Wow - read this re: Film versus Digital debate!

[steve_mcbride2]

<p>It may intrest you all to know that here in the UK (Lancaster/Morecambe) we not only have one of the top Pro D & P companies in the country (David Farnell) but his clientel are going back to film in some high numbers, so much so that he is now stocking & selling Pro Fuji C41/E6 emulsions cheaper than I can get them from my wholesaler!<br>

Also I am the proprioter of Sepia memories classic camera's and November was our best month so far this year!! (Check my website for goodies on www.sepiamemories.com)</p>

[vijay_nebhrajani]

<p>Rishi, if you meant this: "Given that the Canon 40D pixel pitch is considerably smaller than that exhibited by Full Frame cameras, I don't think you, Mauro, can compare these results to a '26 megapixel camera'." then I agree with Mauro: when you do end up packing 26 MP into a FF sensor, the pixel pitch will be the same as 10 MP on the smaller sensor.</p>

<p>The smaller sensor has 1/2.6 times the area of FF, so that's your multiplier.</p>

[Dave Luttmann]

<p >Mauro,</p>

<p > </p>

<p >I laughed like crazy reading your summary of Michael’s assessment of the D30 vs Provia. I used to have an Imacon 343 and still have my old D30. Using good optics on both the D30 and a Nikon F5, I scanned film with the Imacon and converted the Raw file with Capture One.</p>

<p > </p>

<p >Looking at the results at 8x10, the digital file looked soft. Yes, it had sharp edges, but at 180dpi on print, it couldn’t stand up to scrutiny against the Provia. I’ve yet to find anyone comparing prints that agreed with Michael. At 13x19, it was worse. The film began to show grain, but held detail. The D30, at 110dpi, had little detail, virtually no texture, and poor rendering of color because of the interpolation required to even achieve 240dpi on print.</p>

<p > </p>

<p >Most of Michael’s reviews don’t stand up to scrutiny. That said, digital is a fine tool that I use regularly….but to reproduce fine detail, texture, and color, film, especially in MF & LF is what I prefer for high quality output. A lot of pros have been increasing their film use for non-time sensitive work. There has been a tremendous increase in interest, as seen at Photokina, in LF gear as even young people in their 20’s see LF as an opportunity to have the highest quality, for a low cost. Nothing like having $1500 in camera, scanner, lenses, etc, that exceed what a $30,000 back can achieve.</p>

<p > </p>

[mauro_franic]

<p>Yep, I thought also a review of how genuine fractals work was in order too.</p>

[edgreene]

<p>Dave Luttmann: <em>A lot of pros have been increasing their film use for non-time sensitive work. There has been a tremendous increase in interest, as seen at Photokina, in LF gear as even young people in their 20’s see LF as an opportunity to have the highest quality, for a low cost. Nothing like having $1500 in camera, scanner, lenses, etc, that exceed what a $30,000 back can achieve.

 

</em>

<p>Agreed. I'm shooting the Inauguration with my E-3 and E510 and f/2 lenses. But when I want to get paid for my work, my <em>serious</em>-non-PJ work, I shoot film in my Pentax 67. As previously noted, I can shoot at 9AM, take my work to the lab and have prints, CD/DVD negs in hand and images uploaded by the lab by Noon.<br>

I've never yielded to the complications of trying to produce ``perfect``digital images, though my Inauguration digital images will be uploaded on site from my HP laptop with an AT&T wireless card. I do admit that my Olympus DSLRs have a distinct focal length advantage over my film gear in that according to my position on the Mall, I'll likely have to shoot my E3 with 400mm Nikon Nikkor and Olympus T-CON 1.7 @ an EFL of 1360mm @ f/4. <em>No way film gets there without a 20 pound Bazooka of a lens/tripod rig.</em><br>

*<em>Though I take comfort my Pentax LX film SLR still shoots rings around all but the few high end DSLRs, EOS 5D included. </em><em>I also note my 1992 Canon EOS A2 shoots @ 5fps <strong>unboosted</strong>, again knockng the socks off all but high end DSLRs.</em></p>

