gigamacro
Moderators: rjlittlefield, ChrisR, Chris S., Pau
gigamacro
This one is going to meander and ramble a bit, but I have lots of questions and I have to start somewhere. Feel free to comment on whatever you feel passionate about and ignore the rest.
Objective, in the resume sense
Let me begin at the ending. My goal is to create very large, gallery quality inkjet prints of very small subjects. My output device will be the Epson 11880 which takes 64" rolls. To be simpler with the math, let's assume I stick to square format and output 60" x 60" prints. At 300dpi, I will need a sensor size of 18000 x 18000 to fill the print.
So, I'm pretty much signed up for a stack and stitch process with a telecentric lens of some sort to patch together my virtual sensor.
Aside: I'm confident in the stack and stitch + output process because I've done it before with large panoramas as the subject. There I was using a Gigapan robotic mount and a simple Canon G10. The stacks there were for HDR tonemapping - here they will be for focus, obviously. There are significant differences, of course, but it's the same basic idea.
Assets
Equipment-wise, if anyone wants to play Apollo 13 (or The Princess Bride) with me, I have a collection of old but possibly useful parts: Canon 20d, EF-S 60mm, 11-22mm zoom, a monorail horseman 4x5 camera body (takes sinar lens boards), APO-Rodagon 50mm and 90mm enlarging lenses, a holocaust cloak, some ancient Minolta Maxxum lenses - 135mm, 50mm 1:1 macro, 20mm.
So, some purchases need to be made. I suspect that I could buy a few items and make a nice play set out of that, but to do what I really want to do it will be starting over from scratch.
Camera
If possible, I'd like to end up buying a camera body I would be happy using for general photography also. A mirrorless body appeals on that basis, but it may not tether how I would like it to. I would consider Fuji X-T1, Sony a7R, Oly OM-D E-M1. But if it has to be DSLR, I'm willing to use any Canikon that will do the job well. Opinions here welcome, but please say why as well as what.
Lens system
Has to be telecentric something, either purpose built or cobbled together. Does anyone here have experience with the Mitutoyo telecentrics? http://www.edmundoptics.com/microscopy/ ... tives/2481 I presume these can be made to work on camera like other microscope objectives, but I'd rather be sure...which brings me to:
Subjects (choosing a target magnification)
I expect my stitched sensor to be 18000 pixels square. I assume that whatever sensor I buy with have a pixel pitch of approximately 4.75 microns, so my virtual sensor will be about 3.5" square and a 10x lens has me filling that frame with subjects about 1/3" square. Has anyone put together a chart of the various common subjects and their sizes? I'm kind of agnostic about the actual subject material, although I would prefer to show a whole ordinary object with lots and lots of detail - plants, minerals, bugs are all fair game. I just don't want to cripple my choices by choosing the wrong magnification, since I will probably only be able to afford one good lens/objective to start. My intuition breaks down here, because I have very little practical experience. Suggestions wecome.
Lighting
Is there a good thread to read on macro lighting? There are so many variables before I get down here that I don't know whether I should be looking at flash or continuous, how much light I'll need, etc. I've done a fair amount of work with studio strobes and modelling lights, but I don't own any and it wasn't macro subjects, so again I don't trust my intuition fully. I do understand how to light subjects in general, the inverse square law, etc.
Rigging
Luckily, I'm in a ground floor unit with only a layer of paint between the floor and the concrete slab foundation. I live in the Bay Area, so it should be easy to pick up a used breadboard/table, but I have no idea what dimensions would be cramped comfortable, or ridiculously oversized for this work. Similarly, I have no idea whether I should be frankensteining a microscope focusing stage or just getting a stackshot or something else. I do know that I would like the stacks to be motorized/automated and the shifts to be geared (but manual should be fine there), but how that translates into real equipment to consider I am at a loss.
Control / postprocessing
I have a decently powered iMac which should do the trick. I don't mind running a Windows emulator if there's some great Windows-only tool, but I'd prefect not to if I can help it. I will buy and learn Zerene. Already own PTGui and Photoshop.
Time
I'd like to get decisions made and eqipment purchased in the next couple of months. I expect to be making finished prints by December. Any earlier is a bonus. That should leave the better part of a year to work through the issues.
Please, tell me if I'm leaving something out or if I'm making some wrong assumptions or calculations. I am confident that this will work, that it won't be easy, and that getting through it will be a blast.
Due to the print enlargement, I'll get another 17x from the sensor to the wall (60" print/ 3.5" sensor), so about 25 square feet of 170x life size on the print using the 10x example.
