Reversed Teleconverter???

[PaulFurman]

The idea is some macro lenses can project a large image circle, but the resolution isn't that great so grab a larger area and squish it down to fit your sensor. Well, I tried it with a 1.4x Nikon teleconverter and placed the 21mm f/3.5 lens 45mm away like the sensor would be positioned... all that on front of the bellows and well, it seems to work though not conclusive. I might imagine this makes a larger aperture too though I suspect not. Exposures look about the same to me.

[PaulFurman]

Ugh, I may have got those mixed up in their order <g>. Anyways, the idea makes sense doesn't it?

[mgoodm3]

I don't think that reversing the lens will reverse the effect. It's still a negative power lens either way.

Will make the exit pupil smaller. and make the Pmag get smaller.

[rjlittlefield]

I'm a little confused. Does the light go from subject to reversed teleconverter to 21mm f/3.5 to sensor?

--Rik

[PaulFurman]

It's still a negative power lens either way.

<embarrassed look> OK thanks for pointing out what should have been obvious -just holding the converter up to my eye confirms that.

I still wonder if there is a way to do what I imagined though. I was playing around and that JML lens has a very large image circle. On the PB4 bellows I can tilt and shift it to the extreme limits without any vignetting or really obvious sharpness problems. Also with reversed 'normal' lenses, this approach should improve things - since you're only using a tiny central bit of the lens' field of view.

Another case where this idea would apply would be using a medium format lens with (my imaginary converter) to shrink the image down to 35mm format. I guess nobody has ever wanted to do that, but I'm sure it's possible.

[PaulFurman]

I'm a little confused.

You aren't the only one! My sample image above has the two shots reversed so yeah, it looks horrible with this idea. The 2x converter looks much worse.

I'm not easily quashed though, I tried a bunch of things, and finally got an image by mounting the JML 21 directly on the front of a 180mm f/2.8 lens, which gives about 9x mag at infinity or closest focus. Stopping down vignettes badly. Could this somehow be doing what I want? I was thinking about the idea of putting infinity microscope objectives on a tele lens, which I think is a completely different idea.

[rjlittlefield]

Could this somehow be doing what I want?

I can't tell, since I'm still completely unclear about what you want.

Tell me again, please, in different words, starting from the beginning?

--Rik

[PaulFurman]

Stitched panoramas would solve this issue but that requires a rig that has the lens fixed and the camera moves behind. I forget who had such a rig in this group, with all the aluminum stock and the big bag bellows...

That was Elf working with a smaller 4/3 format.

I made something simpler for infinity use, and am tempted to reconstitute it for this purpose.

[PaulFurman]

Could this somehow be doing what I want?

I can't tell, since I'm still completely unclear about what you want.

Tell me again, please, in different words, starting from the beginning?

--Rik

Messages got mixed up somehow. I wrote a reply saying that I want to make a larger format lens convert down to a smaller format, preserving the resolution and aperture. Video people have proved this is impractical - they use a rig that shoots the image projected on ground glass to translate shallow depth of field effects from larger format lenses. I assume this takes much longer exposures and loses resolution, so I've decided my idea is not workable.

[rjlittlefield]

I did see your earlier reply, which seems to have been replaced. Actually I'm reading a copy cached by my local browser, as I type this. I thought the earlier posting was well written, and it helped a lot. What does you mean, "Messages got mixed up somehow"?

Anyway, regarding your desire to have a large lens (wide aperture) project a small image (short focal length), you are correct that that cannot be done while preserving all the light and resolution that would otherwise go along with wide aperture and long focal length. I don't know how to explain why this is impossible, which of course is equivalent to saying I don't understand it myself. The best explanation I know is the one given by Rudolf Kingslake, where he essentially argues that the largest aperture a lens can have is equal to twice its focal length -- that being the diameter of a sphere whose radius is equal to the focal length. This is tied up somehow with wave theory of lenses, relating waves that hit the lens from the front with refracted waves as they hit the sensor. The only way around the limitation, I gather, is to break the intimate connection between incoming and outgoing waves. One way to accomplish that is to stick a diffusing plane into the system, hence the ground glass you mention. This does work to preserve the shallow depth of field of the wide aperture lens, but it definitely does not preserve the resolution in terms of cycles per field across the now smaller field. Whether it inevitably loses intensity as well, I'm not sure. My gut feeling is that loss of intensity is not a fundamental requirement of physics, just impractical and too expensive to be worth getting around. But I could be wrong about that.

The setup I first asked about had a reversed teleconverter out front. That probably is feasible while maintaining good image quality, though whether it does anything useful is a good question. Assuming proper positioning, what a reversed 1.4X TC in front will do is to reduce the magnification of the lens by a factor of 1.4X, while also making its entrance pupil smaller by just enough that the effective f-number (working aperture on the camera side) is unchanged. The camera side working aperture has to be unchanged because that depends solely on the extension and the exit pupil, neither of which are altered by adding the TC out front.

