Poor scans - May explain why some switch to Digital

[robert lee]

<blockquote>

<p>Robert, I'm not sure of the workflow you are using here. Is this blending say three bracketed exposures ...</p>

</blockquote>

<p>For this kind of stuff, I generally script things up for execution on the command line. The example is a contrast blend of 15 frames: groups of 3, +- 1/3EV apart, 5 groups. </p>

<p>At its core, a blended pixel is some weighted average of the same pixel from all of the contributing frames. The more frames there are, the closer the blended pixel value is to the supposed true tonal value. Each doubling in the number of source frames is something like gaining one more bit of tonal precision.</p>

<p>I shot JPEG here because compression of high ISO imager noise is visually really nasty. This is as bad as it gets. If the frames had been taken in RAW (which, by the way is available on all Canon compacts via the CHDK firmware pack) the core of the script sequence would look like something like this:</p>

<blockquote>

<p>dcraw -w -H 0 -o 2 -q 3 -T IMG_*.cr2<br>

align_image_stack -a aligned IMG*.tif<br>

enfuse -o blended.tif aligned*.tif</p>

</blockquote>

<p>The align_image_stack is necessary when shooting hand held. It corrects mis-registration due to camera movement. If you're using a tripod though, remember to disable IS. The stabilization introduces its own mis-registration.<br>

Google dcraw and hugin. DCRaw is stand-alone software while Hugin provides the other two utilities.</p>

<p>By the way, the same technique should work just as well with film. This should get you noiseless small format film captures without noise suppression losses. Of course, once there, you'll probably find it looking awfully like a high quality DSLR frame.</p>

[rishij]

<blockquote>

<p>If you look at my profile you'll find I don't wear glasses . . . ;-)</p>

</blockquote>

<p>Hence why I suggested you get your eyes checked. 20/15 vision seriously rocks, let me tell you.</p>

[rishij]

<p>C Sharon,</p>

<p>Yeah sure, I have absolutely no confidence that a pro lab knows how to use a scanner properly. How many times have you read that a pro lab will 'color correct' your chromes by hand? WTF is that? You don't color correct chromes, you color match them using color management principles!</p>

<p>Furthermore, many 'pro labs' use Nikon scanners, and I've already mentioned a number of times that even Nikon themselves never got their scanners right nor did their software and/or instruction manuals even get <em>close</em> to outlining how one should properly use their scanners.</p>

<p>So, yeah, I doubt <em>all</em> scanning operators.</p>

<p>Furthermore, those links you point to are laughable. One of them shows stepping artifacts (WTF? never seen such crap out of <em>any</em> scans I've <em>ever</em> made) in the MF scan from the Minolta scanner, which scanned a measly 3200dpi, or, 59MP scan of the MF film. An entirely invalid test. The other link showed a Heidelberg Tango scan at, yet again, a measly 3400dpi resolution, or, 67MP scan of the MF film. A Nikon LS-9000, at 4000dpi, will give a 93MP scan of said MF film... that's not a negligible jump in resolution. Furthermore, Mauro & I have clearly shown considerable more detail in a 35mm scan at 5400ppi and at 8000ppi on an Imacon. So imagine how much you could actually get off the MF frame! The fact that the tester verified that only 1 line in that electrical tower was visible under a 12x loupe is <strong>juvenile</strong> ... for Christ's sake, use an optical light microscope man, then talk to me.</p>

<p>So Mauro's initial premise is absolutely valid... all these comparisons on the net showing how older dSLRs outresolve MF are <strong>a joke</strong> . I'm not speaking about the D3x or the 5D Mark II b/c I haven't done those tests myself yet... see? I'm not just a film fanboy, I'm being objective. You, or anyone, pointing to other people's threads claiming that somehow some dude you don't even know scanned properly is just like <strong>faith</strong> : <em>absolutely irrational</em> .</p>

<p>'Nuff said.<br /> -Rishi</p>

[rishij]

<p>Robert:</p>

<blockquote>

<p>The align_image_stack is necessary when shooting hand held. It corrects mis-registration due to camera movement. If you're using a tripod though, remember to disable IS. The stabilization introduces its own mis-registration.</p>

