Yes, I did that in Photoshop. The steps are clearly visible from pic to pic and seems to be ok to me. The pics showed above are no stacks. This are the sharpest single frames for each corner from the stack. But I also tried to stack it with DMap and the results was comparable to the single shots...
It is also reproducable. I tried it several times always with the same result. When it moves, for what reason ever, I would not expect it always at the same time (it also was not always on the same position on the micrometer. I tried it in several positions always with the same results)...
Some trouble with stacking artifacts...
Moderators: rjlittlefield, ChrisR, Chris S., Pau
I think I can see several things.
I'm probably imagining some of them
When you focus "through" the sharp points do the specular highlights go from \ to / direction? If so, the stacker can combine them to X. That's a type of astigmatism.
see this eye:
http://photomacrography.net/forum/viewt ... 679#150679
Is it always OK in the centre?
It would be worth trying flash, because those "long" marks look like movement, too. Also diffusing the light really a lot, to avoid directional refections.
Is it a new Mitutoyo?
I'm probably imagining some of them
When you focus "through" the sharp points do the specular highlights go from \ to / direction? If so, the stacker can combine them to X. That's a type of astigmatism.
see this eye:
http://photomacrography.net/forum/viewt ... 679#150679
Is it always OK in the centre?
It would be worth trying flash, because those "long" marks look like movement, too. Also diffusing the light really a lot, to avoid directional refections.
Is it a new Mitutoyo?
Chris R
-
rjlittlefield
- Site Admin
- Posts: 23598
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Markus, I apologize for the late greeting, but welcome aboard!
I'm the fellow who wrote Zerene Stacker and I answer all the support emails, so hopefully I can address those questions for you.
Taking the problems that you show at http://www.photomacrography.net/forum/v ... 768#182768...
1. DMap of the fibers far in front of background, what you're seeing are abnormally harsh "loss-of-detail" halos caused by the depth map being unable to slew quickly over many source frames. The "perfect" depth map would have cliffs at the edges of the fibers, and that's not a structure that Zerene Stacker (or any other software that I know) is able to compute accurately. Your best approach in this situation is probably to stack in slabs and then use manual retouching to clean up the inevitable visibility errors. The stairstepping along two of the fibers is an artifact that I don't recall seeing before, and I don't know for sure what's happening with that.
2. DMap of the purple surface should be simple for Zerene Stacker to handle. The fact that there is nothing sharp anywhere strongly suggests that the source images are bad. Even if the alignment process were failing, I would still expect sizable regions where the output image was just a copy of a single input image. Assuming that's true, then no sharp output means no sharp input.
3. PMax of the brown surface with bright streaks, those streaks are typical of what PMax does with very bright specular highlights when shooting at high magnification. The basic problem is that such highlights are very often due to light rays that enter the objective far off center, and that causes the specular highlight to move laterally as it goes into and out of focus. Essentially the highlight is being seen through an aperture that is far off center, so it's like looking through one half of a stereo microscope and seeing that the image shifts sideways as you focus up and down. It is very difficult to get a good rendering of such a highlight, because each of the shifted versions looks like detail to be preserved. PMax will try to preserve them all, giving a smooth streak; DMap will probably preserve only a few of them, giving sort of concentric halos. To get a completely clean rendition, retouching from a single source image is the only approach that I know.
Such troublesome highlights can be minimized by using generally quite diffuse illumination, with some slightly brighter areas positioned to put reflections on facets that you care about. Forum member ploum is a master of this approach. His illumination system is shown at http://www.photomacrography.net/forum/v ... 005#175005, and you can see from his images in the same thread what superb results it produces.
4. PMax of bright fibers, those bright streaky halos are typical of PMax and shiny fibers. They result from the same problem with specular highlights that is described above. Again, more diffuse illumination can help. For more explanation, see the discussion at "Reflections of hard and soft light...".
5. PMax of the purple surface. Again, this area should be simple for both PMax and DMap. PMax would be more vulnerable to errors in alignment, because PMax is relentless about trying to show focused detail at every location where it is ever seen. But this result mostly looks like fuzzy source images.
Regarding your odd problem with reproducible blurs, I have no great ideas there. I am not sure that we understand correctly just how reproducible they are.
