CCD flare with Nikon Coolscan 5000ED

lex_harris · August 23, 2008

The Nikon crop with 50% shadow enhance:<div>

</div>

lex_harris · August 23, 2008

And finally the Imacon crop with 50% shadow enhance:<div>

</div>

rob_johnston4 · January 9, 2009

sorry off topic: but couldn't find a better thread.

Well... been trying to research for a while whether to buy the 5000ed of there is something better within fiscal reason... this blog really tried to put me off the Nikon big-time. Looked at the Imacon, but £4000 - £6000 is just out of the question.

Have about 500 slides to scan all velvia and pro-v1 from '92-'98 all snowboarding and Snow/Mountains... and am about 6 days away from buying one on fleaBay. if anyone has a better suggestion, let me know.. if Not, I'll post how I get on.

lex_harris · January 9, 2009

Rob, to put it in perspective, many 5000ED users seem to be quite happy with the performance of their scanner. The results I get with my particular scanner leave a lot to be desired as you can see from the examples, however that is just one scanner. As mentioned earlier in this thread I did have some of my problem slides scanned by a local Pro Lab on their 5000ED and got much the same flare results. So that's two examples where I have personally witnessed this problem. To muddy the waters even further, this Pro Lab claimed they could not see any flare problem in the scans so it seems people have different ideas about what is acceptable.

On my scanner the problem is most pronounced in Kodachromes, and less noticeable in E-6 films like Velvia and Provia. But even with E-6, it is still there. Aside from obvious artefacts at light/dark transitions, it also manifests itself as general loss of contrast throughout. With my Nikon scans I find myself struggling with a general "muddy" appearance, while Imacon scans seem to leap off the screen in comparison. Part of it may be the respective software - Flexcolor has better control of colour than Nikonscan. Other users report better results with Vuescan or Silverfast vs Nikonscan.

Anyway, the upshot is I probably have a "Friday afternoon" 5000ED that is at the bottom of the barrel for performance, you may be lucky and get a comparatively good one.

rishij · January 9, 2009

Every LS5000 I've tested has, to me, unacceptable flare (I tested 9 of them in our lab here at the University of Washington).

The LS9000 performed much better in this regard in a scan of the same slide.

The Imacon 848 performed even better than the LS9000. But the Imacon, I believe, doesn't have glass over the CCD, which leads to other problems. Such as dust on the CCD which then shows up as noise in shadows. Pretty unacceptable to me so I'm inclined to work with the LS9000.

Haven't tested my LS4000 yet, but I bet it's the same as the LS5000.

Rishi

rishij · January 9, 2009

Maybe a naive question, but can you multicoat glass yourself? At least on one side? And isn't the side accessible to us (the side facing the incoming light, away from the CCD) the most important side to coat, so that all the light is transmitted into the CCD housing?

Reference on lens coating:

<a href="http://rick_oleson.tripod.com/index-166.html">http://rick_oleson.tripod.com/index-166.html</a>

Rishi

lex_harris · January 9, 2009

"But the Imacon, I believe, doesn't have glass over the CCD, which leads to other problems. "

Rishi, I've heard this said, but when I last had my 646 open it seemed the CCD did have a cover glass. In fact the service procedure for 646 and 848 scanners recommends cleaning the sensor surface with brush and/or isopropanol and/or lint-free tissue in case of dust accumulation. This would surely not be recommended if there was no cover glass, a naked CCD would be destroyed with such treatment.

Probably not widely known is that Flexcolor includes a hidden diagnostics feature for testing the response of the CCD. This feature can be used to make sure light is reaching the CCD across its full length and to look for signs of dust on the lens, CCD or light source. Send me a PM if you want to know how to do it.

lex_harris · January 9, 2009

"And isn't the side accessible to us (the side facing the incoming light, away from the CCD) the most important side to coat, so that all the light is transmitted into the CCD housing?"

Based on the spatial orientation of the flares (I prefer to think of them as ghost images) which is dependent on location within the image, my conclusion is that these artefacts arise either from multiple reflections occurring between CCD surface and inside of the protective glass, or between the two surfaces of the glass, or both. Take another look at the "pinhole test" and you will see what I mean.

So if this theory is correct you would have have to coat both sides of the glass.

rishij · January 10, 2009

That hidden diagnostic feature -- are you sure it works on the Imacon 848?

Basically, here's the problem I'm seeing with the Imacon 848:

<a href="00RJ35">http://www.photo.net/digital-darkroom-forum/00RJ35</a>

Look further down, as I'm more worried about the subtle lines in shadows than the complete jacking up of the image which, at least, is fixed by rescanning.