</p>

[Dave Luttmann]

<p>Ed,<br>

The Pentax 67 is awesome. Built like a tank. Then again, my RB67 IS A TANK! ;-)</p>

[rishij]

<p>And my EOS-3 has eye-controlled focus, which works great if you don't wear glasses & when you're not shooting bright subjects (that cause your pupil to contract).</p>

<p>Vijay, I see what you're saying about the 26 megapixel camera equivalent. Fair enough.</p>

<p>Also, for the same film, I don't get it: doesn't the larger format, by definition, have to have 'higher resolution' when scanned because you have more 'pixels' (haha) I mean more dye clouds representing the same feature (say an eye) compared to the number of dye clouds that would be representing said feature in 35mm? Or are you saying that because the lenses for larger formats are so much worse than 35mm, that it's all lens-limiting? I'm really confused.</p>

[vijay_nebhrajani]

<blockquote>

<p>Rishi: Or are you saying that because the lenses for larger formats are so much worse than 35mm, that it's all lens-limiting?</p>

 

</blockquote>

<p>Yes, absolutely. Film has the same absolute resolution regardless of format, unless you include some degradation effect from the thicker estar base in LF film. So the difference is the lenses. If a lens is going to provide the circle of coverage that LF lenses do, it can't perform as well as a 35mm lens. If you look at the MTF of LF lenses, it is worse than that of small format lenses. So it does become lens limiting. Also, LF has to be used at really small apertures or your depth of field is a few millimeters - f/22 or smaller for 8x10, and you hit some diffraction related effects too.</p>

<p>So there is a tradeoff, and film no longer becomes the limiting factor. Rather, it is the rate of falloff of lens MTF versus increase of ability to resolve finer detail because of increase of area. One curve decreases as the other rises; so there would seem to be some fairly fixed upper limit on quality.</p>

<p>I just think that 40 MP is way too low for 8x10.</p>

[rishij]

<p>OK cool, thanks for that Vijay. Makes a lot of sense. Looking forward to your response on the dynamic range thread.<br>

Rishi</p>

[rishij]

<p>Guys,</p>

<p>Here's some food for thought re: digital vs. film performance on resolution test charts.</p>

<p>In this thread, we keep dogging film for its great performance on black & white resolution test charts, constantly reminding ourselves that film would perform much worse at lower contrast. But why aren't we dogging digital just the same?</p>

<p>Digital cameras use Bayer pattern sensors. For black, white, and gray data, they can effectively almost reach their physical resolution (read this: <a href="http://www.outbackphoto.com/dp_essentials/dp_essentials_05/essay.html">http://www.outbackphoto.com/dp_essentials/dp_essentials_05/essay.html</a> ). However, only half as much resolution exists for green data, and a quarter as much information for red and/or blue data.</p>

<p>A quarter the resolution? Hmm... sounds familiar:</p>

<p>Performance of Velvia:</p>

<ul>

<li>@ 1.6:1 contrast :: 80 lines/mm (5.53 MP equivalent)</li>

<li>@ 1000:1 contrast :: 160 lines/mm (22 MP equivalent)</li>

</ul>

<p>So, at full contrast, we have 22 megapixels worth of data, but only 1/4 as much at the lowest contrast.</p>

<p>But color film's resolution doesn't change for gray vs. colors.</p>

<p>So maybe test chart performance of digital cameras is just as indicative of 'real-world' resolution as test chart performance of film is!</p>

<p>For that matter, DLT, if you're reading this, can you think of an explanation of WHY resolution drops off so much for lower contrast scenes? Anyone? I don't know that just throwing out 'b/c it's a halftone process of binary information' is good enough... color dyes certainly aren't 'binary', and though silver atoms are (either metallic or not), different tones of gray <strong>are</strong> formed by density of deposits.</p>

<p>Rishi</p>

[rishij]