Instead of shooting a stack and then searching through to find the sharpest parts, you might get more insight by shooting a lot of exposures, all at the same focus position, with nothing changing except the shutter modes and exposure times, and with lots of repetition -- say 10 exposures for each combination of settings. The variation between groups, versus the variation within groups, will probably tell you something. At the least, that test would show clearly that you really are seeing systematic variation and not just misinterpreting random variation as systematic due to not enough samples. (I have made that mistake many times.)
--Rik
I'm the fellow who wrote Zerene Stacker and I answer all the support emails, so hopefully I can address those questions for you.
Taking the problems that you show at http://www.photomacrography.net/forum/v ... 768#182768...
1. DMap of the fibers far in front of background, what you're seeing are abnormally harsh "loss-of-detail" halos caused by the depth map being unable to slew quickly over many source frames. The "perfect" depth map would have cliffs at the edges of the fibers, and that's not a structure that Zerene Stacker (or any other software that I know) is able to compute accurately. Your best approach in this situation is probably to stack in slabs and then use manual retouching to clean up the inevitable visibility errors. The stairstepping along two of the fibers is an artifact that I don't recall seeing before, and I don't know for sure what's happening with that.
2. DMap of the purple surface should be simple for Zerene Stacker to handle. The fact that there is nothing sharp anywhere strongly suggests that the source images are bad. Even if the alignment process were failing, I would still expect sizable regions where the output image was just a copy of a single input image. Assuming that's true, then no sharp output means no sharp input.
3. PMax of the brown surface with bright streaks, those streaks are typical of what PMax does with very bright specular highlights when shooting at high magnification. The basic problem is that such highlights are very often due to light rays that enter the objective far off center, and that causes the specular highlight to move laterally as it goes into and out of focus. Essentially the highlight is being seen through an aperture that is far off center, so it's like looking through one half of a stereo microscope and seeing that the image shifts sideways as you focus up and down. It is very difficult to get a good rendering of such a highlight, because each of the shifted versions looks like detail to be preserved. PMax will try to preserve them all, giving a smooth streak; DMap will probably preserve only a few of them, giving sort of concentric halos. To get a completely clean rendition, retouching from a single source image is the only approach that I know.
Such troublesome highlights can be minimized by using generally quite diffuse illumination, with some slightly brighter areas positioned to put reflections on facets that you care about. Forum member ploum is a master of this approach. His illumination system is shown at http://www.photomacrography.net/forum/v ... 005#175005, and you can see from his images in the same thread what superb results it produces.
4. PMax of bright fibers, those bright streaky halos are typical of PMax and shiny fibers. They result from the same problem with specular highlights that is described above. Again, more diffuse illumination can help. For more explanation, see the discussion at "Reflections of hard and soft light...".
5. PMax of the purple surface. Again, this area should be simple for both PMax and DMap. PMax would be more vulnerable to errors in alignment, because PMax is relentless about trying to show focused detail at every location where it is ever seen. But this result mostly looks like fuzzy source images.
Regarding your odd problem with reproducible blurs, I have no great ideas there. I am not sure that we understand correctly just how reproducible they are.
Instead of shooting a stack and then searching through to find the sharpest parts, you might get more insight by shooting a lot of exposures, all at the same focus position, with nothing changing except the shutter modes and exposure times, and with lots of repetition -- say 10 exposures for each combination of settings. The variation between groups, versus the variation within groups, will probably tell you something. At the least, that test would show clearly that you really are seeing systematic variation and not just misinterpreting random variation as systematic due to not enough samples. (I have made that mistake many times.)
--Rik
Chris, no, there is no astigmatism visible over the entire field when I´m focussing trough. There is only a very small amount of coma in the very outer corners when it´s slightly defocused, but that is to expect on that huge field of view.
Well, if I look to the very small, bright dots on black background they seemed to be a little elongated in the unsharp pics. That let assume some vibrations. But, why only in one or two (unsharp) pics?
When I use live view and go anywhere in the field of view, and focus the pic while using the magnification function of the camera control software, I get pretty sharp results seen on my screen. So I don´t think it´s an optical problem...
Rik, thanks for your welcome. It´s very helpful to have the developer of that pretty nice software here. I think you are right when you assuming that there have not been pretty sharp pics in the input images. I never controlled it all, but it sounds plausible to me (but the yellow needles, I looked for in some pics, seemed sharp to me...). But now the kings question: Why? I did nothing different to other stacks, including the test with that CMOS device above. And as you can see, there are pretty sharp pics included in the stack.
