990 995 CoolPix Shoot Out
Brian Olson & Gordon Couger
December 4, 2002
updated May 30, 2005


Red arrows show artifact caused by spun cast lens. It is much more noticeable than in the 995 then the 990 but it is there in the 990.

 This originally was about image processing artifacts but the discussion has developed more about the ring artifact that is present in many digital camera but most often reported in the CoolPix 995 and newer when used on compound microscopes using Kohler illumination.

The artifacts are worst on light backgrounds. If the image has much strong detail the artifacts will not  be nearly as objectionable.

The artifacts can be completely avoided  by not using any zoom at all..

Arron Messing has reported that CoolPix 5000 with the adapter for the slide copier does not exhibit any artifacts. Several others have confirmed that. I have had one person say they get  artifacts with the 5000. GC 5/30/05

The images have not been enhanced in any way or corrected for color they are as shot if you double click on them you will get full res images.

This change in the image from what I see in the monitor is sometime very noticeable when it switches from the saved image to the monitor image. To be fair there are times the picture is improved but many less are improved than are damaged.

Draw Your own conclusions.

     995 CoolPix                                                              990 CoolPix                                                      990 CoolPix
995 950-1 950-2












gregorI would like to thank Gregor Overney for the picture of the optimized ring artifact taken using a relay lens.

 That the CoolPix 995 has a lens artifact is quite commonly known. But so do all digital cameras that have a non-removable zoom-lens. The effect is worse for the 995 than the 990 since the number of lenses in the zoom increased.While interference patterns of the SONY DSC-S70 are different, they do exist as well.

All digital cameras with non-removable zoom lenses might show this kind of problem. It's not specific to CoolPix. I have attached an image I took with a 995. I tried to "optimize" the appearance of the problems. Therefore the problems appear to be overemphasized. -Gregor March 18, 2003




Further work has shown that proper alignment of the condenser and other elements of the microscope and proper Kohler lighting so no stray or off axis light enters the microscope reduces the artifacts to a degree. If the image has much texture or strength it is usually enough to make them unobjectionable.

They can be eliminated by setting the focal length to 8.2mm with f2.6 aperture and focused at infinity. This wastes a good deal of resolution but there seem to be no artifacts that show up at these setting on my scopes. Here is an unenhanced full frame example of some skin cells scraped from my foot and mounted in immersion oil taken with a 40x .65 objective with Mark Simmons's eyepiece adapter on a Nachet 300 in DIC mode. Using the best illumination I could set up. The brown spots are athletes foot fungus. This is the best image I in regards to artifacts that I have been able to get to date. The second picture is a diatom. Neither has any hint of  artifacts that I can find. There are a some spots of dirt on the condenser common to to both images. The Nachet has no diffusion in the Kohler light system and show every fault in the lighting system. That is one of the reasons I use it in these tests. GC June 16, 2003

<>


Conclusions
Gordon Couger 5/30/05


Many reasons have been put forth for the cause of the artifacts in Nikon  CoolPix cameras. After many many hours of testing  none of them explain the problems I see to my satisfaction. Considering that projected image from the microscope is in focus from any point  from the surface of the eyepiece to infinity any disturbance in the rays of the image as it passes through the camera lens will show up in the image. The fact that it only shows up in transited light microscopes  causes me some real concern about making any generalized statements as to the cause of the artifacts. Form the many hundred images I have taken and been sent by others I have see reflections, refraction, hot spots as well as the blobs, rings and central artifact..

Now at the end of  May in 2005 after 2.5 years  No one has come out with a really good consumer camera for a microscope. The small lens sized of the CoolPix 990, 995, 4500, 5000 and 5500 make an easy match to a microscope. In many cases the eyepiece of the microscope works better then the most expensive modern relay lens. Ted Clark gets almost a full field  with a Olympus 8X WHK eyepiece and a high eye point 10x Zeiss Kpl 20 mm FN and CoolPix 995. Both eyepieces are part of the correction in the optical train and using modern digital adapters will show chromic aberration  in outer third or forth of  image.
.
It is well known that the CoolPix 990 has minimal ring artifact that have worked well for every one that uses it is 3.34 mega pixel camera and  sells used for $100 to $200 dollars. It will produce images good enough for publication in journals.

Arron Messing has been using the CoolPix 5000 with no artifacts problems. It  will use the same hardware as a the CoolPix 990, 995,  & 4500. The CoolPix 5000 sells for $250 to $300 used. If I had to buy something today this is what I would get.

Hopefully a better camera will emerge that will adapt to a microscope but right now the CoolPix 990 and 500 are probably the 2 best choices for  someone that doesn't want to go the digital SLR route.  The well known and  stable feature set, the ease they connect to a microscope and their history of reliability and longevity are hard to beat.