I hope this helps. Confusing stuff to think about.

--Rik

[elf]

Downsized panoramas almost always look better than single images with regard to resolution and sharpness, so I think it's a worthwhile exercise to find a simple method of getting more information per image.

Here's a 3 image pano done with the JML 21mm by shooting the center, then shifting 10mm left and 10mm right (about 90mm extension IIRC):



Original image size is 5320x1917 pixels.

Each frame was shot by moving the camera, not the subject. The left and right frames show quite a bit of smearing that is an artifact of stacking. The individual frames look fine, but the subject features move quite a bit relative to the center. The features at the outside of the image cone have to move more than the center.

To put it another way, in the center frame, the subject features at the left and right side will move towards the center as the bellows extension is decreased. In the left frame, the subject features on the right will move much less than the features on the left. The opposite will happen for the right hand frame.

I don't think Zerene Stacker (or any other stacking software) should be expected to handle this type of movement.

[PaulFurman]

Rik, my guess is I used the back button, replied again and overwrote my first reply.

Ugh, I won't even try addressing that conundrum further. Thanks for shedding some light on it.

I didn't try with the teleconverter on the very front, I still had the 21mm in front of the teleconverter, just that the converter was on the front of the bellows and reversed.

Elf,
I guess the problem is this ends up being a super-wide angle of view. The lens/pupil didn't move, so there's no parallax error when stitching single frames, but the magnification must have changed. I assume you stitched last so it's surprising you got smearing. I had that issue once with a similar shot but the floating flower parts in the middle got smeared while scaling was performed to match the background petals. I finally concluded a longer lens was the solution to such a deep shot, although I still held out hope that holding the lens/pupil stationary would have fixed it but apparently not. This setup should work for you with a flat subject though?

[elf]

I guess the problem is this ends up being a super-wide angle of view. The lens/pupil didn't move, so there's no parallax error when stitching single frames, but the magnification must have changed. I assume you stitched last so it's surprising you got smearing. I had that issue once with a similar shot but the floating flower parts in the middle got smeared while scaling was performed to match the background petals. I finally concluded a longer lens was the solution to such a deep shot, although I still held out hope that holding the lens/pupil stationary would have fixed it but apparently not. This setup should work for you with a flat subject though?

I was assuming you were modifying your tilt adaptor to include shifting plus rise and fall Have you tried using the JML 21mm with tilts?

Yes, the magnification is definitely changing but that's true of any method of stacking. It changes more with fixed lens/move camera technique.

Actually, I think Zerene Stacker could handle stacking shifted frames if it could set an arbitrary center point that may be out of the frame.

[rjlittlefield]

There's a collection of features on the to-do list related to geometry that is known a priori, as opposed to having to be figured out entirely from the imagery. A more common example would be something like "1.0003X per step", which would be handy in dealing with a fair number of microscopy applications with thick subjects. I've added "off-screen center" to the list, just to make sure it doesn't get overlooked.

--Rik

[PaulFurman]

I was assuming you were modifying your tilt adaptor to include shifting plus rise and fall

I used to have those parts assembled for shift in any direction plus up & down tilt with the lens fixed and a bag bellows but I didn't try it with macro work. I doubt it was stable enough for high magnification and it didn't allow much extension. Now I've rearranged and moved all those parts up front to help orient the subject, using the PB4 bellows behind for stacking.

I was thinking of using the PB4 reversed for shifted panoramas but your results are not encouraging. I could achieve that by shifting the lens, then shifting the subject.

Have you tried using the JML 21mm with tilts?

Just preliminary shots checking the LCD zoomed in. It looks fine up until the most extreme limit of the PB4 which is a 30 degree angle. It's sort of rare that I'd actually have use for a tilted stack but maybe. That would be more useful for hand held work. I actually have the JML mounted with a rubber gasket thing that allows a little bit of tilt.

Yes, the magnification is definitely changing but that's true of any method of stacking. It changes more with fixed lens/move camera technique.

Oh yeah, moving the rear standard means more scaling so resolution loss there. Where's my free lunch?!

Actually, I think Zerene Stacker could handle stacking shifted frames if it could set an arbitrary center point that may be out of the frame.

I'm still surprised by these results. You must have had x,y alignment turned off? I don't think Zerene does perspective correction by scaling different parts of the frame like panorama software corrects perspective, does it? In my messed up flower part example, the software would have to scale different parts of each frame based on the distance of that part of the subject from the lens/sensor, which is asking an awful lot if even possible.

There is a slight chance of improvement if you first ran each level of your stacks through pano software, then stacked the resulting large frames but I don't think so.