I am wondering about stacked options for a 55mm diagonal sensor.
How do the coverages of the individual lenses and the interactions between them determine the coverage at the sensor plane?
Coverage and Stacked Lenses - Constraints?
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 115
- Joined: Tue Oct 18, 2011 6:57 am
One criterion is that the designed image circle of the reversed lens, times the magnification of the stacked combo, should not be less than the diagonal of the camera sensor. So if a lens designed for MFT is reversed on a FF lens, the ratio of the "tube lens" focal length to the reversed lens focal length must be greater than about 2. Otherwise the good image circle will not fill the sensor.
But I think the ratio should not be MUCH greater than 2 in this case, because that would mean part of the image captured by the reversed lens would be wasted.
But I think the ratio should not be MUCH greater than 2 in this case, because that would mean part of the image captured by the reversed lens would be wasted.
-
- Posts: 115
- Joined: Tue Oct 18, 2011 6:57 am
Thanks, Lou.
Is it true that if the lenses are close together, the forward-facing lens covers at infinity, and the reversed lens covers at the resultant magnification, that the combination will cover, or is there a potential onion in the ointment?
I fear there is a gap between thin lens theory and the realtiies of pupils.
John
Is it true that if the lenses are close together, the forward-facing lens covers at infinity, and the reversed lens covers at the resultant magnification, that the combination will cover, or is there a potential onion in the ointment?
I fear there is a gap between thin lens theory and the realtiies of pupils.
John
Yes, I think it is always possible to get vignetting or lack of coverage. The condition I gave was a necessary but not sufficient condition.I can tell you for sure that you won't get full coverage if the condition I mentioned is not met. But satisfying that condition is not enough to guarantee coverage.
We often see effects of entrance pupil location causing vignetting, especially in zooms.
We often see effects of entrance pupil location causing vignetting, especially in zooms.
I suspect it goes beyond pupils. Regardless of format, the forward facing (rear) lens has to look through the reversed (front) lens/close up lens/microscope objective. As I currently understand it, this means the entrance pupil of the reversed lens (which, since it's reversed, is serving as its "exit" pupil) has to fill the forward facing lens's field of view for vignetting to be avoided. I've not come across posts from Rik or others with deep optical backgrounds which state this specifically, but it seems to me the default guidance to keep the two coupled lenses close together derives from this condition being more easily satisfied the closer the reversed entrance pupil is to the front principal plane of the forward facing lens.JohnDownie wrote:I fear there is a gap between thin lens theory and the realties of pupils.
My experience with u43 makes me inclined to start from the assumption any stacked candidate being considered would vignette and then look for reasons why that might not be the case. One approach, which is optically naive but seems to have baseline predictive skill, is just to treat the reversed lens like a filter and anticipate mechanical vignetting will occur if its entrance pupil diameter is less than the objective diameter of whatever forward facing telephoto it's mounted on. A testable prediction from this is vignetting decreases with wider reversed lens apertures. This holds for my EL-Nikkor 50mm f/2.8 N from f/16 to f/2.8. I may eventually purchase a similarly designed 50mm f/1.7 or 1.8 (6/4 formula, more or less symmetrical) to confirm increasing pupil diameter from roughly 17mm to 28mm further reduces vignetting. If money, size, and weight aren't constrained one could extend this to f/1.2 lenses or the f/0.95 Nocticrons.
In zooms, my understanding is vignette potential is generally higher because angle of view increases and the front principal plane tends to shift towards the sensor at the wide end of the zoom range, both changes which act to increase the zoom's field of view at the reversed entrance pupil. Within my small sample, zooms where change in vignetting is pronounced are wider range ones with shorter focal lengths at their wide end. Zooms with narrower ranges and longer focal lengths show less variation in vignetting. It seems to me this may be generally expected behavior but, not being a lens collector, I lack the data to make a stronger statement.
As an empirical observation, the one prime I have casts the exit pupil of the infinity corrected objective I have to the sensor at essentially unchanged size. The same holds for the wide end of the zooms I have and, with increasing focal length, pupil enlargement and reduction in vignetting proceed at about half the rate of increase in subject magnification. This suggests to me a sufficient predictor of sensor coverage would consider beam forming by the forward facing lens as well as subject coverage and pupils. However, I'm not aware of any such model having been proposed.
- rjlittlefield
- Site Admin
- Posts: 23602
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
I've never had much success understanding vignetting by starting with the pupils. Instead, I find it simpler to think about what happens to the light as it travels from subject to sensor. Then after I have that model clearly in mind, sometimes I can relate that understanding back to the pupils.
Following that approach...
First, pick some point on the subject. A bunch of light rays leave that point, headed in the general direction of the camera. Some of those rays enter the front lens and get refracted in such a way that they pass through that lens's limiting aperture, get refracted some more, and exit the front lens still heading in the general direction of the camera. Some (not necessarily all) of those rays then enter the second lens, where the process repeats with the second lens's aperture. Light rays that successfully traverse both lenses, passing through both apertures, are the ones that combine to create the image captured by the sensor.
