Jump to content

Poor scans - May explain why some switch to Digital


Recommended Posts

  • Replies 799
  • Created
  • Last Reply

Top Posters In This Topic

<blockquote>

<p>Life is full of surprises Scott! Imacon seem to think cooling worthwhile:<br>

<a rel="nofollow" href="http://www.lexharris.net/documents/Flextight_848_brochure.pdf" target="_blank">http://www.lexharris.net/documents/Flextight_848_brochure.pdf</a></p>

<p>I don't recall how hot is was when I did that scan. The scanner (and me) mostly operate at comfortable room temperature so most likely that scan was done at a moderate temperature. Dark current as you have indicated is a function of temperature, but I'm not sure why you're suggesting a short integration time would lessen any need for cooling. Without knowing the architecture of the device I guess it's all speculation, but generally with CCD arrays I would expect dark current shot noise, photon shot noise and readout noise all to increase with reduced integration times.</p>

</blockquote>

<p>Dark current noise goes up by the sqrt of the integration time. Think of it this way, if you double the integration time you get twice as many electrons from dark current, twice as many electrons and you have sqrt(2) times the noise. Normally people don't start to cool CCDs untill the integration time is over a minute.<br>

Photon shot noise will go up with reduced integration times, but this can be offset with more light.</p>

Link to comment
Share on other sites

<blockquote>

<p>"Lex, you see, electrons .... they do not love each other. And this entire world is built on love.<br /> That is why *electronic* digital cameras have no future. Picture on film is formed, built by a group of particles who love the light and love each other, grouping under the light. Electrons spread as wide as possible from each other." --Sergiy Podolyak</p>

</blockquote>

<p>Sergiy, we understand that like repels like, etc. From this it does not follow that DSLRs have no future. If one thing is quite clear, it is that digital technology continues to advance.</p>

<p>Start the new thread that Lex suggested if you want to discuss this issue seriously.</p>

<p>--Lannie</p>

Link to comment
Share on other sites

<p>Lex, the black area of your scan has surprisingly low noise... lower noise than I've seen on our Imacon 848, lower noise than on my Minolta DSE 5400, LS-4000, LS-5000, etc. You are a lucky guy with a really lucky unit :)</p>

<p>Normally, I see green pixels creeping into dark shadows, on all these scanners, be it Nikon, Imacon, Minolta. Multi-sampling completely gets rid of these green pixels... the Imacon doesn't have multi-sampling, so the green pixels (shadow noise) persists. Lex, these are not those green streaks/banding I spoke of earlier... I'm just talking about random green pixels in blacks that show up on any scanner I use any time I brighten shadows or 'pixel peep'.</p>

<p>Perhaps the Imacon needs the Peltier because it doesn't have multi-sampling (the latter woulda been a much easier fix, of course).</p>

<p>Also, Lex, keep in mind that CCD temperature will rise the more you scan. So if you just turned on your scanner and made that scan in reasonably low temperature weather, I can see why shadow noise is absent. But after 20 minutes of scanning, I'd expect the temperature to rise.</p>

<p>It was just something I was toying with; not that important. Multi-sampling and a brighter light source are much more interesting prospects for me.</p>

<p>Also, you can cool Peltiers further based on the voltage/current you drive them at... you have to be careful though that the hot side doesn't get too hot and is properly cooled.</p>

<p>Again, in general, I think multi-sampling might be a better way to do this... would you agree Scott?<br>

-Rishi</p>

Link to comment
Share on other sites

<p>Sergiy,</p>

<p>I was attacking your idea that the realm of 'digital' itself was limited in the way it represented color.</p>

<p>Sure, Bayer patterns are limited in their color resolution. Charge build-up in the photo-diodes themselves may, and probably do, exhibit different profiles than on film... and one may be more pleasing to one person than the other (honestly though, I think that you'd be hard pressed to be able to tell the difference in a random blind-test). That doesn't mean that one is inherently inferior to the other.</p>

<p>Why? Because you can literally <em>do anything</em> once you've reached the digital realm. The ProPhoto RGB color space represents colors the <em>human eyes can't even see</em> , for Christ's sake. You can stretch the colors of a RAW file to whatever you want, if you were engineering the RAW converter. With proper profiling, you can increase the gamut of a digital capture to a gamut wider than gamut of dyes represented on slide film.</p>