</blockquote>

<p>You're taking of aligning image elements, not simply pixels, correct? In others, mis-registration being where the same image element in the scene (one tree branch, for example) falls on different pixels from shot to shot?</p>

<p>But as I understand it, this mis-registration is beneficial.</p>

<p>PhotoAcute specifically mentions that even tripod-mounted shots have some movement, and this movement helps the software interpolate sub-pixel data. As I understood it, super-resolution actually benefits from what you're terming 'mis-registration'. It is precisely this 'mis-registration' that allows an array of pixels to sample image elements falling on this array that are smaller than the pixel elements themselves. For example, if half of a black line fell on one pixel (rendering that pixel 50% black, or, grey), but when the camera moved that line aligned with the pixel so the full width of the line fell on the pixel (such that the pixel registered 100% black), then that resultant pixel, when averaged, would yield 75% black... which'd basically increase the MTF at that point.</p>

<p>Now, of course you have to perform image alignment for this to work, where images are aligned based on actual scene elements. Otherwise you'd of course run the risk of averaging wildly varying scene elements: <strong>bad</strong> !</p>

<p>Now, when this alignment occurs, you can imagine pixels overlapping other pixels in a fractional manner (i.e. half of one pixel from one image is placed over one pixel in an underlying image, while the other half is placed over an adjacent pixel in said underlying image). This is where subdividing the images into twice as many pixels comes in handy. Hence, the 'super-resolution' process, where sub-pixel alignments are allowed.</p>

<p>Robert, is this what your software is doing? Because I believe it is exactly what PhotoAcute is doing. Blending without sub-pixel alignment would only reduce noise, not allow any increase in MTF of any image elements.</p>

<p>Also, yeah you could do it for film. Man it'd be tedious to do that plus AEB for slide film. Of course, if you did it at the level of scanning, you wouldn't get rid of film grain.<br>

-Rishi</p>

[swilson]

<p>It would be interesting to try it with film, with multible shots it should be pretty effective at getting the grain out, with out loss of low contrast detail.</p>

<p>

<p>I am pretty sure my film scanner is not repeatable enough for the stacking to work, images as very slightly stretched in some places and compressed in other. The stretched and compressed areas change from scan to scan. The effect is small, just a few pixels, but it would be enough to mess up getting registration of the whole image at one time.</p>

</p>

[rishij]

<p>Scott, you seem to be an EE person (sorry Mauro to be completely OT this time), so quick question: If I have a halogen bulb rated for 12V/24 watts, and I plug it into a 12V 3A power source, and the resistance of the bulb is measured to be 1 Ohm, will it attempt to draw 12A, thereby killing the power source, or just draw the 3A, which, at 36W of power, would fry the bulb?</p>

[swilson]

<p>

<p>The resistance of a filament is always much lower when it is cold, so there is an in rush of current when first powered up, in this case 12A. Once the bulb is to temp, which happens very fast, the current will drop down to 2 amps.<br>

Most power supplies will handle a short over current, so likely it would be able to run the bulb without problems.</p>

</p>

[StuartMoxham]

<p>I know that Rishi does not like many of the tests on the internet that have been linked to in this thread. However those results were valid for that person who made that test at that time (assuming they were not trying to bias the results). That was the scan that they had to make the test with at the time and those were the results good or bad.<br>

It would not be very wise to choose film or digital based on a few tests published on the internet, just because scans posted on the internet are bad does not mean that film is bad. On the other hand just because Mauro and Rishi are getting great result scanning themselves those results may not reflect your own especially if you are relying on a lab to do the scanning for you.<br>

Its really up to each person to judge for themselves using each medium and camera system in the way they would normaly photgraph.<br>

I made a print from the crops Mauro posted and adjusted the color and sharpened little on the d2x image. There are differences between the two images as would be expected. The D2x does not fair badly by any means considering it is resampled way beyond its native resolution. The D2x would not be my first choice for photographing maps but if I was a D2x owner I would be well pleased with these results.</p><div></div>

