resolution

What a laugh. An image with notable esthetic quality transcends grain sniffing, anal, microscopic metrics. Your observations might be of passing interest to a reconnaissance 101 class, but only optical benchracers would be silly enough to stay there.

We are not blind and we aren't feeling up elephants.

[Glenn Kroeger]

jj:

No, the point of the elephant fable is that everyone is right and no one has a corner on truth... which is exactly the point of my post.

As for AF charts, I said they are instructive in understanding why digital and analog are different, which if I recall was one of your points. If you will read my post, you will notice that I clearly indicate that subject matter matters in peoples reaction to real images.

Glenn

[Scott Fleming]

This is one of the best of all time threads. I'm going to print it out and file it as well as bookmark it and keep watching it.

This whole subject is as indivdualistic as possible. Digital capture has really tossed the photographic world for a loop. Never since the invention of glass plates has so much changed so fast. Though I enjoy such discussions the personal reality of our photography comes down to what feels good to us and what we can afford. It took me three years of inner machinations to get where I am today on this subject.

As a piper cub flyer I can enjoy listening to the stories of a young jet jocky serving in the Persian Gulf once in a while but dwelling on them would only make me unhappy. I really like making pictures. I love analog equipment. Both handling it and using it. I feel just like jj. Complex digital cameras don't get me off. Give me a young Sophia Loren in a modest bathing suit. Paris Hilton gives me an empty limp feeling.

More than anything I like my favorite prints. They are my friends. I never tire of looking at them. I even forget after a while which camera I used or which lab printed them or how they were scanned . I'll never know or understand the joy of Pt Pd printing in my own dark room nor will I ever own my own 39 MP MF back. But I can make a print I love just as well with my comfortable analog cameras as anything else out there. Speaking relatively. I can punch up a digital photograph in PhotoShop and even get it sharpened ok without owning a third party sharpening program. But I'll probably never learn how to build a complex mask so as to blend two exposures. Screw it. I prefer graduated neutral density filters. I never had a disciplined mind and I can feel the brain cells slipping away by the year now. I'll never be a great photographer nor a Capture One or PS maven. Don't wanna be. I'm very glad however that others of you are and report the latest develpments here and around the web. I'm so glad I don't do this for a living.