Poor scans - May explain why some switch to Digital

[swilson]

<blockquote>

<p ><a href="http://www.photo.net/photodb/user?user_id=643213">Sergiy Podolyak</a> <a href="http://www.photo.net/member-status-icons"></a>, Mar 26, 2009; 07:10 p.m.</p>

 

<p>Ok, Scott, lets talk about tonality.<br />I dont like the tonality of yellow color on you mega-super 200MP shot.</p>

 

</blockquote>

<p>Well first off 200MP is not a mega-super image, for that you really need to be over 1GP.<br>

But as to the yellow, a closer view shows just how much dirt ison the front loader.<br>

<a href="http://www.pbase.com/konascott/image/110649762/original.jpg"><img src="http://www.pbase.com/konascott/image/110649762/original.jpg" alt="" width="700" height="700" /></a><br>

What color it is under all the dirt I don't know, but that is the color as seen with the dirt.<br>

 

<p>But tonality is not the same as color rendition, if it were then no B/W print would have good tonality, but this is clearly not the case. Tonality is the ability to see small changes in either color or brightness in a photo.</p>

</p>

[rishij]

<p>Yeah, um, Sergiy's comment was entirely irrelvant, Scott. I'm surprised you deigned a response :)</p>

[swilson]

<blockquote>

<p ><a href="http://www.photo.net/photodb/user?user_id=423641">Landrum Kelly</a> <a href="http://www.photo.net/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/2rolls.gif" alt="" /></a>, Mar 26, 2009; 07:10 p.m.</p>

 

<p>Mauro, I see that the specs given on DPREVIEW.COM for the images from the Nikon D3X are 6048 x 4032 for a total of about 24.5 MP . For the Canon 5D II the specs are 5616 x 3744 for a total of about 21 MP.<br>

Above I have heard you or someone else say that the actual megapixels delivered by digital cameras are substantially less than such official numbers. On what basis do you say that? I don't understand.</p>

 

</blockquote>

<p>It is almost all due to the Bayer Pattern used to get color from the CCD. Just how much resolution is lost depends a lot on the scene. <br>

Down sizing an image from a camera that uses a Bayer pattern will result in a sharper image. At some point there is no more sharpness to be had with more down sampling. If I use a good lens I seem to hit this point with down sampling at about 83%, which means I now have about 69% of the orginal pixels. But as I said how much you need to down sample depends on the scene, the worst I have come arcoss is red lettlers on a blue background, in that case I have to down sample by close to 50% to get a really clear image.</p>

[rishij]

<p>BTW, Scott, I agree: tons of noise in the sky in my image. Yes, such noise mucks with tonality. But, then again, I never scan (anymore) with the harsh LED + collimator light source of the LS-4000/5000... I use a diffuser in the light path which really helps with the grain.</p>

<p>Furthermore, the mitigated grain is then more amenable to removal by noise-removal software, resulting in pretty clean skies.</p>

<p>But nothing near the clean skies that MF or LF would result in if downsized to a 35mm image... can't beat multisampling, as PhotoAcute itself shows.</p>

<p>But, then again, can't beat the convenience of a 35mm camera over MF or LF.</p>

<p>I am aware that my waterfall image has huge jumps in tonality... which is what makes it harder to detect the effects of grain... which is why so many confused it for a digital image. Smooth tonal transitions in 35mm require a good deal of cleanup, even with slow film.<br>

-Rishi</p>

[swilson]

<blockquote>

<p ><a href="http://www.photo.net/photodb/user?user_id=2381463">Rishi Sanyal</a> <a href="http://www.photo.net/member-status-icons"><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/2rolls.gif" alt="" /></a>, Mar 26, 2009; 08:19 p.m.</p>

 

<p>Yeah, um, Sergiy's comment was entirely irrelvant, Scott. I'm surprised you deigned a response :)</p>

 

</blockquote>

<p>

<p>Through all the back and forth bickering that goes on we also, slowly, I think, are getting a better understanding as to the limits of the different forms of photography. I know I have come away with things that I did not know before. Whether Sergiy will have an open enough mind to learn anything new in this thread I don't know, but I am willing to give him the benefit of the doubt.</p>

</p>

[rishij]

<p>Scott & anyone else:</p>

<p>Why haven't they started building cameras using these:<br>

http://colorlink.com/Page/snapwave.htm</p>

<p>A color filter wheel with no moving parts. Seems you could get an ultra-resolution camera by placing this in front of the sensor, and switching between red, green, and blue transmission. No Bayer pattern necessary...<br>