When I say reproducible, I do it like that:
1. I focus on the highest point.
2. I start an automated stack with 10 exposures
3. I focused new on the highest point
4. I change the shutter mode
5. I start an other automated stack
I do all the steps remote controlled through my computer. I never touched the camera...
Than I turn to an other position of the micrometer and start the procedure above new. I did that several times, and the results are very comparable to each other. It is always the same corner unsharp (depending on the shutter mode).
When I load the 10 pics to Photoshop and look to them in 100% view, I switch between them to find the sharpest one for each of the nearest corner and the deepest one. I´ll do a complete stacked image and will show it here, so you can see, when the unsharp region will beginn to suffer the image. I also tried to start the stack a bit out of focus, let it run through the different focus planes on the object and let it run out of focus at the lowest point of the object, to determine wether it´s only a problem of the first image, but the results are quiet the same. I don´t also think it´s a problem of increasing vibrations through resonances, because, depending on the shutter mode, the unsharp pics are at the beginning of the stack and otherwise at the end of the stack. At all, I think that´s not (only?) a mechanical problem of my rig, but it maybe is a mechanical problem in the camera itself (maybe distorsion of electronic boards while the shutter motors will give a torque moment when opening the different shutters/mirrors?). At the end, I could not imagine what cause such an effect. I´m very curious about that
Cheers Markus
Well, if I look to the very small, bright dots on black background they seemed to be a little elongated in the unsharp pics. That let assume some vibrations. But, why only in one or two (unsharp) pics?
When I use live view and go anywhere in the field of view, and focus the pic while using the magnification function of the camera control software, I get pretty sharp results seen on my screen. So I don´t think it´s an optical problem...
Rik, thanks for your welcome. It´s very helpful to have the developer of that pretty nice software here. I think you are right when you assuming that there have not been pretty sharp pics in the input images. I never controlled it all, but it sounds plausible to me (but the yellow needles, I looked for in some pics, seemed sharp to me...). But now the kings question: Why? I did nothing different to other stacks, including the test with that CMOS device above. And as you can see, there are pretty sharp pics included in the stack.
When I say reproducible, I do it like that:
1. I focus on the highest point.
2. I start an automated stack with 10 exposures
3. I focused new on the highest point
4. I change the shutter mode
5. I start an other automated stack
I do all the steps remote controlled through my computer. I never touched the camera...
Than I turn to an other position of the micrometer and start the procedure above new. I did that several times, and the results are very comparable to each other. It is always the same corner unsharp (depending on the shutter mode).
When I load the 10 pics to Photoshop and look to them in 100% view, I switch between them to find the sharpest one for each of the nearest corner and the deepest one. I´ll do a complete stacked image and will show it here, so you can see, when the unsharp region will beginn to suffer the image. I also tried to start the stack a bit out of focus, let it run through the different focus planes on the object and let it run out of focus at the lowest point of the object, to determine wether it´s only a problem of the first image, but the results are quiet the same. I don´t also think it´s a problem of increasing vibrations through resonances, because, depending on the shutter mode, the unsharp pics are at the beginning of the stack and otherwise at the end of the stack. At all, I think that´s not (only?) a mechanical problem of my rig, but it maybe is a mechanical problem in the camera itself (maybe distorsion of electronic boards while the shutter motors will give a torque moment when opening the different shutters/mirrors?). At the end, I could not imagine what cause such an effect. I´m very curious about that
Cheers Markus
OK... 
These jags/zigzags
remind me of a stack I did where the aligner became confused, because of what happened to be in focus in each frame.
Are these one "jag" per input image?
You could try turning the scaling or x-y (or rotation which is usually best left off) off, to see what you see...
These jags/zigzags
remind me of a stack I did where the aligner became confused, because of what happened to be in focus in each frame.
Are these one "jag" per input image?
You could try turning the scaling or x-y (or rotation which is usually best left off) off, to see what you see...
Last edited by ChrisR on Fri Jan 08, 2016 6:35 am, edited 1 time in total.
Chris R
I don´t know. That was a 300 layer stack...Are these one "jag" per input image?
Ok, I´ll do that.You could try turning the scaling or x-y (or rotation which is usually best left off) off, to see what you see...
I have done an other test with the CMOS device as written above. In the pics you can see first the entire field and than the 100% crops from all edges and the middle. All pics are ZS DMap stacks with 7 layers.