I'll PM you also for your instructions.

Thanks!

Rishi

janrsmit · January 10, 2009

Rob.

The problem with CCD sensor scanners is the glass window that protects the ctual sensor. It sits on top of the sensor housing as an earlier post in this thread shows. This glass causes this ghosting, and it reduces the contrast on the whole. When removed of course the ghosting is gone, but ( to my surpise) it als improved the cntract (micro-contrast), ie the pictures pop out of the screen. (While taken apart i also modified the "dark chamber" between the lens and the sensor by covering its walls with very strong light absorbing material(to remove any chance of reflecting light reappearing at the sensor).

Off all ccd scanners i am aware of the nikons and the imacons are the best. Lex has an 848 that appears not to show ghosting. I had done tests with a much older model imacon and it definitvely had ghosting.

What i suggest is : either get an ls5000 and i help you out in modifications, or if it is just 500 slides 35mm contact me to get them scanned (mentioing UK pounds suggest you are in the UK, i am in Holland). Scanning is one thing gettting the color correct can be another trying thing. At least i have sorted both the ghosting and the color out.

Jan R. Smit

IQ is technology, PQ is YOU.

janrsmit · January 10, 2009

Rishi,

The cover glass on the sensor can be replaced by other glass with varying cotings. In germany there is a company that does just that:

<a href="http://www.imageintensifier.com/neu-english/optoelectronic/services-glas-covers-removal.html">http://www.imageintensifier.com/neu-english/optoelectronic/services-glas-covers-removal.html</a>

Jan R. Smit

IQ is technology, PQ is YOU

lex_harris · January 10, 2009

Jan, are you still working with no cover glass at all or did you have a new coated glass put on the sensor?

Also, do you know where one can get a spare CCD sensor? I would be interested in sending one to Germany for modification but not the only one I have (in case it fails).

Was there any adjustment required in your scanner (e.g. electronic adjustment) after upgrading the sensor?

janrsmit · January 10, 2009

Lex,

Still working with no cover glass. The "dark chamber" is really closed, so basicallyno dust from outside can enter the dark chamber therefore notget on the sensor.

You have to deal with the local Nikon repair service on a spare, but it will be quite difficult. For the LS50 en LS 5000 the optical block (mirror-lens-sensor assembly) is replaced as one whole, and the old one has to be returned to Japan. The price tag to the optical block as about EU220. That is ithout any service labor cost.

I did not do any adjustment (electronic), but basically because i noticed no adverse results and the nikon service manager talked me out of it, because of a possible risk that the automatic tool they use for checking and calibrating could go wrong because of the mod's, and did not know how to restore to the previous good state.

IQ is technology, PQ is YOU

Jan R. Smit

rob_johnston4 · January 10, 2009

thanks guys.

Jan: Funny, I have a Ls50, and did some decent scans back in 2001, but unfortunately over the years lost the drivers... after much search I found some on the internet.. but could never get to work on the Win2000 or XP, Vista Operating systems I have. Hence my current search for a newer, much better? scanner.

With regards to your offer: I have at least as dozen badly under-exposed shots I want to try and create usable images out of. I bought a copy of CS3 and figure between multiple tries re-scanning with different settings and then CS3 'air-brushing' , I should be able to bring out some of the detail in shadows, and then hopefully get some good 'poster' size prints made out of a few (deep red-orange sky sunsets, etc) . So figure I'll just start playing with it myself.

I was going for the 5000ed as the research I could find talked about it being able to pick out a lot more detail in the darks than most scanners of camparable cost

There are 2 on ebay right now and i figure I'll be able to pick one up for no more than £700. Unfortunately not with the extra adaptors I'm looking for (negatives, cartridge film etc) Yes, I am in England. Reading area.

Now back to the subject of this Thread! - after a first scan of the images, figure I'll take the risk and have the company mentioned in Germany try work their magic on the CCD sensor, if needed. anybody know a rough estimate on the cost?

rishij · January 10, 2009

Jan, are you saying that any time you take apart the optical housing, it may need re-calibration?

I've taken apart my LS-4000, but not the optical housing... that seems a bit complicated to take apart. I only proceeded as far as illustrated here:

<a href="http://www.pearsonimaging.com/articles/howto/ls5000cleaning.html">http://www.pearsonimaging.com/articles/howto/ls5000cleaning.html</a>

In other words, I took it apart to this level:

I did this to clean the mirror, and also I noticed that in the LS-4000, there's an area above the mirror housing that's partially 'open', i.e. leaking light, that serves no purpose. So I placed putty all around it. Now cleaning the mirror. Let's see if the scans get better.

