Poor man's Repro-Nikkor (f/1.0, m=1x)
Moderators: rjlittlefield, ChrisR, Chris S., Pau
Poor man's Repro-Nikkor (f/1.0, m=1x)
I've been thinking about how to achieve 80 Mp images at m=1 on an MFT camera without getting diffraction. That means the effective aperture has to be EA=4 or less. As I've written elsewhere, coupled lenses will be best for this, because of their big advantage in EA at this magnification. There are several considerations for choosing the lenses. The forward-facing lens should ideally be made for MFT so that it corrects properly for the thick filter stack in front of the MFT sensor (thick as window-glass!!). The reverse lens should also be made for small sensors, because these should (all else being equal) be made to higher tolerances than lenses for bigger sensors. But the front lens should ideally NOT be MFT because of the very short flange-to-subject distance for these lenses. Also ideally, the two lenses should be identical so that some of their aberrations cancel out. Also the lenses should be cheap...
The Oly 50mm f/2 Four Thirds lens was my choice for this. I don't really know what size sensor filter stack it is designed for, but the Four Thirds Oly sensor stack is probably not that different from the Oly Micro Four Thirds (MFT) sensor stack. It is one of the fastest macro lenses available, which is needed to meet my goal of no diffraction at 80Mp. And best of all, Four Thirds lenses have a similar flange-to-sensor distance as Nikon, twice as long as MFT, so I have lots of working distance between the reversed lens and my subject. One disadvantage is that I should shoot through a sheet of glass with the same thickness as the sensor filter stack that the lens expects. But I can live with that.
Here is my first test. I've left both lenses wide open at f/2.0 and placed a paper iris in the center between them, stopping the composite system down by about 1 stop. When the lenses are wide open, the composite lens acts as if it had an aperture of f/1, with an effective aperture of 2.0, just like the Repro-Nikkor. I have stopped it down to about EA=2.8 with this paper aperture. It is also symmetric just like the Repro-Nikkor.
I see that I get a slight corner vignetting. Apart from that, the results look pretty good. This DMap stack is made under continuous light usings jpgs straight from the camera, and retouched with the PMax image. The result was sharpened slightly in Photoshop. I think it would be even sharper if done with flash. I shot it through two filters in case the lens was expecting a thick filter stack. I should try it again without that...
This is still just a 20Mp image. Overnight I will try the real test at 80Mp, now that I have something which is almost pixel-perfect at 20Mp..
Full image reduced for forum:
100% crops:
The Oly 50mm f/2 Four Thirds lens was my choice for this. I don't really know what size sensor filter stack it is designed for, but the Four Thirds Oly sensor stack is probably not that different from the Oly Micro Four Thirds (MFT) sensor stack. It is one of the fastest macro lenses available, which is needed to meet my goal of no diffraction at 80Mp. And best of all, Four Thirds lenses have a similar flange-to-sensor distance as Nikon, twice as long as MFT, so I have lots of working distance between the reversed lens and my subject. One disadvantage is that I should shoot through a sheet of glass with the same thickness as the sensor filter stack that the lens expects. But I can live with that.
Here is my first test. I've left both lenses wide open at f/2.0 and placed a paper iris in the center between them, stopping the composite system down by about 1 stop. When the lenses are wide open, the composite lens acts as if it had an aperture of f/1, with an effective aperture of 2.0, just like the Repro-Nikkor. I have stopped it down to about EA=2.8 with this paper aperture. It is also symmetric just like the Repro-Nikkor.
I see that I get a slight corner vignetting. Apart from that, the results look pretty good. This DMap stack is made under continuous light usings jpgs straight from the camera, and retouched with the PMax image. The result was sharpened slightly in Photoshop. I think it would be even sharper if done with flash. I shot it through two filters in case the lens was expecting a thick filter stack. I should try it again without that...
This is still just a 20Mp image. Overnight I will try the real test at 80Mp, now that I have something which is almost pixel-perfect at 20Mp..
Full image reduced for forum:
100% crops:
Thanks Curt, I am processing and stacking the hi-res ones right now.
I recently got a 4x 0.20 infinity-corrected Nikon PlanApo but I have yet to test its coverage, because I didn't have the right adapter for it. I figured that it would not be able to cover even the MFT sensor on a 50mm tube. But I will try it. Does your 4x 0.10 cover the sensor well with a 50mm tube lens?