Objective, in the resume sense
Let me begin at the ending. My goal is to create very large, gallery quality inkjet prints of very small subjects. My output device will be the Epson 11880 which takes 64" rolls. To be simpler with the math, let's assume I stick to square format and output 60" x 60" prints. At 300dpi, I will need a sensor size of 18000 x 18000 to fill the print.
So, I'm pretty much signed up for a stack and stitch process with a telecentric lens of some sort to patch together my virtual sensor.
Aside: I'm confident in the stack and stitch + output process because I've done it before with large panoramas as the subject. There I was using a Gigapan robotic mount and a simple Canon G10. The stacks there were for HDR tonemapping - here they will be for focus, obviously. There are significant differences, of course, but it's the same basic idea.
Assets
Equipment-wise, if anyone wants to play Apollo 13 (or The Princess Bride) with me, I have a collection of old but possibly useful parts: Canon 20d, EF-S 60mm, 11-22mm zoom, a monorail horseman 4x5 camera body (takes sinar lens boards), APO-Rodagon 50mm and 90mm enlarging lenses, a holocaust cloak, some ancient Minolta Maxxum lenses - 135mm, 50mm 1:1 macro, 20mm.
So, some purchases need to be made. I suspect that I could buy a few items and make a nice play set out of that, but to do what I really want to do it will be starting over from scratch.
Camera
If possible, I'd like to end up buying a camera body I would be happy using for general photography also. A mirrorless body appeals on that basis, but it may not tether how I would like it to. I would consider Fuji X-T1, Sony a7R, Oly OM-D E-M1. But if it has to be DSLR, I'm willing to use any Canikon that will do the job well. Opinions here welcome, but please say why as well as what.
Lens system
Has to be telecentric something, either purpose built or cobbled together. Does anyone here have experience with the Mitutoyo telecentrics? http://www.edmundoptics.com/microscopy/ ... tives/2481 I presume these can be made to work on camera like other microscope objectives, but I'd rather be sure...which brings me to:
Subjects (choosing a target magnification)
I expect my stitched sensor to be 18000 pixels square. I assume that whatever sensor I buy with have a pixel pitch of approximately 4.75 microns, so my virtual sensor will be about 3.5" square and a 10x lens has me filling that frame with subjects about 1/3" square. Has anyone put together a chart of the various common subjects and their sizes? I'm kind of agnostic about the actual subject material, although I would prefer to show a whole ordinary object with lots and lots of detail - plants, minerals, bugs are all fair game. I just don't want to cripple my choices by choosing the wrong magnification, since I will probably only be able to afford one good lens/objective to start. My intuition breaks down here, because I have very little practical experience. Suggestions wecome.
Lighting
Is there a good thread to read on macro lighting? There are so many variables before I get down here that I don't know whether I should be looking at flash or continuous, how much light I'll need, etc. I've done a fair amount of work with studio strobes and modelling lights, but I don't own any and it wasn't macro subjects, so again I don't trust my intuition fully. I do understand how to light subjects in general, the inverse square law, etc.
Rigging
Luckily, I'm in a ground floor unit with only a layer of paint between the floor and the concrete slab foundation. I live in the Bay Area, so it should be easy to pick up a used breadboard/table, but I have no idea what dimensions would be cramped comfortable, or ridiculously oversized for this work. Similarly, I have no idea whether I should be frankensteining a microscope focusing stage or just getting a stackshot or something else. I do know that I would like the stacks to be motorized/automated and the shifts to be geared (but manual should be fine there), but how that translates into real equipment to consider I am at a loss.
Control / postprocessing
I have a decently powered iMac which should do the trick. I don't mind running a Windows emulator if there's some great Windows-only tool, but I'd prefect not to if I can help it. I will buy and learn Zerene. Already own PTGui and Photoshop.
Time
I'd like to get decisions made and eqipment purchased in the next couple of months. I expect to be making finished prints by December. Any earlier is a bonus. That should leave the better part of a year to work through the issues.
Please, tell me if I'm leaving something out or if I'm making some wrong assumptions or calculations. I am confident that this will work, that it won't be easy, and that getting through it will be a blast.
Due to the print enlargement, I'll get another 17x from the sensor to the wall (60" print/ 3.5" sensor), so about 25 square feet of 170x life size on the print using the 10x example.
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
intjester, welcome aboard!
I wonder, have you talked with the folks at http://gigamacro.com?
