Complete failure and 2 decent pictures

[scitch]

I'd like some advice on two of my most recent stacked photos. One I was really looking forward to it coming out well and it came out horrible. I really enjoy looking at compound eyes under the microscope, but this close-up came out terrible.

It was taken with a Sony A200 at about 3 second exposure ISO 200 at 100X. It is a stack of about 50 images. The eye kept moving to the left every time I adjusted the focus and I believe that was the source of the rainbows at the top. But I'm still not happy with the rest of it. I think the problem is lighting. It was lit with the halogen light from my stereo microscope with a semi-transparent blue diffuser at about a 60 degree angle. In this case, it was not Zerene's fault, the images being stacked were not good.



This image of the fly whose eye is shown above is a little better, but not perfect. It was taken on the same scope with the same camera but at lower magnification (40X). It is a stack of 56 individual photos. On this one, Zerene caused some problems. Many of the individual images are much better than the final. The hairs look terrible compared to some of the source photos. Also, there was a piece of carpet fiber or something coming from its knee to the base of its wing. You can see the shadow on its eye. This fiber looks perfect in the source photos, but is nearly invisible in the stack. I don't care in this case that it's gone, but I was photographing some mold the other day and there were some great looking fibers that were supposed to be in the picture that disappeared.




The next image is a hornet's eye. This one was done on my stereo microscope illuminated from above. I don't know the exact magnification because the magnification through the camera is different than the magnification through the scope with this setup. I'm fairly happy with the potential of this one, but should have taken more photos. It is a stack of 20 images. This scope doesn't have a fine focus knob, so it's harder to take small steps. I didn't go nearly deep enough and boy do I need to get my sensor professionally cleaned. I just keep making it worse. Auto levels in Photoshop helped get the true colors to show.



Thanks in advance for your advice.

[rjlittlefield]

scitch, hello! As introduction, I am Rik Littlefield, the fellow who wrote Zerene Stacker.

On image #1 yes, the streaks are due to movement across the frame. ZS uses the strategy of aligning all images against whichever one appears first in the list after the stack is run. (The list will be in either its original order or the reverse of that, depending on a preprocessing decision.) When an image is shifted to align with its predecessors, missing pixels are filled in as replicates of the closest edge. This produces streaks in that one image; such streaks are liable to propagate into the output. One way to think of the streaks is that they identify portions of the frame that are not covered by all source images and are thus uncertain. Ideally missing regions would be ignored and not filled in by replication, but that's not the way the current version works.

The second image appears to have a couple of problems. First is that the mottled appearance around the front of the face suggests that the image was constructed using the DMap method, not PMax. DMap is not a good idea with this sort of subject. Sometimes it can be made to work well by careful choice of parameters, but that is often difficult and sometimes impossible. One of the features of DMap is that it ruthlessly homes in on the one single depth at which the strongest detail can be seen. When there is a strongly detailed layer such as this fly's cuticle, with a less strongly detailed structure in front of it, the foreground structure tends to disappear entirely. I suspect this is the cause of your disappearing fiber.

I can't tell what you're referring to when you say that "The hairs look terrible...". If you're referring to a broken-up appearance where hairs sometimes disappear entirely, then this may due to the DMap effect described above where foreground structures can completely lose out to background structures that have higher contrast. Using the PMax method, it is more common that foreground hairs will only become "partially transparent", so that the hairs are still visible but background detail shows "through" them. At present, there are no methods in ZS (or anywhere else that I know of) that will avoid this problem. It is an outstanding research issue. See the discussion at http://www.photomacrography.net/forum/v ... php?t=4903 for more information.

Image #3 looks pretty good except for those dust trails that you mentioned. What method(s) have you been using to clean your sensor? I don't clean my sensor very often except for blowing it off, but when I do have to wet-clean it, I've had success with Eclipse and Pec*Pads as described at http://www.cleaningdigitalcameras.com/.

For advice, my best recommendations at this point are to 1) be sure that the area of interest is covered by all frames, and 2) use PMax, not DMap. I don't think we have discussed the importance of diffuse illumination, but that will appear later if it has not already.

--Rik

[scitch]

Thanks for the great information.

I usually have Zerene do both methods and by eye, I choose the better of the two. Next time I'm on that computer, I'll check and see if I have the PMax method saved. I can see from the title of the second photo that it was DMap.

The comment about the hairs was just relative to the original photos before stacking. The hairs have much better detail, texture, and lighting in the originals than they do in the stacked photo. I can try the PMax and see if it helps.

I realized that I've had dust on the sensor for some time, it really didn't become noticeable until I started taking pictures through the microscope. I've tried blowing it out. I've tried special cotton swabs with 91% isopropyl alcohol. I've tried electrostatic brushes. I've tried lens cleaning cloth. I finally got all of the residue off from the wet methods not working. I might just take it in to get it professionally cleaned and then get some Eclipse to maintain it from now on. That picture had far more dust than normal, so I suspect that some of it was on the microscope optics, not the sensor.

I've tried using tissue paper to diffuse the light, but haven't seen much difference. The light has a mostly-transparent, blue glass disc that might filter some of the yellow out of the light but doesn't diffuse much. I've seen the ping-pong ball method mentioned quite a bit and might give that a shot. I've had trouble with shadows too. Would you recommend one of the ring-shaped fluorescent lights to help eliminate that?

Thanks again!
Mike

[rjlittlefield]

I've tried using tissue paper to diffuse the light, but haven't seen much difference.

Hhmm... It occurs to me that although I've often said "use paper", I have not often if ever illustrated what I mean by that. It's kind of late tonight to fill that gap with a picture, so let me try by words. The idea is not to cover the end of the light source in paper. Instead the idea is to enclose the subject in a fairly small "tent" of paper, then illuminate all or most of the tent so that the subject sees light coming from all directions.

Having light reach the subject from a wide range of angles instead a narrow range can make a striking difference in how clearly the subject gets rendered. The example HERE and in the followup posts illustrates how strong the effect can be. But the paper shown there still covers a relatively small area compared to what I usually use, where significant light reaches the subject from a span of typically 180 degrees by 90 degrees -- 1/8 of the full sphere.

What I notice in your second image is that most of the surface of the fly is covered with small colored dots. That is a hallmark of directional light. The colors are "false" in the sense that they do not represent intrinsic color of the cuticle but rather angle- and wavelength-dependent interactions between the illumination, the cuticle, and sometimes the lens. Under diffused illumination, the surface will probably appear textured but not colored.

I would not recommend buying anything new until you have played quite a bit with things that you probably already have available. An ordinary desk lamp placed close to the paper tent may provide enough illumination. For any light source, a white or aluminum foil reflector on the shadow side can greatly increase its coverage. Many of NikonUser's flies are illuminated by a single flash unit that is simultaneously diffused and reflected by a white styrofoam cup. If you don't have a separate flash unit, you can probably use the flash unit built into your camera by building an aluminum-foil lined "snoot" that channels light from the camera's flash down to a diffuser wrapped around the subject. I don't have links handy that illustrate these ideas, but you should be able to find at least a few by using the forum's search facility.

--Rik

[scitch]

Thanks, the lighting discussion is very helpful. I did misunderstand the tissue paper diffuser.

Because I enjoy photographing insects, I recently purchased a really cheap bellows and macro lenses from eBay. The lighting discussion will certainly apply to that as well. I need one more adapter to make them connect to my camera and I'll post some samples. I think macro will be easier for insects than micro although I've seen both on this forum.

Thanks again,
Mike