Rishi

janrsmit · January 10, 2009

Rishi,

Contrary to the peason approach it is not needed to take out the mirror for cleaning. As long as you have a clear access to the mirror. Be aware the mirror is very easily scratched, the mirror surface is quite soft.

The top cover of the optical box is a thin plastic sheet, its front part is visible above the lens.

Opening the optical box is no reason for recalibrating. Removal of the glass cover had more (positive) impact than just the ghosting. I had an extensive demonstration of the calibration process by a Nikon Service Center technician, they use a software tool for it that performs the steps with little manual interaction, including correcting for the infrared leakage. The removal of the glass cover could amongst others have altered the infra-red inpact on the visible red (fortunately i did not notice this, i even assessed it with someone knowledgeable on scanning and profiling of scanners, only to come to the conclusiion that recalibrating was not needed. Mind you if you disassemble the lightsource-assembly, a calibration is needed to re-alining this as misalingments can have a serious inmpact on the scan result.

Removing the ccd sensor assembly and then assemble it again can easily lead to mis-alignment, minute diferences are quite visible. This is something the nikon service center cannot resolve (at least for the Coolscan V (LS50) and the LS5000), this is done in the manufacturing location in japan. So removing it, requires to carefully mark with a fine x-acto knive or similar, and using a loupe on two or three sides such that when assembling these marks help you in getting it correctly aligned again.

Jan R. Smit

IQ is Technology, PQ is YOU.

janrsmit · January 10, 2009

Rob,

In a telecon with Eureca, they mentioned a ball park figure of around 200 - 250 euros, the risk that the sensor gets contaminated or worse damaged beyond repair is for you, they stated the change of 1 to 10.

Yes it appears that the LS5000 is better in the dark reas compared to the V(or LS50). I tried two LS5000, only to find that only if you scan multiple times (4, 8, or 16 times) it would reduce the scanner noise in dark areas and show more detail in the image result. But after removing the cove glass on the LS50 that difference is minimised. So yes modifying an LS500 should improve also the resolution in dark areas of the slide being scanned.

I use Silverfast software as scanning tool, and their Multi Exposure does a good job too of retrieving the most out of dense or dark areas in slides.

However a word of caution, the scanner profile you use could lead to "clipped" dark areas. This is a result of the profiling target being used (and the software), these target (IT-8) was not ment to go all the way to the darkest tones in slides(nor all the way in the highlights), they stop at D=2.4 or so, while i foun that my slides coud go in excess of D=3.2 . So i experimented a lot, even with different profiling software and only after the removal of the coverglass and modificaion of the scanned IT-8 image, i was able to get a correct color profile without blocking the dark areas. I use a profile inspection tool as well (Gamutvision) to visualise problems with profiles, which can be quite profound.

The nikon scan software (i used the latest version) has profiles, but these are canned and you cannot add your own. So when still using this software i switched of color management to get the max in dark areas (and highlights), but at the expense of colors being not quite close to the real thing.

IQ is Technology, PQ is YOU.

Jan R. Smit

rishij · January 10, 2009

Jan,

Thanks. I'm trying to understand what you call leaks in the 'dark chamber'. Though I haven't taken the scanner completely apart, I can see one source of light leakage, below:

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/FixingLightLeakageWithPutty-1.jpg">Link to Full-Size Image</a>

That is, I think, the thin sheet you are talking about. It doesn't lay flat, so light can leak under it and into the lens. So I covered it up with putty:

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/FixingLightLeakageWithPutty-2.jpg">Link to Full-Size Image</a>

Are there other source of light leakage also? Everything else seems pretty well sealed to me. But, as I said, I haven't taken the optical housing apart.

Hearing your description though, it seems a bit dangerous. I'm kinda close to giving up on the Nikon scanners. I just tested the horizontal resolution (along the CCD axis) and it came out to a dismal 10 megapixel equivalent from a 35mm frame of Velvia. The resolution in the other axis is significantly higher (15 megapixels-ish). The Imacon 848 pulled out a whopping 25 megapixels worth of data (1000:1 contrast test chart) from the 35mm Velvia... which exceeds even the stated resolution of Velvia film (160 lines/mm at 1000:1 contrast, which, for a 35mm frame, is the theoretical equivalent of 22 megapixels).