I recently got a 4x 0.20 infinity-corrected Nikon PlanApo but I have yet to test its coverage, because I didn't have the right adapter for it. I figured that it would not be able to cover even the MFT sensor on a 50mm tube. But I will try it. Does your 4x 0.10 cover the sensor well with a 50mm tube lens?
I've done the hi-res stack. Looks nice. At the reduced size of this forum it looks just like the non-hi-res version I posted above. To see the difference it has to be blown up to 100%, and compared with the previous image blown up to 200%. Here is the same section of the wing, blown up to 100% from the hi-res version, and 200% from the normal (but still very nice) version. Note that some of the difference (especially in tonal range) is due to the use of jpgs in the normal image and Tiffs made from Raw files in the hi-res case.
First, the 200% enlargement of the normal image:
Next, a 100% enlargement of the same section (same FOV) of the hi-res version:
First, the 200% enlargement of the normal image:
Next, a 100% enlargement of the same section (same FOV) of the hi-res version:
Lou,
Very interesting & clever solution!! How do you think resolution-wise this 50mm back to back same lens type will compare to a dedicated macro at 1:1, say maybe the Nikon 105mm VR on a DX or APC sensor?
I have an old Nikon 50mm F1.8D that might be a start, these are pretty cheap (~$120 new), and have 52mm filter threads.
Best,
Mike
Very interesting & clever solution!! How do you think resolution-wise this 50mm back to back same lens type will compare to a dedicated macro at 1:1, say maybe the Nikon 105mm VR on a DX or APC sensor?
I have an old Nikon 50mm F1.8D that might be a start, these are pretty cheap (~$120 new), and have 52mm filter threads.
Best,
Mike
Zzffnn, yes, these are not very expensive nowadays on the used market, though you can still get them new for $500.
Mike, the filter glass has to go between the rear element of the reversed lens and the subject. The lens was designed with the expectation that there would be a thick filter stack on top of the sensor.
In this test I used a single filter instead of two. I still don't know how big the ordinary Four Thirds filter stack is. All I can find is the depth of the Micro Four Thirds stack. Probably they are the same....
Mike, the filter glass has to go between the rear element of the reversed lens and the subject. The lens was designed with the expectation that there would be a thick filter stack on top of the sensor.
In this test I used a single filter instead of two. I still don't know how big the ordinary Four Thirds filter stack is. All I can find is the depth of the Micro Four Thirds stack. Probably they are the same....
Mike, I think this 50mm lens is very good; it is a macro lens, not just a regular 50mm. I chose it because it was reasonably sharp wide open, has a very flat field, might be approximately corrected for an MFT sensor filter stack, and has double the working distance in reverse versus any native MFT lens. It uses ED glass too. Reviews have said this was one of the sharpest lenses ever tested, and user comments were mostly glowing. Plus it's cheap used.
Here it is:
https://www.bhphotovideo.com/c/product/ ... ro_ED.html
Here it is:
https://www.bhphotovideo.com/c/product/ ... ro_ED.html
I think this would beat it handily. Most consumer macro lenses are optimized for 1:10-1:2, not 1:1. Since this combo does so well on a camera with tiny pixels, I expect it would be just about flawless on a sensor with larger pixels.How do you think resolution-wise this 50mm back to back same lens type will compare to a dedicated macro at 1:1, say maybe the Nikon 105mm VR on a DX or APC sensor?
Except that it won't cover the whole sensor!!
But there are ways to do this same sort of thing in other formats.
Hi Lou,
The image comparison is night and day. That HR mode should get a lot of use.
Yes, the low power Nikon & Olympus Planapo objectives have a very small FOV. The 4x Plan has an unusually large FOV. It will cover a Canon APS-C sensor at 1x with a 50mm tube lens. You could probably push it a bit further with a 40mm on an MFT sensor. You have to unscrew the front housing of the objective or it will suffer from severe vignetting.
The image comparison is night and day. That HR mode should get a lot of use.
Yes, the low power Nikon & Olympus Planapo objectives have a very small FOV. The 4x Plan has an unusually large FOV. It will cover a Canon APS-C sensor at 1x with a 50mm tube lens. You could probably push it a bit further with a 40mm on an MFT sensor. You have to unscrew the front housing of the objective or it will suffer from severe vignetting.