-Rishi</p>

[swilson]

<p>

<p>BTW I a while ago I wanted to see just how far I could push a digital image and still have reasonable tonality. The link below is about the limit, down load the photo and look at its histogram.<br>

<a href="http://sewcon.com/temp/6bit.tif">http://sewcon.com/temp/6bit.tif</a><br>

That image uses only 64 level for each colors, which looks like a comb when viewing the histogram. Looking at the sky you can see that it does not look exactly right, but not bad either.</p>

<p> </p>

</p>

[rishij]

<p>I bet you could also swap in some ND filters in the path using something like that to get a greater dynamic range...<br>

-Rishi</p>

[Landrum Kelly]

<p>Thanks to everyone for their answers to my sometimes dumb questions. I ask a lot of questions because I really don't know the answers, but I have been trained enough about suspending judgment that I don't always swallow the answers I get--nor do I reject such answers. I simply try to hold the competing views in my mind until I can make a judgment for myself.</p>

<p>That has been the beauty of this thread, that the numerous competing exchanges have generally been the most civil and enlightening (or at least provocative) that I have seen anywhere on the forums, and trying to keep up with you guys has been and continues to be a great learning experience.</p>

<p>I will say this: the more I read about film, the more I understand digital. I am sure that that is all the more true for those who not only read about film but who regularly have used it, as I really have not in a great many years (and I was never much more than a snapshot shooter with it). That is one reason that I want to try to get more into the use of film and film scanning. I see it as very nearly essential for my growth as a photographer, regardless of the medium employed. Scanning and other processing issues force one to come to grips with theoretical issues that never face those who happily snap away with only digital cameras.</p>

<p>I think that there is truly a lot to be said for using both.</p>

<p>--Lannie</p>

[mauro_franic]

<p>Lannie, Scott explained the resolution question. AA filter, noise reduction and demosaicing also resolution.<br>

Scott, the smoothness on tonality from a DSLR is not an advantage, it is the result or of averaging color information. </p>

<p>On film, you can do this as well. It all depends on film size and print size; if you are to print from 35mm at 16x20 or larger, and depending the paper, and you see grain on the print, then you can reduce it with software. Also grain, at least to my eye, is always more pleasing thant pixels because of the randomness in shape and size.</p>

<p>I wish I had taken that picture of the little girl with the 40D along with Tmax to show you side by side that much of the smootheness is actullaly erased tonal gradation (and obviously some areas would be just flat whit due to DR limitation).</p>

<p> </p>

[swilson]

<p>Nasa use to use color wheels for the video camera, which is why you would see a whole lot of color for moving objects.</p>

<p>While I was at Tektronix we make displays using a color shutter, which is much like the SnapWave device, but we used it for output rather then input. <br>

 

<p>We put a monochrome monitor behind a color shutter, it is far easier to get really high resolution with a monochrome display then it is with color, at least it was in the days of CRTs.</p>

</p>

<p> </p>

[swilson]

<p>shoot ended up with a double post, sorry</p>

[mauro_franic]

<p>To ilustrate, this is Rishi's image with noise reduction:</p>

<div></div>

[mauro_franic]

<p>Now is that making you believe you have captures better tonal gradation. The prints will not agree.</p>

[mauro_franic]

<p>In Rishi's scan, I don't believe it is grain I am seeing. The size seems too large for Velvia although it could be due to the particular reaction to that color in the sky but I see it on the mountains too.</p>

<p>Rishi, did you observe that?</p>

[rishij]

<p>Mauro, I don't understand the sentence structure in your last post -- please rephrase?</p>

<p>Also, thank you for the NI treatment :) I was gonna do that and post back to appease Scott's distaste for grain, but got sidetracked by my job, haha.</p>

<p>NI does a good job at retaining detail while getting rid of noise, but I hate those stupid isotropic crystal-like arrangement of noise it enhances. ARGH!</p>

<p>The problem, though, is that some of the areas where grains 'clumped', or grain aliasing occurred, end up still showing up after NI noise removal. Look at the magenta sky in the noise-reduced image... you'll see clumps of darker sky. If you can't see it at first, just scroll up and down and your eyes will pick up these clumps moving back and forth. This sort of 'noise' cannot be removed by noise reduction software... and I run into it every time I try to reduce grain in the sky.<br>

<br /> It's better with the Scanhancer though, as the diffuse light source to begin with helps. Also better with the LS-9000, due to its diffuse light source.</p>

