The Mysterious El-Zoom Nikkor
Moderators: rjlittlefield, ChrisR, Chris S., Pau
Re: The Mysterious El-Zoom Nikkor
After too many adjustments to count, these measurements are the closest to parfocal I could get with the lens in its current configuration. I suspect im not even close. Maybe everything is backwards I dont know. Any less extension and the working distance instantly ventures into what seems to be a small telescope, or any less working distance (tried to get close to what the 150 f2.8 has) and the required extension goes wild.
Both of these images were at a working distance of 425mm. Lowest magnification was .28x with 185mm extension. Highest magnification was .80x with only 110mm extension. Neither in true focus.
From the first image, here is a 400% corner crop w/ nearest neighbor resampling of a piece of hair from my dog. Just put in an order for a bunch of Thorlab's 95mm rail components in hopes to better test this thing.
Both of these images were at a working distance of 425mm. Lowest magnification was .28x with 185mm extension. Highest magnification was .80x with only 110mm extension. Neither in true focus.
From the first image, here is a 400% corner crop w/ nearest neighbor resampling of a piece of hair from my dog. Just put in an order for a bunch of Thorlab's 95mm rail components in hopes to better test this thing.
Re: The Mysterious El-Zoom Nikkor
Update: I have managed to obtain parfocality. I dont remember which method I used at this point, but I can confirm optically it works just like a normal stereo microscope. Minimum magnification is when the 2 elements come together towards the center of the lens & outwards for maximum magnification. Extension was roughly 108mm, and working distance was 23-24 inches. I had to laugh at that; the original way I had the lens mounted had a maximum working distance of 20 inches, so it took a whole new setup just to find focus 3 inches away. I also noted this: at maximum zoom --- smallest working distance (~10mm) was with an extension of 23 inches, and shortest feasible extension of ~1 inch(camera grip got in the way of the massive flange) yielded a working distance of a whopping 45 inches. I couldn't reach either of the extremes set at minimum zoom. I had the camera at 10 feet & still couldn't find anything. Photos below are of a vintage combination square I had laying around.
Minimum Zoom 100% crop Maximum Zoom 100% crop The maximum zoom photos are slightly out of focus, I still need to fine tune camera distance etc. But I wanted to show it is indeed parfocal so I didn't move the stage between the two sets of photos. Any glare or halos are almost certainly a result of the "lens tube" being cheap black felt. Magnifications are approximately 0.216x & 0.473x in the parfocal setup.
Minimum Zoom 100% crop Maximum Zoom 100% crop The maximum zoom photos are slightly out of focus, I still need to fine tune camera distance etc. But I wanted to show it is indeed parfocal so I didn't move the stage between the two sets of photos. Any glare or halos are almost certainly a result of the "lens tube" being cheap black felt. Magnifications are approximately 0.216x & 0.473x in the parfocal setup.
Last edited by J_Rogers on Sat Sep 04, 2021 3:10 pm, edited 1 time in total.
Re: The Mysterious El-Zoom Nikkor
-Here are some more pictures for those curious of how I have everything setup just for this one lens. I attempted to take some really nice "product" photos but ended up going with some quarter effort work. Stand is currently 6 feet tall, total combined weight is well north of 100 pounds. I have to give some respect to all those commercial product photography guys because trying to take a professional picture of something this big and heavy makes all those food bloggers look like they are playing a children's game.
-I scrapped the Thorlabs 95mm rails after getting them. Way too big for what I wanted to use them for. Also the cost / scalability is rather steep. Have inquired about a custom cut surfacing plate as a more permanent base in lieu of the platform currently being used. $500 plus freight. If I keep this itll be what it mounts on top of.
-Other than that, enjoy my messy workshop and failed photoshoot.
-I scrapped the Thorlabs 95mm rails after getting them. Way too big for what I wanted to use them for. Also the cost / scalability is rather steep. Have inquired about a custom cut surfacing plate as a more permanent base in lieu of the platform currently being used. $500 plus freight. If I keep this itll be what it mounts on top of.
-Other than that, enjoy my messy workshop and failed photoshoot.
-
- Posts: 1706
- Joined: Sat Apr 14, 2018 10:26 am
Re: The Mysterious El-Zoom Nikkor
That's quite a monster! I'll admit I'm a bit surprised that it only has about a 2:1 zoom ratio given all the optical work they obviously put into it, but that should be a useful mag range.
Re: The Mysterious El-Zoom Nikkor
thanks for the update with all the pictures, I very much enjoy seeing the journey with this exotic lens
so if I got this right and you're still using a D3300 that would be 15mm sensor height, or roughly 0.2x and 0.5x.
that would make sense if the lens was used in a scanner with a 12mm high sensor, since the object area would cover 135 and medium format film.