I have access to an Imacon 848, but it has its own host of problems (see here:<a href="00RJ35"> http://www.photo.net/digital-darkroom-forum/00RJ35</a> )

Cheers,

Rishi

rishij · January 10, 2009

However a word of caution, the scanner profile you use could lead to "clipped" dark areas. This is a result of the profiling target being used (and the software), these target (IT-8) was not ment to go all the way to the darkest tones in slides(nor all the way in the highlights), they stop at D=2.4 or so, while i foun that my slides coud go in excess of D=3.2 . So i experimented a lot, even with different profiling software and only after the removal of the coverglass and modificaion of the scanned IT-8 image, i was able to get a correct color profile without blocking the dark areas. I use a profile inspection tool as well (Gamutvision) to visualise problems with profiles, which can be quite profound.

</blockquote>

Wow. Jan, I'm so glad someone else discovered this problem also, though Don Hutch also mentions it in one of his manuals.

Yes, profiles made from HCT or IT8 targets often lead to 'clipping' in dark areas of scans. It's quite unacceptable. Some profiling software tries to compensate this using their own proprietary techniques... by best guess at what they do is: they probably darken the scan of the target itself before making the profile. Of course, I don't know for sure... I mean, this is such an esoteric field and basICColor software itself costs a ridiculous $500 and I haven't found a way to even get in touch with the software engineers to talk about what they do with this clipping problem. If they'd even be willing to communicate that with me.

I've had much more luck with the software devs for LPROF. They're very nice folk and we chatted back and forth about possible ways to avoid this dark clipping. Darkening the target scan seemed to solve the problem... however, I have not yet found a way to apply 'black clipping' in 16-bit (48 bit color) space... Photoshop only allows you to apply black clipping on an 8-bit scale... do you have any idea of how I could black clip the target scan on a scale of 1-65,536? I know Cinepaint does it, but for the life of me I can't get it to compile correctly on my Mac to safe the final TIFF file. Argh!

I've confirmed that unexposed Velvia scans as significantly darker than Dmax on a HCT or IT8 target.

Perhaps this is also due to optical flare (which'd be much greater on the IT8 target than an unexposed portion of film!), and perhaps this is why profiles made for the Imacon 848 show much less dark clipping than those made for the LS-9000 (which shows more optical flare).

Don Hutch suggests some inane way to get rid of the flare by scanning the target in 4 different orientations... this is a dumb idea and doesn't work (just think about it... I myself also proved it to myself by trying... and failing). What DOES help though, is to take every individual dark patch in a scan, New Layer via Cut --> rotate 90 degrees --> set Blend Mode to 'Darken'... repeat for 90 degrees, 180 degrees, and 270 degrees.

Of course, no sane person in the world would have the patience to do that for every dark patch by hand. Perhaps one could code some software to do it... I would if I had the time.

Rishi

rob_johnston4 · January 10, 2009

thanks guys.

Jan: Funny, I have a Ls50, and did some decent scans back in 2001, but unfortunately over the years lost the drivers... after much search I found some on the internet.. but could never get to work on the Win2000 or XP, Vista Operating systems I have. Hence my current search for a newer, much better? scanner.

With regards to your offer: I have at least as dozen badly under-exposed shots I want to try and create usable images out of. I bought a copy of CS3 and figure between multiple tries re-scanning with different settings and then CS3 'air-brushing' , I should be able to bring out some of the detail in shadows, and then hopefully get some good 'poster' size prints made out of a few (deep red-orange sky sunsets, etc) . So figure I'll just start playing with it myself.

I was going for the 5000ed as the research I could find talked about it being able to pick out a lot more detail in the darks than most scanners of camparable cost

There are 2 on ebay right now and i figure I'll be able to pick one up for no more than £700. Unfortunately not with the extra adaptors I'm looking for (negatives, cartridge film etc) Yes, I am in England. Reading area.

Now back to the subject of this Thread! - after a first scan of the images, figure I'll take the risk and have the company mentioned in Germany try work their magic on the CCD sensor, if needed. anybody know a rough estimate on the cost?

janrsmit · January 11, 2009

Rishi,

The mirror housing is the visible part of the optical box, you have put putty between the optical box and the frame. Better to remove it, as the optical box moves in the frame. It moves an innerframe holding the optical box and the lightsource to scan a slide, it does not move the slide. The putty will only cause problems.

Wrt to the resolution, how did you measure this? Would like to be able to reproduce it on my scanner to se the results.