<p>For that matter, does anyone know the 1.6:1 contrast resolution for any digital SLR? Has it been measured somewhere?</p>

[rishij]

<blockquote>

<p>Also, if absorption is mediated by mobile electrons, and you have X electrons within a given area of film, but you shine 2X photons at this same given area that arrive <em>at the same instant</em> , then, would your surface be able to absorb all those photons or would some go through? What if you shone 10X, or 100X, or 10,000X photons at this same area containing X mobile electrons? How would all those photons still be absorbed at the same instant?</p>

</blockquote>

<p>I'm posing my question again for some optics/engineer/physicist to tackle. Rich?? You still here?</p>

<p>I have a feeling that my error might be in the assumption that 10,000X photons can arrive <em>at the same instant</em> ? Light is more like a flux of photons? So perhaps electrons can be excited, come back down, and then absorb another photon... and this can happen on a timescale (short) that makes the object effectively opaque?</p>

<p>-Rishi</p>

[bernardwest]

<blockquote>

<p>However, only half as much resolution exists for green data, and a quarter as much information for red and/or blue data.</p>

 

</blockquote>

<p>Rishi, I don't reckon it is as clear cut as this. The demosaicing will result in interpolation of colour data. Obviously this won't give 100% accuracy, but neither will it give 0% accuracy. It will be somewhere in between these two.</p>

[rishij]

<p>Bernie!</p>

<p>Thanks for being one of the, uh, one remaining still engaging in these esoteric discussions :)</p>

<p>I think we've boiled it down to a phase misalignment that would lead to 4000ppi scanners not really extracting 4000ppi of detai out of film (where 4000ppi represents the limit of film's ability to record high contrast -- 1000:1 -- detail)... I think you're right here about demosaicing & interpolation resulting in better results than just 'a quarter the resolution'.</p>

<p>This all probably explains, in part, why digital really does look better than 35mm film counterparts. Remember for 'real world' contrast (1.6:1), film is only 5.5MP, at best.</p>

<p>Does anyone know anything about 1.6:1 resolution test charts? Do these exist?</p>

<p>Rishi</p>

[bernardwest]

<p>Vijay.... Now <a href="http://tesladownunder.com/TeslaColorREDBilat00811000.jpg">this</a> is what I call a Faraday cage!</p>

[rishij]

<blockquote>

<p><em>So, for 35 mm images, effective pixel count is typically on the order of 4.5 megapixels (40 divided by 9) for colour images.</em></p>

</blockquote>

<p>Yeah Vijay, definitely disagree with Merklinger here. In fact, to me, he's just all over the place in his discussion, though he does have some valid points (such as the resolution drop when scanning film). But 2.6MP for a final 35mm scan? That's a joke.</p>

<p>But, to get right down to the point, why's he dividing 40MP by 9 to get the MP rating for 35mm color, as opposed to B&W, film? That's inane, to say that B&W is 40MP but color film is 4.5MP. I can understand him dividing 40 by 3, since he said earlier that the resolution of color film is about a third that of B&W (which is probably itself debatable). So that'd put color film at 13.3MP. But divide by 9? Huh? Can anyone explain the logic there?</p>

<p>Finally, looking at Mauro's film shots of resolution charts under the light microscope, Velvia color slide clearly resolves ~21MP of information on a 35mm frame for 1000:1 contrast. For real-world contrast, it falls anywhere in between 5.5MP and 21MP.</p>

<p>Man, if you're not objective enough, you can rationalize almost anything.</p>

<p>And I pose this question again (which I posed on other threads): Is there such a thing as a lower contrast resolution test chart? Like a 1.6:1 contrast resolution test chart?</p>

<p>Rishi</p>

<p>P.S. WTF is up with the quotes feature on the website now? It's all jacked up! I can't get out of the quotes without altering the HTML.</p>

[bernardwest]

<blockquote>But divide by 9? Huh? Can anyone explain the logic there?</blockquote><p>

 