I'm probably missing something here, your pictures show a ruler with roughly 7cm image hight on min and 3cm on max mag.
so if I got this right and you're still using a D3300 that would be 15mm sensor height, or roughly 0.2x and 0.5x.
that would make sense if the lens was used in a scanner with a 12mm high sensor, since the object area would cover 135 and medium format film.
chris
Re: The Mysterious El-Zoom Nikkor
You are correct. It goes to show what a 70+ hour work week will do. I thought I could take the easy way out and measure 1 millimeter in photoshop and cross multiply it by sensor height divided by total pixels. After double checking it, this time the long way, I got 0.216x & 0.473x. Used photoshop to draw out a scale bar for the clipped sections of the ruler. Give or take 72mm and 33mm respectively.chris_ma wrote: ↑Sat Sep 04, 2021 12:38 amthanks for the update with all the pictures, I very much enjoy seeing the journey with this exotic lens
I'm probably missing something here, your pictures show a ruler with roughly 7cm image hight on min and 3cm on max mag.
so if I got this right and you're still using a D3300 that would be 15mm sensor height, or roughly 0.2x and 0.5x.
that would make sense if the lens was used in a scanner with a 12mm high sensor, since the object area would cover 135 and medium format film.
Re: The Mysterious El-Zoom Nikkor
Decided to test the telecentricity of the lens this past week. I am ultimately curious of the accuracy of my "tests" as I recall reading a post on here from Rik mentioning higher magnifications allow for more variation in the angle vs lower magnifications (on testing telecentricity.) And if I have interpreted that correctly, that was with respect to the error in perfectly perpendicular axis between test target, lens, and camera. Without a true dedicated mount for my camera to couple to the lens correctly, it is still just hanging above the lens as seen in earlier photos. Given the lower magnifications that the El-Zoom is parfocal at, I can imagine there is a rather limited amount of accuracy you can get with its current camera mounting configuration.
Another interesting situation arose when adjusting the focus of the image for testing. Either the field of view is (too large?) for a small crop sensor, or the lens is completley out-resolving my camera. From "perfect focus" it takes a step size of 250 microns in either direction, up or down, until there is any noticeable difference in image. I'm not talking about a live view preview either. There is about 500 microns where you can take any raw image into photoshop, zoom in 1600% or 3200% and search the entire image, of a flat target, for a change in pixels. Other than camera noise there is no change between any image whatsoever. At 25 micron limits of either side of that 500 micron range the only noticeable change is the introduction of the faintest chromatic aberrations in -A- corner of the image, again seen only looking at pixels at 1000%+ zoom in photoshop. So that is how I decided what was "perfect focus" and what wasn't. Split the difference. And then use 250 micron steps in either direction. I am inclined to say it is out-resolving due to the fact it only took 10 steps at 250 microns in either direction past the perfect focus limit to reach an image that was entirely out-of-focus.
As to the test itself: aperture was wide open, lens was set on maximum zoom. Glued down a sheet of this graph paper on a piece of glass and positioned it completley flat. From the center of one grid line to the next is 2.5mm. Used an image in the perfect focus range as the index image and then stepped down from 500 microns and up from 0 microns in 250 micron steps. 10 steps in either direction. So a total of 21 images.
I then took each image and cropped 2600 x 1800 pixel sections, bottom left -> bottom right -> top left -> top right -> center for comparisons. In Zerene used the index image and set each frame to be scaled independently to the index.
Here is the target and outlines of each place that i cropped for a test Scale as reported by Zerene. Lowest frame number corresponds to furthest focus, and highest to closest focus Change in scale factor. I omitted the jump in the middle of the scale numbers to make it a little easier to view the numbers Each 2.5mm square in the images is 330 pixels x 330 pixels. Which comes out to 0.0075757576mm/pixel.
Multiplying the 2600 (pixels) by the average change in scale of each crop region yields (same order as labeled in the charts):
------>1.0917197298347514
------>0.9747199562157579
------>1.0914473725812892
------>1.0401310665191175
------>1.1124955855413747
Divide each by two then multiply by mm/pixel number of 0.0075757576 cancels the pixels and should give roughly the change in image size in mm. This is Delta H
------>0.0041353020
------>0.0036921211
------>0.0041342704
------>0.0039398904
------>0.0042139984
Step size is 250 microns = 0.25mm. This is Delta S
Arctan(theta) = Delta H / Delta S ==> gives Telecentricity in degrees.
------>0.9476549830 BR
------>0.8461103150 BL
------>0.9474186223 TR
------>0.9028816239 TL
------>0.9656858414 C
- Someone can check the math. But I guess at the moment the conclusion with those degrees, would be "somewhat" telecentric?