Yes i also tried Don Hutchinsons advices, but with little success. What i did was do a straightforward scan of an it-8 target and in PWP (Picture Window Pro) stretched the dark end to maximum darkness (0,0,0) making sure it was not clipping either channel(RGB). Then the best reuslts are with PM (Profile Mechanic) of the same supplier as PWP (DL-C.com). Tried whatever profiling tool i could lay hand on(demo, evaluation, basiccolor, some open source stuff, oops i forgot some) even the profiling function of Silverfast, the tool i use for scanning. PM works best for me. I shoot mostly with Fuji, Don's targets arequite expensive and there is no Fuji version. Also i figured out that at best with these targets you can reduce the brand typcial color anomalities, which i close enough for me. All i want is to get the maximum out of the image into a digital master.

IQ is Technoloy, PQ is YOU

Jan R. Smit

janrsmit · January 11, 2009

Rishi,

The mirror housing is the visible part of the optical box, you have put putty between the optical box and the frame. Better to remove it, as the optical box moves in the frame. It moves an innerframe holding the optical box and the lightsource to scan a slide, it does not move the slide. The putty will only cause problems.

Wrt to the resolution, how did you measure this? Would like to be able to reproduce it on my scanner to se the results.

Yes i also tried Don Hutchinsons advices, but with little success. What i did was do a straightforward scan of an it-8 target and in PWP (Picture Window Pro) stretched the dark end to maximum darkness (0,0,0) making sure it was not clipping either channel(RGB). Then the best reuslts are with PM (Profile Mechanic) of the same supplier as PWP (DL-C.com). Tried whatever profiling tool i could lay hand on(demo, evaluation, basiccolor, some open source stuff, oops i forgot some) even the profiling function of Silverfast, the tool i use for scanning. PM works best for me. I shoot mostly with Fuji, Don's targets arequite expensive and there is no Fuji version. Also i figured out that at best with these targets you can reduce the brand typcial color anomalities, which i close enough for me. All i want is to get the maximum out of the image into a digital master.

IQ is Technoloy, PQ is YOU

Jan R. Smit

rishij · January 13, 2009

Ok Jan, thanks for the tip.

I removed the putty, and put in double sided tape instead. So, now, the setup looks like this (and seems to effectively, or potentially, block stray non-image-forming light from entering the lens):

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/LS-4000_TapingPotentialLightLeakShut_1.jpg">Link to Full-Size Image</a>

And here's another view of it:

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/LS-4000_TapingPotentialLightLeakShut_2.jpg">Link to Full-Size Image</a>

I put the scanner back together & it ran just fine, as that little flimsy piece of whatever (plastic?) moves along with the optical housing.

I hardly see any flare at high contrast boundaries. These same areas show quite a lot of flare on 4 of the 6 LS-5000s at the lab. Not much flare on the LS-9000 either. But all still show a little more flare than on the Imacon 848 in the direction of the CCD axis. The Imacon, for some crazy reason unbeknownst to me, however, shows flare in the axis perpendicular to the CCD (temporal flare?)... as can be seen below:

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/ImaconFlare.jpg">Link to Full-Size Image</a>

It is annoying though how often I have to clean the mirror. It's like a regular thing. Yet some people apparently pay people $250 to do it? LOL. It's really an easy job. Though incredibly annoying.

Jan, do you think I could get inside the optic housing and apply a coating to the glass myself?

Cheers,

Rishi

lex_harris · January 13, 2009

"The Imacon, for some crazy reason unbeknownst to me, however, shows flare in the axis perpendicular to the CCD (temporal flare?)"

Rishi, I don't think it's a temporal issue. This looks like a dispersion effect, possibly due to dusty lens. Light from bright parts of the image is reaching the sensor when it should not be. I'd suggest you look at the IR filter that sits on top of the lens, it may need cleaning.

rishij · January 14, 2009

Yeah, Lex, I think you're right. Basically, as the scanner approaches a dark area next to a light area, light from the light area leakes onto the CCD while it's still scanning the dark area. Probably a really dirty lens. Problem is, I'm not allowed to take it apart... *sigh*. Even though I've fixed much more complex equipment.

Here's an even worse example of the flare:

<a href="http://staff.washington.edu/rjsanyal/ScanningWoes/ImaconBanding/Imacon848_Flare&Ghosting.jpg">Link to Full-Size Image</a>

I must say, I absolutely hate the Imacon.

I'm *really* hoping the Minolta Dimage Scane Elite 5400 turns out to be some sort of miracle. Haha. Oh brother.

Rishi

Sign In

CCD flare with Nikon Coolscan 5000ED

Recommended Posts

Link to comment

Share on other sites

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

Posted Images

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Create an account or sign in to comment

Create an account

Sign in