I haven't read the Merklinger article, but I wonder if this has to do with the concept of vertical and horizontal resolution? As mentioned a lot, I don't give much of a fig about all the technicalities of resolution, so I can't be definitive here. But I do recall reading a thread recently about resolution being tied to either of the horizontal or vertical image dimensions. That is, if you compare say a 10000 pixel image to a 5000 pixel image, the resolution difference isn't half as you might intuitively think, but is infact less than this. Say the dimensions are 200x50 pixels and 142x35 pixels. In either the horizontal or vertical dimensions the resolution difference is (200-142)/200 = 29%.<p>

 

So, I am wondering if the reason he is dividing by 9 is because it is a resulting resolution of 1/3rd in EACH dimension. ie. 1/3 x 1/3 = 1/9. This is just a guess, and from memory, the thread I refer to contained some counter arguments to this approach to resolution (none of which I bothered to think about much because, as I said, I don't really care).<p>

 

Re. the gripe about the quotes thing - you might already know this, but you can opt out of the new text box editor and revert to the old way by default. I did this, as the new editor just caused too many hiccups. You can change the preferences in your workspace homepage.

[rishij]

<p>Yeah but that would mean that he was initially saying that the resolution of color film is 1/3 that of B&W <em>in each dimension</em> (both horizontal & vertical). That's retarded. That doesn't make any sense. And I won't believe it until someone provides me with some evidence for it.</p>

<p>Right here under the microscope, Kodak T-max is approaching 26MP worth of resolution, and Velvia 50 is approaching 21MP worth of resolution, looking at Mauro's shots of test charts.<br>

<br /> So empirical evidence would suggest that Merklinger is way off.</p>

<p>Bernie, I think what you're thinking about is this: when you view an image at 50% (or resample to 50%) in photoshop, it's actually at 1/4 the resolution, since you're cutting both dimensions in half, and 1/2 * 1/2 = 1/4.</p>

<p>Rishi</p>

<p>P.S. Never got a chance to make it thru Mees over the holidays b/c I got too interested in motion judder & motion blur as they relate to modern LCD technology, and how dark frame insertion and all this other crap is supposed to fix it. Anyway, sometimes I'm scatter-brained. Which is probably why I'm on this ridiculous thread to begin with :)</p>

<p>Long live this thread! Haha.</p>

[bernardwest]

Is the T-max black and white? If it is, then hunt around and see if you can find those elusive gray grains. We never did reach a satisfactory conclusion. Or did we? I'm can't even remember if this thread or that other one is the gray grain thread!

[rishij]

<p>I don't have a powerful enough optical microscope here (that I know of).</p>

<p>That being said, under the microscope, under these sort of what I'll call 'macroscopic' magnifications, dark areas are just areas of more dense silver deposits... light areas just have a lower NUMBER of interspersed silver deposits (lower density per unit area).</p>

<p>So the halftone process as an explanation, to me, is irrefutable.</p>

<p><strong>HOWEVER</strong> , looking at the smallest individual specks I can see, none of them look 'truly black'... I mean, what is 'truly black'? In light areas of the film (almost clear), I still see some silver specks... but they're not like some 'pure black' absorbing all light hitting them. I mean, think about it, if a sheet of metal is thin enough, it'll transmit light.</p>

<p>Same here. The size and density of the silver specks determine how dark that area appears. Some specks are larger than others. Some appear denser than others when viewed from a distance because silver from, probably, multiple sensitivity sites merged. So yeah, I guess different silver specks can take on different tones when viewed from a distance, but it's a pretty pointless debate.</p>

<p>Maybe the more relevant question is 'for every individual silver deposit that grew from one sensitivity site, can there be different densities of the resulting speck'. My guess is yes. Will these different-density specks contribute differently to tones formed at the macro-scale? Yes.</p>