<p>There *is* a maximum saturation and/or hue you can reach of any color on slide film, determined by the max number of dye molecules that can be deposited within a given area of film. The only limits on hue & saturation in the 'digital realm' is that defined by the limits of human vision... or you can go beyond it if you want.</p>

<p>So to knock digital as somehow inferior is ignorant, and I was challenging your broad claim. When you make such bold broad claims, you need to back them with evidence. Else, you'll be challenged.</p>

<p>Now, if you're talking of what is displayable in the digital realm, sure, as of now, a pathetic LCD can't reach the dynamic range of a slide on a backlight. But you think it's gonna stay that way for long? You gotta be kidding me. Look up some of the research going on in HDR displays. They <em>will</em> outdo what slide is capable of now. Why? Because of <em>rational design</em> . To say it won't happen, Sergiy, is countering the wonders of human intelligence.</p>

<p>Pointing out colors as being inaccurate in digital photos vs. the film counterpart is childish. Have you developed the RAW converter yourself using shots of proper color charts? Or used the Adobe DNG converter at least to build a profile for your camera? Had you done that, you could accurately represent the real-world colors, and the yellows of your Komatsu would've been spot on, assuming you had a calibrated monitor. After all, if you're comparing to scanned film, your comparison is even more childish. You're gonna trust the colors of some automated scanning software that inverted the negative, removed the orange mask, then re-expanded the compressed tonal range on the negative?</p>

<p><em>You kidding me?</em></p>

<p>Now, if you're talking of color-gradation, I'm willing to buy that you might see some inherent advantage in some films over a Bayer pattern sensor... I don't see it in the examples you posted, but that could be due to a number of factors... my monitor? The scanned image? Non-color-managed browser?</p>

<p>My point is-- your broad sweeping generalization is entirely invalid until you've provided some theoretical evidence to back your claim. Mathematical equations and all.</p>

<p>Present me that, & I and others will be more willing to give your hypothesis more credit.<br>

-Rishi</p>

Link to comment
Share on other sites

<p>Sergiy:</p>

<p>Check out this link: <a href="../photo/8215134">http://www.photo.net/photo/8215134</a></p>

<p>As well as the rest of Marc Adamus' portfolio... & then explain to me how somehow the colors of digital will always be inferior to that of film...</p>

<p>-Rishi</p>

<p>P.S. I commented earlier that Bayer sensor filter transmission profiles were narrow bands... now I'm not so sure I was correct, and want to state that for the record. Newer filters may have broad transmission bands for red, green, and blue. Scott, or anyone else, have any links to new Bayer sensor filter transmission spectra?</p>

Link to comment
Share on other sites

<p>Re: ccd cooling.</p>

<p>I had not thought about the CCD temp going up with scanning, that is a good point. I see that the Imacon can produce 1.2 GB files at a rate of 50 MB/minute. That is 24 minutes, I can see how the CCD might warm up during that time.</p>

<p>As for mult-sampling vs cooling, I would go with cooling, film scanners are already painfully slow.</p>

<p>You just want to make sure you don't cool below the dew point, assume that the CCD is in open air.</p>

<p> </p>

Link to comment
Share on other sites

<blockquote>

<p>Lex, you see, electrons .... they do not love each other.<br /> And this entire world is built on love.<br /> That is why *electronic* digital cameras have no future. Picture on film is formed, built by a group of particles who love the light and love each other, grouping under the light.<br /> Electrons spread as wide as possible from each other.</p>

</blockquote>

<p>HAHAHAHA!<br /> <br /> <br /> You've clearly demonstrated that you don't even know how your own beloved film works. Perhaps, Sergiy, you should read the thread that you yourself linked to: <a href="00ROOo">http://www.photo.net/film-and-processing-forum/00ROOo<br /> <br /> </a><br>

There, a bunch of us hashed out the theory behind film. It was a bit traumatic, but a great learning experience.</p>

<p>Sensitivity specks on silver halide crystal (collections of Ag+ and Br- ions) defects serve as a sink for electrons that get knocked off of halides upon productive absorption of a photon. As electrons build up at these sensitivity sites, guess what, since electrons repel electrons, less electrons are likely to build up! That can put a 'cap' on the exposure of a grain... but probably only momentarily because a silver ion can accept an electron and become silver metal. So now you have silver metal building up at a sensitivity speck. But they can only grow so large because these clumps start disrupting the crystal lattice structure. So, yet another 'cap' on exposure of a grain.</p>