<p>Mauro, I've done my own comparisons of LS-4000/5000/9000, yes. They all have pretty much the same resolution, as tested by your Velvia frame. LS-4000/5000 have more 'microcontrast' due to lack of diffuser... but now I scan with Scanhancer all the time so it's the same as the 9000. 5000/9000 have much better DOF so sometimes you can actually nail focus across the full frame without flattening the film... but my success rate is something like 20%. That's unacceptable, so I use glass. The LS-4000 is useless without glass flattening the film due to its dismal DOF. However, to use AN glass you have to have a Scanhancer in there, and the Scanhancer causes a lot of IR falloff around the edges, which causes the AN etch pattern to show up upon IR channel subtraction. So, basically, I haven't gotten either the LS-4000 or LS-5000 to work yet with Scanhancer + AN glass + ICE. For now, it's been no ICE. I believe I'd have to upgrade the IR source of the LS-4000/5000 to get it to work with Scanhancer, as, right now, it's a point source & is too weak or does not have even enough illumination across the field (maybe the condenser is not as effective with IR... there's certainly spectral dependence of refraction, so...)</p>

<p>LS-5000 light source:<br>

<img src="http://staff.washington.edu/rjsanyal/Photography/35mm_Scanners/Nikon5000-LightSource.jpg" alt="" width="564" height="800" /><br>

You can see the point source towards the bottom of the image. The condenser collects & collimates the light before it is bounced down by the mirror. Thanks Erik de G (of Scanhancer fame) for the image :)</p>

<p>On the other hand, here's the Minolta DSE 5400's superior light source (CCFL):<br>

<img src="http://staff.washington.edu/rjsanyal/Photography/35mm_Scanners/MinoltaDSE5400-LightSource.jpg" alt="" width="800" height="600" /><br>

<em>Image courtesy: me :)</em><br>

<em><br /> </em><br>

Note the row of LEDs at the bottom... there's 6 of them. Those are the IR LEDs. They are collected and collimated by the element placed in front of them. Just the fact that there are 6 of them spread across the field results in more even IR illumination, I suspect. Because with the Scanhancer in place, I have no IR fall-off, and can scan with ICE. Probably wouldn't be too hard to incorporate this into the LS-4000/5000, but the biggest problem would be the distance the IR light would have to travel, as I'd have to put the LEDs behind the already-present point source. Hmm... maybe if I collimated the IR light first, I'd have less fall-off... Certainly a possibility. Dunno where to get these collimator/optics parts from though... anyone have a good idea?</p>

<p>Wow, that was a lot of rambling. Still there...?<br>

-Rishi :)</p>

[rishij]

<p>Well, Mauro, it's Velvia 100. More grainy.</p>

<p>If it's not grain, what'd it be? Grain aliasing? Perhaps you're not used to seeing it b/c you use a LS-9000. As I said, it's less pronounced when using Scanhancer.</p>

<p>Like I said, I hate the LS-4000/5000 light sources. They're absolutely dismal. Which is why a Minolta DSE 5400 is a bargain, if you can find one, for 35mm film. I wouldn't use any other small format scanner.<br>

-Rishi</p>

[Landrum Kelly]

<p>Speaking of noise reduction, I notiiced a long time ago that photos made with the original 5D that people often raved about were simply smoothed out beyond belief. We were told that it was the low noise sensor. There was some truth to that, but there was also the NR algorithm that was greatly responsble. The problem was even worse with a Kodak 14n that I once had. In good light it could be incredibly sharp, but one could never turn the NR completely off, and sometimes, even in the best light, the needles on evergreens in particular simply went toward a watercolor effect. It was very frustrating.</p>

<p>Here is one shot (a crop, actually) that I got some good comments on. The reality is that I was somewhat disappointed with how flat it looked. There was just so much more detail as well as variations in color tone in what I saw compared to what I captured. That said, it isn't a bad image. It just isn't particularly true to what I saw (or had seen up close rather than through a telephoto).</p>

<p>http://www.photo.net/photo/6649708&size=lg</p>

<p>In addition, the images from the 5D came out soft and had to be sharpened significantly using unsharp mask.</p>

<p>--Lannie</p>

[philipp_wetzler]

<p>Mauro,</p>

<p>I'm one of those people who kept hearing how 35mm film is much better still than the current selection of DSLRs but never really bought it -- because in all the scanned photos I saw, no matter the detail that's technically there, because of large amounts of grain and lack of sharpness the <em>perceived amount of detail</em> on film was always much lower (even though film grain has its own aesthetic appeal, but that's a different story). I kept this opinion even after spending some time trying to get good results out of a Nikon LS-5000. Even with 16x multisampling, and at different resolutions, the result couldn't really compare to what I got from my simple D40x.</p>