<p>So, no, I can't really say it's binary, in the end. I also can't say that one speck itself from one sensitivity site can lead to thousands of tones. No way. But maybe all the permutations of developed vs. undeveloped sensitivity sites corresponding to an area of film the size of the original grain can give rise to hundreds of tones... but again, limiting ourselves to a discussion of the area of film corresponding to one grain is kind of dumb... because a grain is just there to present sensitivity sites, and with stacked layers of film, areas of grain can overlap, so sensitivity sites themselves are dispersed throughout the entire film... it is the exposure and development of these tiny sensitivity specks that determines the density of film at any given area. And this is a halftone process where the resolving element can be of any size and perhaps of different densities.</p>

<p>So, it's largely analog in nature. Pun intended.</p>

[steven_kokensparger]

<p>Whoever looks at the article which started this thread and says "it's an argument about the superiority of film over digital" is, of course, missing the point entirely and is going to conclude, as every good Computer Programmer should, that "it's a load of hogwash."<br>

I enjoy programming and I enjoy photography. Programming is not photography and photography is not programming. Computer Programmers think photography is about "capturing" an event. Photographers think about photography as painting with light and actually "creating" the event through the mind's eye. For Programmers, the equipment makes the shot. For photographers, the equipment is but a means to an end.<br>

I started out in photography a long time ago and bought my first SLR camera when I was 15 years old. I had admired my uncle's Hasselblads, but could only afford a Pentax K-1000. I haven't purchased a different film SLR since. I've had showings, won art shows, even went to college for photography before taking a sabatical to go to law school and start my practice. I've had the money to buy a different camera, but I've never really had a reason. The Pentax K-1000, with the right lens, lighting equipment, film, developing equipment, paper, and photographer, can do anything any other film SLR can do. The only difference is I don't have to fuss around with all of the automatic programming crap that goes along with more expensive cameras. <br>

I equate it to word processing programs. The only thing I leave on when I buy a new Office Suite of software is the automatic spell check. I do not even allow the program to change the spelling of the word automatically, because I may have meant to spell it that way. There is an art to writing just as there is an art to photography. I let it alert me to a spelling error by underlining the word and if I did not mean to spell it that way, I'll look up the correct spelling. You would be surprised how many briefs I would have screwed up had I allowed my word processor to simply change what I have written automatically.<br>

What makes me chuckle is how many articles I've read on line which refer to the Pentax K-1000 as a "beginner's camera" simply because it was the choice of photography teachers for their beginning photography courses. The Pentax K-1000 wasn't chosen for these courses because it was for idiots, it was chosen because it was a fine camera which also just happened to be FULLY MANUAL.<br>

I know plenty of programmers are going to talk about shutter speeds and mega pixels, but I've also used the Pentax K-1000 to photograph football games from the sidelines and horse races. I've also used other cameras, both film and digital. One of my finest pieces from the K-1000 was a sepia toned print of a horse race where you could make out the individual hairs on the horse's mane as he raced past ... you just need to know how to use your equipment to get what you want.<br>

Is a math student a better math student because you placed a calculator in his hand? Not necessarily, but if he has the basic "craft" down, then he sure will use the new tool you gave him to make his work easier. Similarly, an "advanced" photographer isn't an "advanced" photographer just because he knows how to work his way through the menus of a company's firmware in order to disable the automatic aperture setting on his dandy new fully automatic Minolta, he's an advanced photographer because he knows how to work with these options as tools to order to further his art.<br>

I'm ranting in this way because I logged on recently in order to see what everyone else is doing through the forums online. What I've found were voluminous discussions of which high priced lens to buy for a particular model of new digital SLR. I'm definitely am not going to argue that a quality lens is not important, but I'm stunned when an advanced photographer is arguing to "avoid that lens at all costs because it it compatible, but the auto-focus will not work." ... What??? Who cares? If you can save some money, get a higher quality lens, focus the lens yourself, you'll get your own results rather than the snap shot the manufacturer thinks you want.<br>

No offense is intended here, as I have also recently purchased my own digital slr to experiment with and I am also a computer geek. However, I wonder how many of those digital camera enthusiasts who consider the K-1000 to be a "beginner's camera" could actually get a print to turn out from using one. <br>