<p>Furthermore, as layers of film get exposed forming silver metal at sensitivity specks, it's possible that these areas, now opaque, prevent the penetration of photons to an unexposed grain immediately below it (this is a hypothesis I put forward; haven't heard it backed up yet)... which can further 'cap' exposure.</p>

<p>In fact, it is my feeling & understanding that this sort of 'cap' on exposure makes the exposure of film hyperbolic at the higher log E end... which is <strong>one</strong> of the reasons it may be so hard to overexpose film (the other major reason being that the inherent dynamic range is just huge... a property of the system of silver halides itself... much too theoretical a discussion to get into here). How much of that data in the further compressed highlight region is recoverable is up for debate; Mauro & Bernie if you remember Vijay & I argued about this for a while. My opinion is that it is not negligible.</p>

<p>So, Sergiy, what is your point of attacking electronics when it's literally the basis of... like... everything we use?</p>

<p>Groups of particles that love each other? You're showing yourself to be quite ignorant, Sergiy. I certainly don't know everything... in fact, I hardly know anything in this world of infinite knowledge... but, dang, I don't make broad general sweeping comments without any claims to back them up.</p>

<p>-Rishi</p>

Link to comment
Share on other sites

<blockquote>

<p>You just want to make sure you don't cool below the dew point, assume that the CCD is in open air.</p>

</blockquote>

<p>Silicone sealant would have to be used to seal the interface between Peltier (TEC) and back of CCD so that no moisture forms there. Probably would have to put a thermosensor in there too to make sure it doesn't cool too much. Then a fan on top of the Peltier... luckily, there's a lot of open space inside my Minolta... just dunno if I want to be blowing dust around in it. Probably this is one of those situations where Bob Atkins would say 'hobbyists often like to tinker for the sake of tinkering, not for the sake of the final product' :)<br>

-Rishi</p>

Link to comment
Share on other sites

<p>Re: Bayer filters,</p>

<p>The curves I have seen have always been fairly broad, which is really what you would want. See page 1-5<br>

<a href="http://www.graftek.com/pdf/Manuals/basler/A600fmanual.pdf">http://www.graftek.com/pdf/Manuals/basler/A600fmanual.pdf</a><br>

 

<p>As long as the curves * CCD spectral response are linear combinations of the tristimulus curves you can record colors as the human eye sees them. What this means is the curves could be wider then the tristimulus curves and corrected by multiplying a vector made up of the RGB values by a transformation matrix.<br>

In reality this matrix multiplication is done any time a raw file is converted, it is needed to get the colors into a given color space. If the color space is CIE 1931 then RGB values are simply XYZ that you would get from the tristimulus curves, other color spaces have different basis.</p>

</p>

Link to comment
Share on other sites

<p>There's absolutely nothing more natural looking about the image after your 'filmoscope' filter compared to Scott's original image.</p>

<p>I have no idea what you're talking about, and I doubt anyone else does. Anyone else see a difference in Sergiy's crop?</p>

<p>Plus, still no response on your ludicrous assertions about the inferiority of digital, or that somehow the involvement of electrons limits digital's capability but not film's...</p>

<p>-Rishi</p>

Link to comment
Share on other sites

<blockquote>

<p>"Picture on film is formed, built by a group of particles who love the light and love each other, grouping under the light. Electrons spread as wide as possible from each other." --Sergiy Podolyak</p>

</blockquote>

<p>Sergiy, explain what happens in the lowly 1s2 orbital. How is that two electrons manage to co-exist within that orbital. That will suffice for now. We can go on up to more complicated (and electron-filled) atomic and molecular orbitals later.</p>

<p>If you need to get into quantum mechanics to explain, we'll try to be patient.</p>

<p>--Lannie</p>

Link to comment
Share on other sites

<blockquote>

<p ><a href="http://www.photo.net/photodb/user?user_id=423641">Landrum Kelly</a></p>

<p ><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 28, 2009; 07:24 p.m.Sergiy, explain what happens in the lowly 1s2 orbital. How is that two electrons manage to co-exist within that orbital. That will suffice for now. We can go on up to more complicated (and electron-filled) atomic and molecular orbitals later.</p>

 

<p>If you need to get into quantum mechanics to explain, we'll try to be patient.</p>

 

</blockquote>

<p>Far too simple a question, one up one down.</p>

<p>

<p>As rather why electrons in pairs can "love" to be together but electrons by themselves do not.</p>