They're in the business of making equipment that does pretty much exactly what you want, and they're physically very close to you. I've got to think it would be a lot more efficient to rent time on their gear rather than struggling to cobble something together yourself.
--Rik
I wonder, have you talked with the folks at http://gigamacro.com?
They're in the business of making equipment that does pretty much exactly what you want, and they're physically very close to you. I've got to think it would be a lot more efficient to rent time on their gear rather than struggling to cobble something together yourself.
--Rik
That's certainly a backup plan, if plan A proves untenable. I think going that way alleviates some technical difficulties, but will introduce some stifling creative restrictions. If I were certain of the subject mattter, that way would be better, but I need time to explore and that time is available to me from 11pm to 1am on most days. There is the possibility to put together a conventional rig, work on a highly cropped section with just stacking, and then buy time for the final stack and stitch. But for that my list of questions is very much the same as it is now...rjlittlefield wrote:intjester, welcome aboard!
I wonder, have you talked with the folks at http://gigamacro.com?
They're in the business of making equipment that does pretty much exactly what you want, and they're physically very close to you. I've got to think it would be a lot more efficient to rent time on their gear rather than struggling to cobble something together yourself.
--Rik
Hello
Maybe this note in my blog (sorry, it's in french) can help you.
http://macromicrophoto.fr/dotcl/index.p ... roscopique
Motorized & automatized XYZ seems to be compulsory.
Panorama and stacking are very difficult.
One of my attempt
http://www.nikonsmallworld.com/gallerie ... etition/57
8 stacks of 250 shots were used - i stitched them with Photoshop, by hand, because my softs for panoramas were unable to do that.
Maybe this note in my blog (sorry, it's in french) can help you.
http://macromicrophoto.fr/dotcl/index.p ... roscopique
Motorized & automatized XYZ seems to be compulsory.
Panorama and stacking are very difficult.
One of my attempt
http://www.nikonsmallworld.com/gallerie ... etition/57
8 stacks of 250 shots were used - i stitched them with Photoshop, by hand, because my softs for panoramas were unable to do that.
A fellow who went on a somewhat similar journey. Not "micro".
here
(I wouldn't pick a Canon 180mm macro, the Sigma's sharper)
This is probably stating what you already know, but doing much of a range of sizes of subject with a small number of lenses is difficult. The "smaller " you go, the harder that gets. The Canon MPE-65 does cover, at high quality, 1:1 to about 1:4 on an APS sensor. Then microscope objectives take over. It's not telecentric of course.
here
(I wouldn't pick a Canon 180mm macro, the Sigma's sharper)
This is probably stating what you already know, but doing much of a range of sizes of subject with a small number of lenses is difficult. The "smaller " you go, the harder that gets. The Canon MPE-65 does cover, at high quality, 1:1 to about 1:4 on an APS sensor. Then microscope objectives take over. It's not telecentric of course.
The photo of using the binocular to help correctly set the added lens for telecentricity was very instructive. Thank you.ChrisR wrote:A fellow who went on a somewhat similar journey. Not "micro".
here
(I wouldn't pick a Canon 180mm macro, the Sigma's sharper)
It can be made to be so with an added lens, no?ChrisR wrote: This is probably stating what you already know, but doing much of a range of sizes of subject with a small number of lenses is difficult. The "smaller " you go, the harder that gets. The Canon MPE-65 does cover, at high quality, 1:1 to about 1:4 on an APS sensor. Then microscope objectives take over. It's not telecentric of course.
So that would cover subjects 3.5" down to a little smaller than 1".
If I start with mpe-65 for flexibility, what's the best way for motorizing the z axis? I remain unconvinced that x and y will require full automation since overlap is overlap, and parallax will be eliminated by making the optical system telecentric. However, I'm not sure how to build a focusing sled that shifts left / right and can also raise and lower in repeatable increments.
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Yes, with side effects. See HERE.intjester wrote:It can be made to be so with an added lens, no?
Reading between the lines about your skills and inclinations, I think you have two reasonable options. Both of them involve StackShot. One approach is to use the StackShot package, both controller and rail, using the rail for focus stepping. The second approach is to use just the StackShot controller, combined with a separate stepper motor set up to drive the fine focus knob of a microscope focus block. This latter option does not need to be complicated. See for example HERE. Note also the technique shown HERE, in which an extension platform is used to remote the XYZ position of a microscope stage so that it can be used with "open" illumination.If I start with mpe-65 for flexibility, what's the best way for motorizing the z axis?