In the ideal case there is exactly one limiting aperture in the entire chain of lenses. This might be in the front lens, or it might be in the rear lens, or it might be added between the two lenses as an iris or Waterhouse stop. In any case, when there is only one limiting aperture, all points on the sensor see the same effective aperture and there is no vignetting.
However, it is common that for off-center image points, two apertures will "fight with each other". That is, the front aperture will block some rays that otherwise would have made it through the rear aperture, and for the very same image point the rear aperture will block some rays that did make it through the front aperture. Where both apertures are in play like this, the resulting effective aperture as seen by the sensor becomes smaller and the image becomes dimmer. This is vignetting.
Now, relating this back to the pupils... As you know, the exit pupil of the front lens is the apparent position of the front lens's limiting aperture, as seen from the back of the lens. Likewise, the "entrance pupil" of the second lens is the apparent position of the second lens's limiting aperture, as seen from the front of that lens. The only time we can really say that one of these pupils fits inside the other is when they both occur in the same plane along the optical axis, and that almost never happens when you're stacking ordinary lenses. So in practice, we're always really talking about a beam of light fitting inside a pupil, not one pupil fitting inside another.
Of course the beams depend on the angles of light rays as well as locations of the pupils, so in the end everything is connected to everything. I don't know any simple way to make accurate predictions in every case.
As a general rule, vignetting will be minimized or eliminated when either (A) one pupil is much larger than the other, or (B) the pupils are close together so the beams can't spread much between them.
Perhaps the discussion and illustrations here will help: http://www.photomacrography.net/forum/v ... 667#131667
--Rik
Following that approach...
First, pick some point on the subject. A bunch of light rays leave that point, headed in the general direction of the camera. Some of those rays enter the front lens and get refracted in such a way that they pass through that lens's limiting aperture, get refracted some more, and exit the front lens still heading in the general direction of the camera. Some (not necessarily all) of those rays then enter the second lens, where the process repeats with the second lens's aperture. Light rays that successfully traverse both lenses, passing through both apertures, are the ones that combine to create the image captured by the sensor.
In the ideal case there is exactly one limiting aperture in the entire chain of lenses. This might be in the front lens, or it might be in the rear lens, or it might be added between the two lenses as an iris or Waterhouse stop. In any case, when there is only one limiting aperture, all points on the sensor see the same effective aperture and there is no vignetting.
However, it is common that for off-center image points, two apertures will "fight with each other". That is, the front aperture will block some rays that otherwise would have made it through the rear aperture, and for the very same image point the rear aperture will block some rays that did make it through the front aperture. Where both apertures are in play like this, the resulting effective aperture as seen by the sensor becomes smaller and the image becomes dimmer. This is vignetting.
Now, relating this back to the pupils... As you know, the exit pupil of the front lens is the apparent position of the front lens's limiting aperture, as seen from the back of the lens. Likewise, the "entrance pupil" of the second lens is the apparent position of the second lens's limiting aperture, as seen from the front of that lens. The only time we can really say that one of these pupils fits inside the other is when they both occur in the same plane along the optical axis, and that almost never happens when you're stacking ordinary lenses. So in practice, we're always really talking about a beam of light fitting inside a pupil, not one pupil fitting inside another.
Of course the beams depend on the angles of light rays as well as locations of the pupils, so in the end everything is connected to everything. I don't know any simple way to make accurate predictions in every case.
As a general rule, vignetting will be minimized or eliminated when either (A) one pupil is much larger than the other, or (B) the pupils are close together so the beams can't spread much between them.
Perhaps the discussion and illustrations here will help: http://www.photomacrography.net/forum/v ... 667#131667
--Rik
Hi Rik, thanks for the link. It's probably no surprise my anecdotal experience with vignetting is consistent with your figures there. I guess my previous post can be more compactly stated as one way of getting around far to the rear element placement is to fill the front element of the forward facing (rear) lens in the stacked pair with light. A a method, it's increasingly impractical beyond the roughly 35mm diameter pupils of the Raynox DCR-250 and 150 and 50mm f/1.4s. I'm not sure where that gets the OP with respect to covering a 55mm diagonal but it does provide some indication of the amount of beam expansion required compared to APS-C or u43.
55mm does make me think Fuji GFX, where Fuji's cross sections suggest the 110 f/2, 120 f/4, and 100-200 f/5.6 adhere to the trend towards far rear elements. The 250 f/4 does accommodate the 1.4x teleconverter so perhaps that would be a native mount starting point, either directly or with TC's beam expansion. Considering the 250's price presumably other lenses would be financial options. If so, Nathan's PhaseOne data and finding no camera lens did well due to vignetting may be relevant.
55mm does make me think Fuji GFX, where Fuji's cross sections suggest the 110 f/2, 120 f/4, and 100-200 f/5.6 adhere to the trend towards far rear elements. The 250 f/4 does accommodate the 1.4x teleconverter so perhaps that would be a native mount starting point, either directly or with TC's beam expansion. Considering the 250's price presumably other lenses would be financial options. If so, Nathan's PhaseOne data and finding no camera lens did well due to vignetting may be relevant.