<p>Or why helium 4 can form a super fluid but helium 3 can not. </p>

</p>

Link to comment
Share on other sites

<p>Well, then, we shall just have to manage to capture paired electrons so that they will do our bidding, Scott. At some point we may even be able to manage quantum effects on crowded sensors. Who knows what the future will bring?</p>

<p>Although I am obviously being facetious, I wonder what impelled Charles Townes to think that he could manage to get light to act in a "coherent" fashion. The guys who make the bridge from theory to practice are even now working on the digital miracles of the future. I suspect.</p>

<p>What will they come up with? No one knows, not even Sergiy.</p>

<p>--Lannie</p>

Link to comment
Share on other sites

<p>Sergiy,</p>

<p>With all due respect, are you a troll? Let's take a look at your 'filmoscope' filter in a meaningless comparison due to the fact that you didn't provide a full 1:1 image...<br>

<img src="http://staff.washington.edu/rjsanyal/Photography/Sergiy.jpg" alt="" /></p>

<p>Wow. Great. <em>Congratulations</em> . Looks like you decreased the highlights, increased the shadows, and added some saturation.</p>

<p>Big whoop-dee-doo. What does this have to do with your argument that digital can't nail colors? In fact, I think you just proved yourself wrong... because you took a digital image, then manipulated, to satisfy yourself. Umm... yeah... that was exactly my point. That you can take a digital file and literally do anything with it.</p>

<p>Hence, again, I find your comments of 'dizitalie's color' ridiculous, inane, and unsubstantiated.</p>

<p>I also can't believe I just wasted 4 minutes of my life composing & writing this post.<br>

-Rishi</p>

Link to comment
Share on other sites

<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="" title="Frequent poster" /> </a> , Mar 28, 2009; 11:11 p.m.<br>

Sergiy,<br>

With all due respect, are you a troll?</p>

 

</blockquote>

<p>Dude, if you have to ask...</p>

<p>Anyway, I've been following this thread with interest, not contributing since I don't have much to say that hasn't already been said, but whats all this hatred for electrons? Yeah, I admit, too many of them surging through your brain has the potential to fry it, but electrons don't kill people, people kill people.</p>

<p>Hahaha, never thought that electrons would be the center of such a political maelstrom. Yeah - lets organize a protest against those evil electrons.</p>

Link to comment
Share on other sites

<blockquote>

<p>Dude, if you have to ask...</p>

</blockquote>

<p>I was just bein' cautious... I've never met a real-life troll before :)</p>

<p>-Rishi</p>

<p>P.S. Good to see you Vijay... er... yeah you're right, must be Stockholm syndrome ;)<br>

P.P.S. I bet Bernie's even happier to see you :)</p>

Link to comment
Share on other sites

<p>

 

<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 28, 2009; 01:44 p.m.</p>

 

<blockquote>

<p>Also, Lex, keep in mind that CCD temperature will rise the more you scan. So if you just turned on your scanner and made that scan in reasonably low temperature weather, I can see why shadow noise is absent. But after 20 minutes of scanning, I'd expect the temperature to rise.</p>

 

</blockquote>

<p>Not necessarily, providing ambient temp is stable and the CCD has reached thermal equilibrium. In the 646 there is no Peltier cooler but the Kodak KLI-8023 sensor package is still thermally bonded to the CCD carrier, a decent chunk of metal in itself and a fine heatsink. Sure, from cold you may get a rise in temp before reaching equilibrium but thereafter I expect you could scan all day without any further change. The vertical form factor of the scanner does mean that heat (from lamp for example) will rise up to the CCD and no doubt this is a good part of the reason for adding the cooler.</p>

<p>The KLI-8023 spec sheet is useful in understanding why Imacon uses such things as Peltier coolers and IR filters:<br>

<a href="http://www.lexharris.net/documents/KLI-8023LongSpec.pdf">http://www.lexharris.net/documents/KLI-8023LongSpec.pdf</a></p>

<p>Cooling is the best way of reducing dark current noise. Some examples of CCD cooling:<br>

CCDs in laboratory optical spectrometers: -45C to -80C<br>

Hubble CCD: -40C<br>

Galileo CCD: -120C<br>

Imacon: +18C<br>

DSLR: none :-(</p>

<p> </p>

 