BTW, at 10X NA 0.25 (and wider), you don't really need to worry about telecentricity of the optics. That's because the depth of field is so shallow that even if the optics are technically not telecentric, there still is no significant change of scale within the in-focus slab. In this case you can just turn off Scale correction in Zerene Stacker. However, to get perfectly orthographic projection, it's also necessary to turn off Shift X and Shift Y correction, and in order for that to work correctly your setup has to be very stable against shifts in the framing. That argues in favor of using a microscope focus block, as opposed to the StackShot rail which may have slight lateral shifts from frame to frame due to stickiness of the bushings on the side rails.
--Rik
Other way round, it goes from life-size, down.So that would cover subjects 3.5" down to a little smaller than 1".
Sorry, I meant 4x.
So with (say) a 20mm wide sensor that's a subject 20mm down to 5mm wide. It actually goes to 5x (4mm) but the aperture's a bit of a quality limit because of diffraction, with small sensor pixels. It was designed for 24x36 sensor film cameras of course, and still "fits" a larger sensor, larger pixel camera, well.
It only talks to a Canon body, and as I see Rik has (mid typing) covered, it wouldn't be easy to make TC. For thin subjects - a leaf say, it could be useful though.
Large stitched and stacked macro panoramas aren't hard to do, the failure rate is just high
While a telecentric lens would be nice, it's not necessary. Start with a spherical panorama setup. Mount a bellows such that the lens is fixed on the rotation point. Do the focus stacking by changing the bellows draw. Repeat for each frame in the panorama.
Here are a couple that I have printed at 30"X60". Both needed to be downsampled to print that small.
https://photosynth.net/view.aspx?cid=39 ... 67cfceae6a
https://photosynth.net/view.aspx?cid=0a ... d6ed5f3203
And a thread about the setup: http://www.photomacrography.net/forum/v ... p+panorama
While a telecentric lens would be nice, it's not necessary. Start with a spherical panorama setup. Mount a bellows such that the lens is fixed on the rotation point. Do the focus stacking by changing the bellows draw. Repeat for each frame in the panorama.
Here are a couple that I have printed at 30"X60". Both needed to be downsampled to print that small.
https://photosynth.net/view.aspx?cid=39 ... 67cfceae6a
https://photosynth.net/view.aspx?cid=0a ... d6ed5f3203
And a thread about the setup: http://www.photomacrography.net/forum/v ... p+panorama
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Much as I respect elf's work, that rotating bellows rig seems unnecessarily complicated for the purposes described by intjester. Low mag telecentric combos are quick and easy to make, once you learn the trick, and they let you use much simpler scanning mechanics that can be purchased off the shelf. It's a convenient coincidence that telecentricity becomes unnecessary, due to shallow DOF, at roughly the same point where telecentric combos become difficult to assemble.
intjester, I think you've been pointed to the major threads on this site regarding telecentric lenses. To recap, they are my series of three threads from 2006-2007 (OLDER, MIDDLE, NEWER), plus recent description HERE of a telecentric combo at 0.8X to 1.69X (and many other combinations, in the ensuing discussion), and the discussion HERE and following, in danibo's thread that ChrisR pointed to. The folks at gigamacro.com learned how to do their telecentrics from the NEWER thread.
By the way, it's hardly apparent at the size usually seen, but that moth wing image I use for my avatar is actually a severely reduced version of a 55 megapixel stack-and-stitched pano, shot with a 6.3 megapixel camera back in 2006. I ended up printing it two feet square at 305 dpi; it's been hanging on my wall ever since. The story on that is HERE. No telecentrics were used for that, but at about 2:1 at f/4 and a 38mm lens, it was right at the limit of relying on shallow DOF to hide perspective.
--Rik
intjester, I think you've been pointed to the major threads on this site regarding telecentric lenses. To recap, they are my series of three threads from 2006-2007 (OLDER, MIDDLE, NEWER), plus recent description HERE of a telecentric combo at 0.8X to 1.69X (and many other combinations, in the ensuing discussion), and the discussion HERE and following, in danibo's thread that ChrisR pointed to. The folks at gigamacro.com learned how to do their telecentrics from the NEWER thread.
By the way, it's hardly apparent at the size usually seen, but that moth wing image I use for my avatar is actually a severely reduced version of a 55 megapixel stack-and-stitched pano, shot with a 6.3 megapixel camera back in 2006. I ended up printing it two feet square at 305 dpi; it's been hanging on my wall ever since. The story on that is HERE. No telecentrics were used for that, but at about 2:1 at f/4 and a 38mm lens, it was right at the limit of relying on shallow DOF to hide perspective.