Mirror lockup with 10s delay to the shots:
EFSC with 10s delay to the shots:
Hope this helps to understand. I´m further confused...
Cheers Markus
Hi Marcus
It is possible to get (shutter) vibrations that looks different in different
parts of the picture. If I do not use EFSC my system is sensitive to this
around 1/100 sek.
Below is a series of pictures to show what I mean.
The first to pairs; 1/125 sek non diffused LED vs non diffused flash viewed at 200%
first bottom left then upper left.
The last pair is a comparison between flash and diffused flash.
Canon 5 dII, mitutoyo 10x, reversed Raynox DCR-150.
It would be interesting if you could test with a diffused flash.
Regards Jörgen
It is possible to get (shutter) vibrations that looks different in different
parts of the picture. If I do not use EFSC my system is sensitive to this
around 1/100 sek.
Below is a series of pictures to show what I mean.
The first to pairs; 1/125 sek non diffused LED vs non diffused flash viewed at 200%
first bottom left then upper left.
The last pair is a comparison between flash and diffused flash.
Canon 5 dII, mitutoyo 10x, reversed Raynox DCR-150.
It would be interesting if you could test with a diffused flash.
Regards Jörgen
Hi etalon,
Lots of good advice here. I may be able to add somethings as I typically shoot silicon chips at very high resolutions. I use a Nikon D800 camera so I don't have the EFSC.
I have spent years studying, calculating and experimenting trying to get better chip images with the help from some of the folks here. What I have evolved to is a setup with 2 light tents, one inside the other, four 2 foot square light boxes and a bunch of umbrellas with strobes to even the light out. Uniform lighting is very important to get good images of chips.
I use a sequence of: move the camera/lens, wait a couple seconds, flip the mirror up, wait a couple seconds, trigger the shutter opening with a 1~2 second exposure and trigger all the strobes on the rear curtain closing. This is done in the dark so the shutter being open for 1~2 seconds does not expose the sensor, but the strobe flash of light does and it allows the front curtain vibrations to die off before the actual image is captured. Leaving the shutter open for this long does introduce some noise but with proper exposure the noise isn't an issue.
Sources of vibration include Air Conditioning, walking around, TV sound, doors opening/shutting, pool pump, to name a few. I try and minimize these vibration sources as best I can.
Hopefully some of this will help, please keep us posted on your progress.
Cheers,
Mike
Lots of good advice here. I may be able to add somethings as I typically shoot silicon chips at very high resolutions. I use a Nikon D800 camera so I don't have the EFSC.
I have spent years studying, calculating and experimenting trying to get better chip images with the help from some of the folks here. What I have evolved to is a setup with 2 light tents, one inside the other, four 2 foot square light boxes and a bunch of umbrellas with strobes to even the light out. Uniform lighting is very important to get good images of chips.
I use a sequence of: move the camera/lens, wait a couple seconds, flip the mirror up, wait a couple seconds, trigger the shutter opening with a 1~2 second exposure and trigger all the strobes on the rear curtain closing. This is done in the dark so the shutter being open for 1~2 seconds does not expose the sensor, but the strobe flash of light does and it allows the front curtain vibrations to die off before the actual image is captured. Leaving the shutter open for this long does introduce some noise but with proper exposure the noise isn't an issue.
Sources of vibration include Air Conditioning, walking around, TV sound, doors opening/shutting, pool pump, to name a few. I try and minimize these vibration sources as best I can.
Hopefully some of this will help, please keep us posted on your progress.
Cheers,
Mike
Hi Jörgen,
that is a very interesting comparison. I´ll try it with a flash, but it will take time, because I have at the moment no possibility to make the setup completely dark (the PC monitor is always illuminating).
Mike, thanks for your help. That sounds also interesting, but that require a lot of room and a lot of work before you can take a pic if you can´t let it assembled. So this is sadly not possible for me.
Chris, I did the stack you required with disabled x, y and rotation, but the results are more worst:
DMap
PMax
Yesterday I also tried three stacks from an Azurite crystal, captured with very diffuse light, but what the stacking software have made from the frames are scary.
Well, it seems to me, that a DSLR is not a good choice for microphotography, so I´ll looking for a mirrorless camera. I think, the Olympus EM5 mkII would be a good choice...