<blockquote>

<p ><a href="../photodb/user?user_id=940372">Scott Wilson</a> <a href="../member-status-icons"><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/1roll.gif" alt="" /></a>, Mar 28, 2009; 12:50 p.m.</p>

 

 

 

<p>Dark current noise goes up by the sqrt of the integration time.</p>

 

</blockquote>

<p>And signal increases linearly (hopefully) with integration time. So SNR increases with integration time, or conversely decreases with shorter integration time. Thus the degrading effect of dark current noise increases as integration time is decreased. Sorry, I didn't explain myself well before.</p>

 

 

 

<blockquote>

<p ><a href="../photodb/user?user_id=940372">Scott Wilson</a> <a href="../member-status-icons"><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/1roll.gif" alt="" /></a>, Mar 28, 2009; 02:40 p.m.</p>

 

<p>Re: ccd cooling.<br>

I had not thought about the CCD temp going up with scanning, that is a good point. I see that the Imacon can produce 1.2 GB files at a rate of 50 MB/minute. That is 24 minutes, I can see how the CCD might warm up during that time.</p>

 

 

<p>You just want to make sure you don't cool below the dew point, assume that the CCD is in open air.</p>

<p>As for mult-sampling vs cooling, I would go with cooling, film scanners are already painfully slow.</p>

</blockquote>

<p>Scott, see p.21 of the spec sheet linked above. With the higher data rates of Imacon 848 upwards, heating of the CCD device may well be another reason why these higher models have the cooler. Actually, thinking about it, and considering the very low background on the (slower) 646, could it be the ONLY reason?<br>

Condensation / ice formation is clearly a problem at low temps. Lab spectrometers enclose the CCD in a purged (usually Ar) environment to prevent this. Obviously with DSLRs/scanners it is more problematic...</p>

<p>Multi-sampling still has some attraction though - it's the only way of reducing ALL types of temporal noise.</p>

<p> </p>

 

 

 

 

</p>

Link to comment
Share on other sites

<blockquote>

<p>And signal increases linearly (hopefully) with integration time. So SNR increases with integration time, or conversely decreases with shorter integration time. Thus the degrading effect of dark current noise increases as integration time is decreased. Sorry, I didn't explain myself well before.</p>

</blockquote>

<p>But in the case of the scanner the 10ms integration time should be enough to fill the quantum wells of the CCD and yet be a short enough time that not many electrons are going to leak into the cells.</p>

<p>

<p>If I am shooting with my DSLR and have enough light for a 1/100 second exposure I am not worried about dark current at all. If on the other hand I am exposing for 1 minute dark current becomes a much bigger deal. In both cases I have about the same signal level, if I am exposing for 1 minute it is going to be because not much light is reaching the CCD. In a scanner there is enough light that the ratio of electrons from photons to electrons from dark current is very high.</p>

</p>

Link to comment
Share on other sites

<blockquote>

<p ><a href="../photodb/user?user_id=940372">Scott Wilson</a> <a href="../member-status-icons"><img title="Frequent poster" src="http://static.photo.net/v3graphics/member-status-icons/1roll.gif" alt="" /></a>, Mar 29, 2009; 08:49 a.m.</p>

 

 

 

<p>If I am shooting with my DSLR and have enough light for a 1/100 second exposure I am not worried about dark current at all.</p>

 

</blockquote>

<p>Of course, if all you're shooting is a grey card. But what about those deep shadows in a real life scene? Are you not interested in dark current in those shadows?</p>

<p>The point is, the pixels in the highlights at 1/100sec get heaps of photons and SNR is high. The pixels in the shadows get few photons but have the same dark current and the same short integration time - SNR is low. But increase integration time, sure you blow the highlights because exposure is the same across the whole chip, but SNR in the shadows is better is it not?</p>

<p>Put it another way: Your dark scene that needs a one minute exposure will probably look good with a 1 minute exposure, but try upping the ISO and shooting the same scene at 1/100sec. Does it look a tad noisier? </p>

<p>Anyway, back to scanners, my 646 takes about 10 minutes to scan 135 portrait at 6300ppi. The frame is 36mm long so that's about 14.8 lines/sec or 67msec per line. It never occurred to be before, but could this longer integration time (compared to faster scanners) be a significant factor in producing lower noise scans? My Nikon LS5000ED scans the same 135 portrait in 20 seconds but with vastly more noise not just in shadows but in all parts of the scan.</p>

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...