[rishij]

<p>Scott Wilson -- you 'da man! Thank you, got my light box up & running and my self-taught (ray) optics course can now ensue... at 2 a.m.</p>

<p>Oh time, thou art a heartless...</p>

[rishij]

<p>Stuart, you make some good points. But then these people shouldn't put up self-righteous titles & statements like 'Canon 1Ds Mark II vs. Medium Format Velvia film!'</p>

<p>Instead, they should state it as 'Canon 1Ds Mark II vs. my poor & limited scan of MF film'. People should know that if they have that great landscape shot shot on chrome years ago, that they can get more out of it. Not be disheartened about their own film shots due to these comparisons.</p>

<p>Yes pros may already know. But many may not. Certainly many amateurs may not know. Therefore these tests are misleading.</p>

<p>You're right though that the choice of any format is in the hands of the photographer -- if in his/her hands, film workflow stinks or requires too much effort, then that must be taken into consideration.</p>

<p>-Rishi</p>

[Landrum Kelly]

<p>Stuart, the rivers are still blue on the film scan (although barely visible on the smaller version you have posted), but not so on the D2X image. The relative clarity of the stars (country capitals) is also still visible.</p>

<p>Most of all, however, you have by now gone through so many "layers" of imaging that you are getting further and further removed from the original files, so that I don't think that a "photo of a print" then digitized and shown on our monitors is going in the right direction to show what you want to show.</p>

<p>That said, yes, many D2X owners would be quite satisfied with the results.</p>

<p>--Lannie</p>

[Landrum Kelly]

<p>Still, the handwriting does seem to be on the wall: although MF film is still beating digital in terms of resolution (and possibly other things), digital has now closed the gap so much that one wonders when many will say, "Why should I scan? Neither I nor anyone else can see the difference, even with the best 'pixel-peeping.'"</p>

<p>The question is whether FF DSLRS will ever catch MF high res film, and, if so, when?</p>

<p>Another question: at what point is anyone going to care anymore? I hate to sound cynical, but it was not so long ago (about seven years) that my Olympus E-20 with 5 MP and a built-in zoom Zuiko lens was almost state of the art--and we used to ask, "When will digital catch up with film?" The answer is that it already has at larger "formats" (if one wants to call stitched images or digital backs a "format" for purposes of argument).</p>

<p>When is it going to happen on a light-gathering medium 24mm x 36mm?</p>

<p>I don't think that we are there yet, but with 24+ MP full-frame sensors already on the market right now (Nikon and Sony), we have to be getting close.</p>

<p>If someone were starting out today in photography, is there really any compelling reason, based on what we have seen above (so far), to say, "You really ought to put your money and time into film processing and film scanning"?</p>

<p>I am not playing the devil's advocate here. For those who already have the equipment and the know-how, staying with film is still a very good option. But for the rest, newbies starting out with no experience of either, what advice could one give in good conscience?</p>

<p>--Lannie</p>

[alan_goldhammer]

<blockquote>

<p ><a href="../photodb/user?user_id=2381463">Rishi Sanyal</a> <a href="../member-status-icons"><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/2rolls.gif" alt="" /></a>, Mar 17, 2009; 09:42 p.m.</p>

 

<p>Alan - ha! It *is* in fact... well, medicinal chemistry technically, but mostly biochem/biophys -- I'm working on understanding the packaging motor of viruses...<br>

I'll make a deal with you: help me purify my intractable viral protein & I'll take some time out to write a book :)<br>

The sad thing is, by the time I (we, because certainly it'd be a joint effort with all the help & work provided by folks like Roger Smith, Erik de G, Mauro, Mendel Leisk, Scott Turner, Robert Budding... countless wonderful people in these threads have contributed to my knowledge & efforts) ever get around to writing one, film will probably be dead.<br>

I'll try anyhow, for the sake of posterity. But first I'm trying to finish my prototype holders for the Nikon LS-4000/5000 and the Minolta DSE 5400 aimed at holding film as flat as wet-mounting would allow. Still not sure if the concomitant light source modification required for the LS-4000/5000 would allow ICE functionality though... so I've got some kinks to work out. Famous last words.</p>