<p>Well, you've convinced me. Some of the scanned images people posted (for example your TMAX 400 and Ektar 100 scans above) are truly impressive, even at large magnifications. Now my question is: what do I need in order to get results like that?<br>

Let's not talk about what camera and film to use, that's been much discussed elsewhere. The big question is the scanner. You (and other participants of this thread) have established that having film scanned at a photo lab is not really an option. From my own experience I can say that it's not enough to just buy an expensive film scanner (like a Nikon LS-5000), either. This is also underlined by the many, many people who keep disputing the claim that film can compete against digital, because they've tried scanning film using expensive equipment and didn't get good results.</p>

<p>So, starting with nothing (and keeping to 35mm for now), what scanning equipment would you personally recommend to get for results like the ones you posted? Maybe the equipment isn't everything. Taking classes, maybe? Lots of practice? (How much?) What else would help?<br>

And finally, having all that equipment, how much work would you say is involved in getting one very high quality scan of one 35mm photo? How long does it actually take you, for the whole process?</p>

<p>I'm not trying to be sarcastic at all -- I know everything will take longer with film, and that expensive equipment is required, but I also know (and have been convinced by this thread) that it may well be worth it. But in order to decide <em>if</em> it's worth it for someone, let's talk about what it takes to get those convincing results.</p>

<p>Philipp</p>

[swilson]

<p>

<p>I have use programs like NI, but have not had a lot of luck with them.</p>

<p>For something like a sky it is pretty easy to remove the noise, from areas of an image that have more subtle texture you might not even see the texture against the noise.<br>

<br />In this image there is fairly subtle textures in the sand and water, going from a fair bit near the people feet do a very smooth area in wet sand on the right hand side. If there was much noise in this image it would be hard to get everything to look right using NI.<br>

<img src="http://www.pbase.com/konascott/image/110652379/original.jpg" alt="" /><br>

Link to full res photo<br>

<a href="http://www.pbase.com/konascott/image/110652509/original.jpg">http://www.pbase.com/konascott/image/110652509/original.jpg</a></p>

 

 

<p>

<p>In both my film scans and my digital shots, where I go to iso 800 or above, I have to deal with this, I pretty much leave the noise in the image, but grumble a lot. I would go for something like the 5D not for the extra resolution but rather the ability to shoot at iso 800 and above with a lot less noise then I am getting now.</p>

</p>

</p>

 

[Landrum Kelly]

<p>Scott, that beach shot is incredible at full size. What did you shoot it with, and how did you process it?</p>

<p>--Lannie</p>

[mauro_franic]

<p>Thanks Les. It is tempting to copyright the soup but I have to custom tweak it to each scanner like I did with the CS 9000. I will make one for the CS 5000 as promissed.<br>

Philipp, capture (and developing for B&W) is most of it. For scanning, I turn all adjustments off and just feed it through. Honest that is all I do.</p>

<p> </p>

[mauro_franic]

<p>Scott,</p>

<p>First regarding you picture, it looks like a nice place to be in.</p>

<p>But more importantly, after spending the last several days looking at pictures on this post, can't you tell by looking at your picture how much it is just a color average, amost like water painting, void of small detail? Looks at the blond hair, the trees or hotel in the back at 100%....</p>

[swilson]

<blockquote>

<p ><a href="http://www.photo.net/photodb/user?user_id=423641">Landrum Kelly</a> <a href="http://www.photo.net/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/2rolls.gif" alt="" /></a>, Mar 26, 2009; 10:46 p.m.</p>

 

<p>Scott, that beach shot is incredible at full size. What did you shoot it with, and how did you process it?</p>

 

</blockquote>

<blockquote>

<p>--Lannie</p>

</blockquote>

<p>Hi Lannie,<br>

It was shot with a 350D using the 18-55m kit lens, at 18mm f/8, ISO 100, color temp 6250.</p>

<p>I used RawShooter essentials to convert the raw file, Exposure comensation set to -.63 and Sharpness set to 25, everything less set to the middle position.</p>

<p>In true it is often the lighting that matters the most in how sharp an image looks, the photo was taken at 9:30 ap in Feb. so the sun was no too high and the angle was pretty good, off to the side.</p>

<p> </p>

[mauro_franic]

<p>Scott, my previous post was meant in an inquisitive tone - nothing else. I hope it dosan't sound like a critisism of the picture.</p>