Cheers Markus
that is a very interesting comparison. I´ll try it with a flash, but it will take time, because I have at the moment no possibility to make the setup completely dark (the PC monitor is always illuminating).
Mike, thanks for your help. That sounds also interesting, but that require a lot of room and a lot of work before you can take a pic if you can´t let it assembled. So this is sadly not possible for me.
Chris, I did the stack you required with disabled x, y and rotation, but the results are more worst:
DMap
PMax
Yesterday I also tried three stacks from an Azurite crystal, captured with very diffuse light, but what the stacking software have made from the frames are scary.
Well, it seems to me, that a DSLR is not a good choice for microphotography, so I´ll looking for a mirrorless camera. I think, the Olympus EM5 mkII would be a good choice...
Cheers Markus
It shouldn't be a problem. If you use a small "wide angle from subject viewpoint" diffuser like a table tennis ball, with tubes of paper around it as well, you can put the flash very close.no possibility to make the setup completely dark
You're working at effective f/20 or so, 1/250th, low iso...
It seems something is fundamentally different about what you're doing, which is giving an unusual combined effect.
Chris R
Why?etalon wrote:Well, it seems to me, that a DSLR is not a good choice for microphotography, so I´ll looking for a mirrorless camera.
Most people here use DSLRs. Of course for most micro work and many macros you don't need the optical viewfinder and a modern mirroless with EFSC will work nicely, but a Canon EOS 5D mkII also has it and is an excellent tool for this work. The shutter of a mirrorless is similar to the one of a DSLR.
I think you will do better concentrating in illumination, stacking parameters and retouching, and isolating any posible vibration source.
The corner unsharpness maybe due to the lack of good coverage of your objective in FF sensor (you may need to crop the image) but also to some misalignement of the optical setup or from a defective objective (one corner sharper than the other doesn't sound right). I think very unlikely that it could be due to the shutter, if I understand well you've tested with different shutter speeds and illuminations.
Pau
Chris, if I understood right, than it works only when exposure time is in seconds range with the flash on the second curtain. This require to shelter from ambient light to avoid exposure the sensor in the time vibrations are not damped out, or I´m wrong?It shouldn't be a problem. If you use a small "wide angle from subject viewpoint" diffuser like a table tennis ball, with tubes of paper around it as well, you can put the flash very close.
I don´t understand this...It seems something is fundamentally different about what you're doing, which is giving an unusual combined effect.
Pau, because the 5DII have big and heavy shutters compared with a smaller camera and also a high mass. It also have a limited live time of the mechanical shutters and I make a lot of pics with focus stacking...Why?
Yes, I agree completely. But that can also be done while change the camera.I think you will do better concentrating in illumination, stacking parameters and retouching, and isolating any posible vibration source.
No, that´s not correct. If there where a damage in optics, I never can get a sharp image over the whole FOV without making changes of the camera and optics. But I did! Depending on the shutter mode, on time is the lower part sharp, one time the upper without touching/changing anything (except the shutter mode via remote control)!The corner unsharpness maybe due to the lack of good coverage of your objective in FF sensor (you may need to crop the image) but also to some misalignement of the optical setup or from a defective objective
That´s correct.if I understand well you've tested with different shutter speeds and illuminations.
I agree completely, but not for every application...the 5D2 is fundamentally good!
Cheers Markus
Hi Markus
My suggestion about testing with a flash is just to see if there is vibrations that causes the problems you see in your pictures. If it is - there is on this forum a lot of discussion about how to minimize it - one easy way is flash.
If you shoot with a flash at flash sync speed there is probably not a problem to do it in daylight. Very little ambient light will reach the sensor and the flash will freeze the movements. If there is vibrations every picture will be misaligned. Usually the misalignment is taken care of by the stacking program. Continuous flashing, several hundred times in a row can of course be a disturbance to neighbors so a dark room might be a good idea anyway.
Regards Jörgen
My suggestion about testing with a flash is just to see if there is vibrations that causes the problems you see in your pictures. If it is - there is on this forum a lot of discussion about how to minimize it - one easy way is flash.
If you shoot with a flash at flash sync speed there is probably not a problem to do it in daylight. Very little ambient light will reach the sensor and the flash will freeze the movements. If there is vibrations every picture will be misaligned. Usually the misalignment is taken care of by the stacking program. Continuous flashing, several hundred times in a row can of course be a disturbance to neighbors so a dark room might be a good idea anyway.
Regards Jörgen