Digital SLR


For the money buying a camera for a microscope I think a dedicated camera for a microscope  makes more sense that dSLR. I would only put a dSLR on a microscope if I was buying one for some other reason. Just as SLR camera are not good fits for microscope neither are dSLR cameras. When I want high resolution images I use a  CoolPix 995 to compose the shot and determine the exposure and then use a 35 mm  microscope camera.

Some people are havening good results with digital SLR camera on microscope. Jan Hinsch  wrote About the use of Digital Single Lens Reflex Camera on Microscopes for the McCrone Group's Modern Microscopy. If you decide to go this route be sure the camera has a mirror lock up, a way to view the image with the mirror locked up and meters the light with out a lens. Some Canon users are very happy with the low light performance of the CMOS sensor Canon uses.  Be aware the Canon Rebel does not have mirror lock up feature.
G. Couger 5/30/05<>
 
Update 10/12/2005

The adapter makes no difference. I have not determining the cause of the rings to a point that I can say with any certainty anything but they get worse as you increase the zoom and that they are inherit to the Coolpix 995  through 4500 and the lighting  not anything added in between. I suspect that the rings and blobs are artifacts caused by internal refection of tooling marks or something. It is contestant over a long period of time  from 995 thou 4500.  The Coolpix 5000 and 5400 do not have the artifacts I documented. I believe that everything I know about a 5000 applies to a 5400 everthing I read leads to believe that. I would check and be sure that you can't crash the lens into the eyepiece with the zoom of the movable lens in case it is differs the than the 5000.

All cameras with lens in the path between the eyepiece and the sensor or film have the potential to distort the image to some degree.

I have built a number of parts to mount my Coolpix 995 to a Zeiss dovetail that goes in the camera pot  and I almost always use Mark Simmons' adapter with my on Leitz and Zeiss objectives. A   Nikon Art.Nr UR-E6 is need with a CoolPix 5000 & 5400 that cost less than 10 dollars  most solutions such as Mark's adapter  for CoolPix camera used on microscopes at Perspective Image www.perspectiveimage.com/1/nikon.php


It has been report that artifacts are absent in images made by late 4500's. I have not seen any pictures taken by late run 4500's. But certain conditions must be met to make the rings show up so the people reporting them may or my not have thing right to produce them. The better the lighting on transmitted light microscope is the better they show as the depth of focus increases to include the light path above the first lens in the eyepiece near the subject and below the first   lens near the subject in the condenser.  When this condition is reached every speck of dust show up in the image. This not the same as depth of field which for a microscope is very small and is inversely proportional to the size of the numeric aperture of the entire microscope.

To get the best solution for a CoolPix  has an artifact problem as the ones shown above is to use an adapter/relay lens/eyepiece that fills the field with the largest image possible. In my expertise there are no artifacts the widest angle setting of the zoom lens on a CoolPix 995 or  4500. For my testing I use a Nachet 300 DIC transmitted light scope that I believe to be one of the best scope possible to show lens artifacts or any problem in the light train. It has true Koehler lighting with no diffusion in the light train and the DIC color gradient with out a green filter really high lights any problem.

The second choice is digitally removing the artifacts. It also has the effect of correcting for uneven lighting, a dirty condenser or anything else on the slide used to subtract out the artifacts and the back ground. I believe that this can be automated in  ImageJ a public domain Java program from the National Health Service that has a script language and program plug in interface.  There is an active programming and users group that will help you with problems. A much better solution than in most paint programs.

To the best of my memory to digitally remove the artifacts shoot an image of a blank slide with the mountant and cover slip a near the subject as you can get, save the as subject-bl.tif. The to what ever you have to do to it make a smooth even background and save it as subject-bk.tif.  It may just take heavy low pass filters or it may take hand editing. Then take the picture of the subject and save it as subject-m.tif an write protect it and make back ups of all 3 files in at least 2 other places at lest one off premises. I upload mine to my server and when I was working took a back up ever week to an off site fire poof safe.

Then copy subject-m.tif to subject-mw.tif use the ImagJ image math tool differencing to remove subject-bl.tif from subject-mw.tif and save subject-dif.tif and save.  You may have to go to Paint Shop Pro or Photoshop at that this point to make background with subject-bl.tif and over lay subject-mw.tif

I have only worked this out on monochrome images. So you may  have to split the stack into R, G & N or C, M & Y stack and process each one separately but once a scrip is written for it is will only take a time to run. But I am on another project at the moment but I would be gad to help anyone with the problem.

Today I  have a Coolpix 995, several Zeiss 35 mm cameras and Leitz 4X5 for a microscope. For high quality I would use the 995 as a light meter and use film. If I was in a hurry I would use 4X5 and contact print the negative or scan them directly in with a scanner or possibly try using the 995 to scan the negative on a light table and try to get the best of both worlds using black and white film and a set of red, green and blue filters I have.