My new digital SLR was expensive, and for good reason! It's so advanced, it will take a quality picture to the manufacturer's specifications whether I am holding it or off taking a crap somewhere. It doesn't care whether I am involved in the process or not, and I am sure I am going to spend countless hours wrestling with it to return it to a fully manual status before I become accustomed to letting it do some of my thinking for me. I would note, that I've taken several hundred digital photos from other cameras and many so far from this one and I have yet to agree with more than a couple of the results from the "auto correct" option in the software that comes with any of them. How then, can I trust the "auto focus" and "auto aperture" settings in the camera itself?<br>

It's a new world, but I think we need to remember why things work the way they do in our cameras so we can refuse to allow the manufacturer to usurp our role as photographers.</p>

<p> </p>

[rishij]

<blockquote>

<p>but the auto-focus will not work." ... What??? Who cares? If you can save some money, get a higher quality lens, focus the lens yourself, you'll get your own results rather than the snap shot the manufacturer thinks you want.</p>

</blockquote>

<p>Tell me you're joking. Are you trying to imply that manual focus is somehow inherently better than auto-focus?</p>

<p>Do you understand the principles of manual focus; i.e., contrast and/or phase detection? If you do, then tell me how our eye-brain system is somehow better at doing edge detection on-the-fly, while trying to figure out, oh I dunno, exposure, composition, etc. Or explain to me how our less-sensitive eye-brain system is supposed to perform this contrast/edge detection in darker scenes, with small objects in the composition as the focus target, when a more-sensitive CCD/computer could do it in a fraction of the time?</p>

<p>Hatin' on auto-focus is truly hatin' on progress. It's absolutely inane. A highly-tuned sensor/algorithm is just much better at it than we are. Even with exposure, you can't tell me the world of photography would be a better place without light meters. Sure I use manual, I even have a fully manual Minolta X-700 SLR, but for any shot that I actually care about, I whip out my Canon EOS-3 (manual mode, of course), and start partial metering (using eye-control, hehe) different spots around the frame... the sky, the foreground, the sun, etc. and arrive at an ideal exposure estimate. But that doesn't mean I didn't use the massive help of the camera's meter, and the camera's auto-focus, and the camera's auto exposure-bracketing set to +/- 2/3 of a stop... all so I could focus on the more important aspects of photography.</p>

<p>Composition, for example. And, especially with the highly finicky (small exposure latitude, high gamma) Velvia slide film, exposure.</p>

<p>I understand the limits of automated systems very well. But there's a limit. There's <em>no way</em> manual focus is better than auto-focus b/c, at the very least, the auto-focus can act as an <em>aid</em> : it most probably gets the focus right, then you can fine tune, if you want, with the manual focus ring (on L series lenses anyway). Since most of the time the camera gets it right (well, sometimes after a few trials anyway), auto-focus is a huge advantage. Even if it doesn't get it right, you can help it out by getting it in the 'ballpark'. From there the camera's auto-focus can fine-tune. Our eyes are just not as good at fine-tuning at this point, largely b/c our the focus target is too small in the frame for the resolution capabilities of our eyes to discern small shifts in focus.</p>

<p>Which is why rangefinders were developed, to aid in focusing. Are you going to diss that too? B/c auto-focus is just a step beyond that... think of autofocus as, albeit a simplification, the alignment algorithm that aligns the edges in a rangefinder-like system. Do you really think that somehow our eyes are better at that than some simple software algorithm? Especially in a low contrast scene? In the dark?</p>

<p>You gotta be kidding me.</p>

<p>Use tools. Feel Human.</p>

<p>:)</p>

<p>-Rishi</p>

[James G. Dainis]

<I>"Tell me you're joking. Are you trying to imply that manual focus is somehow inherently better than auto-focus?"</I><P>

 

What he is saying is that if you can't get an auto focus lens, get a manual focus lens and focus it yourself and save money. I don't know if you are deliberately misconstruing what he said to start another argument but this thread is now closed.