 

<p>-Rishi</p>

</blockquote>

<p>Rishi, Cool, my thesis work was in membrane biochemistry but over 30 years ago! just have a desk job these days in drug regulatory affairs and doubt I would be of much help in the lab. :-)<br>

Keep us apprised about the film holder for the 5000. I'm using the Nikon strip holder for the B&W negatives but it doesn't hold things completely flat which can be a problem is there is curl. I hope film never dies (though good printing paper seems to be more difficult to find). I continue to shoot both digital and film and it's interesting to to compare and contrast findings.<br>

Alan</p>

[james_symington1]

<p>"If someone were starting out today in photography, is there really any compelling reason, based on what we have seen above (so far), to say, "You really ought to put your money and time into film processing and film scanning"?</p>

<p>Almost certainly not - film is something that newcomers would go to only after having cut their teeth on digital. I haven't met a beginner in years who was remotely sympathetic to the idea of buying a K1000 and taking it from there as many of us have done (actually a Zenit EM in my case).</p>

<p>And indeed why would they want to start like that when everyone is taking fun pictures, putting them on websites etc etc. I think any desire to shoot film now, from the perspective of a beginner, is a sign of photographic maturity and curiosity about alternative processes which by definition a beginner would need to acquire.</p>

<p>In contrast, I am sure that I am not alone amongst the 'old-timers' (although I am only 40) to still feel, occasionally, that I am cheating when I use a DSLR. I know it makes no sense but I do get a pang of guilt from time to time. I'll get over it eventually...</p>

[mauro_franic]

<p>Stuart,<br /> Thank you for running the print test.</p>

<p>I see almost all the short comings of the digital file in the snapshot of the print you posted. <br /> Also remember that although on the film it represents how the print would look on a 24" printer, on the D2X it represents only an 18" tall print (because the D2X was frames on a narrower portion of the map vertically).</p>

<p>YOUR CONCLUSION IS:<br /> Is your conclusion that the differences are observable on the print 18 inches tall or not?</p>

[mauro_franic]

<p>Stuart, when you comment on your conclusion, please quote what you see on the prints (not the snapshot of it).</p>

[swilson]

<blockquote>

<p ><a href="../photodb/user?user_id=423641">Landrum Kelly</a> <a href="../member-status-icons"><img title="Subscriber" src="http://static.photo.net/v3graphics/member-status-icons/sub8.gif" alt="" /><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/1roll.gif" alt="" /></a>, Mar 18, 2009; 08:46 a.m.</p>

 

<p>Still, the handwriting does seem to be on the wall: although MF film is still beating digital in terms of resolution (and possibly other things), digital has now closed the gap so much that one wonders when many will say, "Why should I scan? Neither I nor anyone else can see the difference, even with the best 'pixel-peeping.'"<br>

The question is whether FF DSLRS will ever catch MF high res film, and, if so, when?</p>

 

</blockquote>

<p>

<p>This is a tricky question to answer since film and digital are different. For much of the shooting I do a long lens is needed and reach is more important then the total number of pixels captured, in this case my 20D will out do any film camera out there.</p>

<p>

<p>The test that started this whole thread is where film does best, a very controlled environment, a high contrast subject, camera tripod mounted and a result where resolution is the most important out come of the image. For this kind of photography it may be a while before a FF digital passes 6x7 MF, not because it would be hard to do so but because the market would be small for such a camera.</p>

</p>

</p>

 

<blockquote>

<p>Another question: at what point is anyone going to care anymore? I hate to sound cynical, but it was not so long ago (about seven years) that my Olympus E-20 with 5 MP and a built-in zoom Zuiko lens was almost state of the art--and we used to ask, "When will digital catch up with film?" The answer is that it already has at larger "formats" (if one wants to call stitched images or digital backs a "format" for purposes of argument).<br>

When is it going to happen on a light-gathering medium 24mm x 36mm?<br>

I don't think that we are there yet, but with 24+ MP full-frame sensors already on the market right now (Nikon and Sony), we have to be getting close.</p>