Another interesting situation arose when adjusting the focus of the image for testing. Either the field of view is (too large?) for a small crop sensor, or the lens is completley out-resolving my camera. From "perfect focus" it takes a step size of 250 microns in either direction, up or down, until there is any noticeable difference in image. I'm not talking about a live view preview either. There is about 500 microns where you can take any raw image into photoshop, zoom in 1600% or 3200% and search the entire image, of a flat target, for a change in pixels. Other than camera noise there is no change between any image whatsoever. At 25 micron limits of either side of that 500 micron range the only noticeable change is the introduction of the faintest chromatic aberrations in -A- corner of the image, again seen only looking at pixels at 1000%+ zoom in photoshop. So that is how I decided what was "perfect focus" and what wasn't. Split the difference. And then use 250 micron steps in either direction. I am inclined to say it is out-resolving due to the fact it only took 10 steps at 250 microns in either direction past the perfect focus limit to reach an image that was entirely out-of-focus.
As to the test itself: aperture was wide open, lens was set on maximum zoom. Glued down a sheet of this graph paper on a piece of glass and positioned it completley flat. From the center of one grid line to the next is 2.5mm. Used an image in the perfect focus range as the index image and then stepped down from 500 microns and up from 0 microns in 250 micron steps. 10 steps in either direction. So a total of 21 images.
I then took each image and cropped 2600 x 1800 pixel sections, bottom left -> bottom right -> top left -> top right -> center for comparisons. In Zerene used the index image and set each frame to be scaled independently to the index.
Here is the target and outlines of each place that i cropped for a test Scale as reported by Zerene. Lowest frame number corresponds to furthest focus, and highest to closest focus Change in scale factor. I omitted the jump in the middle of the scale numbers to make it a little easier to view the numbers Each 2.5mm square in the images is 330 pixels x 330 pixels. Which comes out to 0.0075757576mm/pixel.
Multiplying the 2600 (pixels) by the average change in scale of each crop region yields (same order as labeled in the charts):
------>1.0917197298347514
------>0.9747199562157579
------>1.0914473725812892
------>1.0401310665191175
------>1.1124955855413747
Divide each by two then multiply by mm/pixel number of 0.0075757576 cancels the pixels and should give roughly the change in image size in mm. This is Delta H
------>0.0041353020
------>0.0036921211
------>0.0041342704
------>0.0039398904
------>0.0042139984
Step size is 250 microns = 0.25mm. This is Delta S
Arctan(theta) = Delta H / Delta S ==> gives Telecentricity in degrees.
------>0.9476549830 BR
------>0.8461103150 BL
------>0.9474186223 TR
------>0.9028816239 TL
------>0.9656858414 C
- Someone can check the math. But I guess at the moment the conclusion with those degrees, would be "somewhat" telecentric?
- rjlittlefield
- Site Admin
- Posts: 24009
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: The Mysterious El-Zoom Nikkor
I can't tell exactly how you're thinking about telecentricity.
So instead of checking your math, let me start by doing some of my own, then bring the two back together.
As I understand your test, what you've done is to hold the lens-to-camera distance fixed, while varying the lens-to-subject distance by steps of 250 microns (0.25 mm).
Then, Zerene Stacker's alignment algorithm computes a scale change that is roughly 0.00044 per step (measuring from midpoints of the clusters at +-2500 microns).
You've mentioned that the effective DOF of the lens is at least 500 microns, so the scale change per DOF is at least 0.00088. Assuming an image width of 6000 pixels, that would mean a change in image width of almost 5 pixels per DOF. That would produce unacceptable smearing in a stacked result, if we did not allow Scale to be corrected in alignment.
So, for purposes of stacking, the lens definitely cannot be treated as telecentric.
Estimating the entrance pupil location gives another way of looking at this issue. The entrance pupil location can be estimated as being stepSize/deltaScale distance away from the subject. Plugging in the numbers, 0.25mm/0.00044 = 570 mm from subject to entrance pupil.
Enrico Savazzi's page lists the maximum focal length as 230 mm. I assume this is the setting at which you've observed maximum parfocal magnification of 0.473x (reported in the previous post). These two numbers, if correct, give us the location of the front principal plane, which would be located at 230*(1+1/0.473) = 716 mm from subject to front principal plane.
This is another indication that the lens is not telecentric. With a truly telecentric lens, the entrance pupil is located at infinity. With a lens that is nearly telecentric, the entrance pupil is located closer than infinity, but still much farther away than you would expect based on its focal length and working magnification. With this lens, that relationship is reversed -- the entrance pupil is even closer than predicted by focal length and working magnification.
But you've written "somewhat" telecentric, and I gather that this is based on an average value of roughly 0.92 degrees that you've calculated. Let me look at that from a couple of different standpoints.