If you do this on images for publication I would make copies of every step and the complete environment that did the work including a full copy of installation of Imagej and all its plugins and the copy of Java runtime in use and zip it up and put in escrow with a third party with an unimpeachable reputation and make the whole package available to the readers as well via a web site so anyone with a question about your methods could see for them selves what you did it. This is a good piece on the ethics of image processing  By the
Pharmacy Department at Arizona  University  Digital Image Ethics The example I  just described is a very extreme one case of image processing and at the least the original should be the origninal should be easily availble to the reader.

If I were buying a digital camera just for use on a microscope today. I would choose from one of  these three. A Coolpix 990 with the 3.3 mega pixels and a zoom that only cause the faintest hint of rings that are extremely difficult to see and cost from $120 to $200 on the used market.  A 990 will capture all there is to see in a microscope and the articulated  head makes a fantastic field microscope an often over looked feature of the Coolpix 600/950 thru 4500 series.

Aaron Messing continual  praises of  the 5 mega pixel  Coolpix 5000. Aaron has the credentials to have a opinion that counts on microscope cameras an "aaron messing" "nikon small world"  search on Google brings up 61 hits.  The 5000 cost $200 to $300 used on ebay and needs Nikon UR-E6 step down adapter to make fit 28 mm Coolpix accessories.  I wouln't buy Nikon's high priced relay lens. I use sometingh tht uses the eye piece that matchs the microsope objects or Mark Simmons adapter perspectiveimage He desiged, makes and tests it himself. He runs a small low over head business and when you talk to some there it is Mark and he does not make extravagant claims about his product. He is straight forward about the problems with chromatic aberration of older Zeiss, Lomo, Leitz and Olympus microscopes. I can show you images that look bad but in reality it is small problem in most cases as the aberrations are most noticeable in the outer 1/3 of the image and the feature must be shown in both red and blue light not red or blue light as most features are. If you are using the microscope for measurement and using colored light this would be a serious  concern but most images are cropped to a square image from a round image made by the eyepiece so only the 4 corners present a problem.

If chromatic aberration is a problem then the only choice is to use an eyepieces that matches the objectives. I have built a number of parts to mount my Coolpix 995 to a Zeiss dovetail that goes in the camera pot  and I almost always use Mark's on Leitz and Zeiss objectives.

Spike Walker needs no intodtion in photomicroscopy. Spike's work says it all. Spike has gone from the Coolpix 5000 to a Sony W7 for a small digital camera. Spike has a small fortune in camera equipment. I am not sure if he became a professional photographer to make money or to try to pay for his obsession. Ether way the we are better off for both Spike and Aaron's experience and their stunning images. Even though at times they had to beat me about the head a bit to get me to listen.  At 7.1 mega pixels the W7 the $400 camera and $225 for an adapter is much less than I paid for the Coolpix 995 to get in the game. In private correspondence Spike has convenience to take a very hard look a the W7 before I buy another camera. If I was starting form scratch I am fairly sure it would get the nod over anything but a good deal on Coolpix 990 if I was trying to get by on the cheap.

All of this bring up how many pixels is enough.  In the McCrone Group's Modern Microscopy Ted Clarke's shows that for capturing data 2 mega pixels will do www.modernmicroscopy.com/main.asp?article=31  and Nikon U puts it 2.2 Mega pixels for a 0.5x image and 617 pixels for 100x 1.4 n.a. objective at http://www.microscopyu.com/tutorials/flash/pixelcalc/ with a bit more user friendly interface.

So for a microscope what good are the extra pixels. They determine the size the image can be printed. While 617 pixels gets al the data on 1400x image it makes lousy 4x5 print.

Here is a scanning calculate http://www.scantips.com/calc.html to find the various resolution to scan film for the resolution you need. Magazines use 300 dots per inch dpi, news papers use 60 to 75 dpi and photographic prints use 240 to 400 dpi and  I believe JPGs and printer's magic software relaxes that to some degree. But you don't get to cheat at the print shot that uses halftones and ink on paper. I have been there, done that and it took 5 years to get the ink out from under my finger nails.

Here is a simple table   on the Lulu Blog showing 8 mega pixels needed for a 300 dpi 8x10 page. That's why the commercial buyer want slides or very large digital images as a 36 x 24 inch photo quality poster takes at lest a 223 megabyte image.

I think this shows if one is serious about selling ones work the digitally age is not truly here yet. Using a 3 to 7 mega pixels camera to do  small images and use it to set up shots for film caners is going to be the way to work. Unfortunately the market for microscope images is small.

Now the down side if I have my math right is an operation that takes 10 seconds on 3.2 megabyte image 223 megabyte image takes 180 days on 223 megabyte image not counting swapping out to disk. The same 10 second operation on a 3.1 mega pixel image takes 3 minutes and 15 seconds on a 7.1 mega pixel image. So the bigger the image the faster the computer, video card and monitor you need. While Moore's law seems to still hold that the speed of computers are doubling and the size of chips grows smaller when the number of operations is the time it took at the old sizes raised to the power of the new size divided by the old size large numbers being anything to  molasses like slow crawl. <10/12/2004>