</blockquote>

<p>

<p>I know a die hard MF film user that reasonably bought the new Sony, he is now in the process of selling off $7,000 of both Hasselblad gear as well as his Nikon 9000 scanner. Keep in mind the Hasselblad is 6x6 not 6x7, but in any case the Sony is good enough that he sees no reason to shoot with the MF camera anymore.</p>

</p>

 

<blockquote>

<p>If someone were starting out today in photography, is there really any compelling reason, based on what we have seen above (so far), to say, "You really ought to put your money and time into film processing and film scanning"?</p>

</blockquote>

<p>

<p>There is one good reason to put your time and money into film, if you enjoy using film.</p>

</p>

 

<blockquote>

<p>I am not playing the devil's advocate here. For those who already have the equipment and the know-how, staying with film is still a very good option. But for the rest, newbies starting out with no experience of either, what advice could one give in good conscience? </p>

</blockquote>

<p>

<p>The advice I have given people thinking about buying into film is not to rush out and buy a $2000 scanner. You can get some pretty cheap MF camera that are decent and have the film professionally scanned. Go out with the MF camera and someone who has a decent DSLR and shot the kind of things you like to shoot. Get a few of your best shots scanned on a drum scanner and make a print that is the largest you might think of making and compare to what the digital guy got.</p>

<p>The other option is to join a camera club and view the prints that people are getting with high end DSLRs and with MF cameras.</p>

<p>But what does not work very well, IMO, is looking at photos of test charts or maps hanging on walls, I really believe it is best to look at photographs of the kind of things you plan to photograph.</p>

</p>

[mauro_franic]

<p>Scott, yes there is no replacement for direct experience.</p>

<p>Who is selling the 9000 and where? I know several people interested.</p>

[swilson]

<p>The guy is Alan Browne,</p>

<p>Here is his ad for the stuff<br>

<a href="http://www.aliasimages.com/Hasselblad500Forsale.html">http://www.aliasimages.com/Hasselblad500Forsale.html</a><br>

I know he really wants to sell it all together, I have not heard how that is going.<br>

He really like the Hasselblad, before getting the Sony FF DSLR</p>

<p> </p>

[jeff_z.]

<p>I'm late to this thread, but read it quickly with interest. The original premise makes sense to me, but perhaps this betrays my bias towards film to a certain extent. Shooting film is wonderful for me, I see absolutely no reason to switch for my photography. I won't recount all of the myriad reasons why, as all have been discussed many times before.</p>

<p>What motivated me to post is what Scott said in reply to Roman's thoughts, which were basically, that Nikon does not offer the Coolscan V anymore because perhaps it is not in their best financial interests to do so; they would rather have customers purchasing their latest digital cameras as there is more profit in going that route... Scott gave a lengthy rebuttal saying, basically, that this was not so, that "...there is no company or publication out to get film.." I think perhaps the most cogent point for Roman's argument relates to Scott's last comment, ".. In the early part of this decade the digital camera manufactures were mostly trying to get people to switch from shooting film to digital, but now they are mostly trying to get people to upgrade from older digital cameras to newer ones."</p>

<p>Imho, Roman hits on one of the most important points- I think his post strikes at the heart of the whole larger debate. That is, that there is huge money at stake... Once the various parties involved got the ball rolling on the selling of digital cameras, their business models were radically altered... They now are focused on selling the "latest and greatest" digital cameras which "need" "upgrading" much more often than film camera technology. Then, there are the many digital photography accoutrements to sell... These companies have largely and consciously moved to a "razor blade business" model, much more lucrative than the old. For people like me using film cameras ranging in age from 17 to 40 years, and with no plans to buy new, there just isn't any profit... I think this is a very valid point. </p>

<p>And the magazines not being involved, too?? Well, they've <i>always</i> derived very substantial revenues from advertising; I remember being warned almost twenty years ago when I first became interested in photo that the articles were often worth reading, but that their ads will, "make you think that you need new equipment". The advent of digital photography has to have been like manna from heaven for them!</p>