--Rik
I think we're in agreement. After all the stitching, the sensor is virtually embiggened to 3.5" square, so the 3.5" subject is 1:1 life size and the slightly under 1" subject projects at 4x to cover the same area.ChrisR wrote:Other way round, it goes from life-size, down.So that would cover subjects 3.5" down to a little smaller than 1".
Sorry, I meant 4x.
OK, I am convinced. But now I have to go down the microscope rabbit hole...rjlittlefield wrote:Yes, with side effects. See HERE.intjester wrote:It can be made to be so with an added lens, no?
Reading between the lines about your skills and inclinations, I think you have two reasonable options. Both of them involve StackShot. One approach is to use the StackShot package, both controller and rail, using the rail for focus stepping. The second approach is to use just the StackShot controller, combined with a separate stepper motor set up to drive the fine focus knob of a microscope focus block. This latter option does not need to be complicated. See for example HERE. Note also the technique shown HERE, in which an extension platform is used to remote the XYZ position of a microscope stage so that it can be used with "open" illumination.If I start with mpe-65 for flexibility, what's the best way for motorizing the z axis?
BTW, at 10X NA 0.25 (and wider), you don't really need to worry about telecentricity of the optics. That's because the depth of field is so shallow that even if the optics are technically not telecentric, there still is no significant change of scale within the in-focus slab. In this case you can just turn off Scale correction in Zerene Stacker. However, to get perfectly orthographic projection, it's also necessary to turn off Shift X and Shift Y correction, and in order for that to work correctly your setup has to be very stable against shifts in the framing. That argues in favor of using a microscope focus block, as opposed to the StackShot rail which may have slight lateral shifts from frame to frame due to stickiness of the bushings on the side rails.
--Rik
I really like the Olympus setup in your second link, above, with the exception of the lights not moving with the stage. Even with all the diffusion, I would still expect to get smearing of fine detail in the overlap regions if I move the stage instead of the camera since the lighting would then vary from stack to stack (unless they moved together with the stage as I mentioned). Maybe I'm worrying about nothing, but I'm trying to talk it out in the hopes of avoiding 4-digit mistakes.
- rjlittlefield
- Site Admin
- Posts: 23625
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
I share your concern at the conceptual level. However, take a look at http://www.photomacrography.net/forum/v ... php?t=1032 , visit the full size image HERE, examine it at actual pixels, and see if you find the transitions between rows troubling. I'm not sure how horizontal shift was done in that case, but I'm quite sure that vertical shift was done by moving the subject and not the lights. Bear in mind that in that case there was substantially less diffusion than you'll be needing to get good results at 10X, and the movement of the subject was roughly 5 times larger than you're talking about.intjester wrote:I really like the Olympus setup in your second link, above, with the exception of the lights not moving with the stage. Even with all the diffusion, I would still expect to get smearing of fine detail in the overlap regions if I move the stage instead of the camera since the lighting would then vary from stack to stack (unless they moved together with the stage as I mentioned). Maybe I'm worrying about nothing, but I'm trying to talk it out in the hopes of avoiding 4-digit mistakes.
What matters, of course, is angular shift in the lighting. So you'll be wanting to use relatively large diffusers, placed relatively far away from the subject. The classic pingpong ball is not the best approach for this problem, although to be honest, I suspect even the pingpong ball would work fine with only about 2 mm lateral shift per tile.
--Rik
Believe me, I have spent a lot of time considering that particular photo! It was what initially convinced me that I will be able to reach a satisfactory endpoint in this endeavor. It also reminds me that I'm years late to the party.rjlittlefield wrote:I share your concern at the conceptual level. However, take a look at http://www.photomacrography.net/forum/v ... php?t=1032 , visit the full size image HERE, examine it at actual pixels, and see if you find the transitions between rows troubling. I'm not sure how horizontal shift was done in that case, but I'm quite sure that vertical shift was done by moving the subject and not the lights. Bear in mind that in that case there was substantially less diffusion than you'll be needing to get good results at 10X, and the movement of the subject was roughly 5 times larger than you're talking about.
Cognisys now have their new three axis controller so you could use three Stackshot rails. Stack two rails on top of each other at 90 degrees and use the third for Z stacking. I've tried this and it works very well. http://www.photomacrography.net/forum/v ... +stackshot An example of two rails stacked is here http://www.photomacrography.net/forum/v ... hot+linden
Linden Gledhill http://www.flickr.com/photos/13084997@N03/