<p>To Kodak's credit, they are continuing the research and development of film technology and have introduced truly great films in recent years. Hopefully, posts like this will help to lend the necessary financial support to these wonderful efforts. </p>

[mauro_franic]

<p>Thank you Scott. I will pass along.</p>

[rishij]

<blockquote>

<p>If someone were starting out today in photography, is there really any compelling reason, based on what we have seen above (so far), to say, "You really ought to put your money and time into film processing and film scanning"?</p>

</blockquote>

<p>HECK NO. I like to think of digital as a well engineered process built from the ground up. Film is more 'dirty, random, wet' chemistry that was tweaked, like much of chemistry & biology, to do what we wanted it to do. Digital will surpass film, save for perhaps the largest film formats, b/c, in my limited knowledge, an 8" x 10" electronic sensor seems a little ridiculous (if not ridiculously expensive!). Therefore if you're starting off, <em>start digital</em> .</p>

<p>But that doesn't mean that many of us that have been shooting film won't continue to do so, myself included, for the time being. It also doesn't mean that film technologies should desist, b/c like I've said many times before, many of us have precious film lying around that we wanted digitized in this 'digital age'.</p>

<p>Jeff, you make good points. I agree that the business model changed drastically. This is evident in film scanner projects being killed. Look at the new film scanners coming out today. They're <em><strong>toys</strong> </em> . Scanning circa 2MP images from your film, but selling for a measly $100-$200. If the general public is somehow, appallingly, satisfied with these new products, just to keep digital snapshots of their old memories, then how can Nikon & Minolta/Sony justify pouring more money into R&D and production for $1000-$2000 film scanners? Don't get me wrong, I really wish they did. But the market just doesn't seem to be there. The market is furthermore frustrated by the fact that these $1000-$2000 scanners are typically bought by discerning photographers, & discerning photographers <strong>do</strong> care about things like film flatness, soft light/grain reduction, etc. But not one of these scanners ever got all those things right (some of them got pieces of 'em right, sure). So I think there was substantial stigma to even entering the world of 'higher-end desktop scanning'.</p>

<p>In an ideal world, scientists & designers over at Nikon/Minolta would care enough, like Mauro & I do (I think I can safely speak for you, Mauro, here), about bettering film scanning, and the business side of things would actually listen to them, and a new, user-friendly, high-resolution, non-headache-inducing scanner would be designed and released. One where you never worry about film flatness or grain grittyness. One where <strong>you the user</strong> never have to worry about color accuracy; the scanner worries about it. One that would actually extract the full dynamic range present on the film using simple ideas from 1993-1994 patents that have, now, expired. One that would jump from 4000ppi to 8000ppi resolution even in a desktop scanner using a 4000ppi CCD employing recently developed technologies that I don't wish to get into here. Etc. etc.</p>

<p>That would be a wonderful world. Just not the one we live in.<br>

-Rishi</p>

[mauro_franic]

<p>Jeff,</p>

<p>you have good points, also due to advertising people have switched from film to Digital, not now (when the 5DII can give results comparable in resolution to 35mm film) but several years ago where digital cameras didn't come near the quality of film. </p>

<p>Those who did not give up landscapes with 35 film (an obviously MF) are left today with a library of much higher quality images than possible with 2MP, 3MP, 6MP, 8MP 10MP, etc cameras.</p>

<p> </p>

[david_littleboy__tokyo__ja]

<p>"(when the 5DII can give results comparable in resolution to 35mm film)"</p>

<p>What alternative universe are you stuck in? The 5D1 wiped the floor with 35mm, and the 5DII pulls even further ahead. The 5D1 makes lovely 12x18 prints; 35mm isn't even close at that size.</p>

 

[mauro_franic]

<p>David,<br>

You have missed out from the previous comparisons. This is how 35mm negative film (Ektar) compares to a 10MP DSLR (40D). The raw file from the 40D was processed by multiple people with different tools and the best was picked.</p>

<p>The difference is substantial enough that going from 10MP to 12MP will not make a difference.</p>

<p>Films like Velvia and TMX obviously have an even larger gap.</p><div></div>