For reference, let's use https://www.edmundoptics.com/knowledge- ... ification/ , which says that
First, I note that Edmund's page contains "Figure 3: A telecentricity plot for a typical telecentric lens", which shows values smaller than 0.1 degrees. That's almost 10X smaller than the number that you've computed.
Second, I think the number that you've computed is conceptually different from the number that Edmund is talking about. What you've really computed is a difference of angles, between the center of each crop and the edge of that same crop. That's the same as Edmund's angle only for the crop that is centered on the optical axis, and only at the edge of that crop.
Noting that the width of your crops is about 43% of the total image width, the angle at the edges of the total width would be 1/0.43 times larger, giving about 2.1 degrees at the edge.
That same number could be computed a different way, based on subject size and distance to the entrance pupil. Using 2.5 mm per square on the graph paper, the subject size looks to be about 45.2 mm total width. That would give a half-width about 22.6 mm, which combined with the estimated 570 mm to entrance pupil, calculates to be an angle around 2.3 degrees (=DEGREES(ATAN(22.6/570))). (I expect the difference between 2.1 and 2.3 degrees is a matter of accumulated rounding and approximate inputs.)
I hope this trip through Calculation Land is of some help. Let me know if I've messed something up.
But anyway, I don't see this lens as being telecentric in any sense. It's just a long lens, used with a small sensor.
--Rik
So instead of checking your math, let me start by doing some of my own, then bring the two back together.
As I understand your test, what you've done is to hold the lens-to-camera distance fixed, while varying the lens-to-subject distance by steps of 250 microns (0.25 mm).
Then, Zerene Stacker's alignment algorithm computes a scale change that is roughly 0.00044 per step (measuring from midpoints of the clusters at +-2500 microns).
You've mentioned that the effective DOF of the lens is at least 500 microns, so the scale change per DOF is at least 0.00088. Assuming an image width of 6000 pixels, that would mean a change in image width of almost 5 pixels per DOF. That would produce unacceptable smearing in a stacked result, if we did not allow Scale to be corrected in alignment.
So, for purposes of stacking, the lens definitely cannot be treated as telecentric.
Estimating the entrance pupil location gives another way of looking at this issue. The entrance pupil location can be estimated as being stepSize/deltaScale distance away from the subject. Plugging in the numbers, 0.25mm/0.00044 = 570 mm from subject to entrance pupil.
Enrico Savazzi's page lists the maximum focal length as 230 mm. I assume this is the setting at which you've observed maximum parfocal magnification of 0.473x (reported in the previous post). These two numbers, if correct, give us the location of the front principal plane, which would be located at 230*(1+1/0.473) = 716 mm from subject to front principal plane.
This is another indication that the lens is not telecentric. With a truly telecentric lens, the entrance pupil is located at infinity. With a lens that is nearly telecentric, the entrance pupil is located closer than infinity, but still much farther away than you would expect based on its focal length and working magnification. With this lens, that relationship is reversed -- the entrance pupil is even closer than predicted by focal length and working magnification.
But you've written "somewhat" telecentric, and I gather that this is based on an average value of roughly 0.92 degrees that you've calculated. Let me look at that from a couple of different standpoints.
For reference, let's use https://www.edmundoptics.com/knowledge- ... ification/ , which says that
Those words are a bit ambiguous, but I think they're referring to the angle between the optical axis and a ray drawn through the edge of the field. (The total angle of view would be twice that, counting both sides of the optical axis.)The other relevant specification is telecentricity, which is generally specified in degrees, and can be thought of as the residual angular field of view of the lens.
First, I note that Edmund's page contains "Figure 3: A telecentricity plot for a typical telecentric lens", which shows values smaller than 0.1 degrees. That's almost 10X smaller than the number that you've computed.
Second, I think the number that you've computed is conceptually different from the number that Edmund is talking about. What you've really computed is a difference of angles, between the center of each crop and the edge of that same crop. That's the same as Edmund's angle only for the crop that is centered on the optical axis, and only at the edge of that crop.
Noting that the width of your crops is about 43% of the total image width, the angle at the edges of the total width would be 1/0.43 times larger, giving about 2.1 degrees at the edge.
That same number could be computed a different way, based on subject size and distance to the entrance pupil. Using 2.5 mm per square on the graph paper, the subject size looks to be about 45.2 mm total width. That would give a half-width about 22.6 mm, which combined with the estimated 570 mm to entrance pupil, calculates to be an angle around 2.3 degrees (=DEGREES(ATAN(22.6/570))). (I expect the difference between 2.1 and 2.3 degrees is a matter of accumulated rounding and approximate inputs.)
I hope this trip through Calculation Land is of some help. Let me know if I've messed something up.
But anyway, I don't see this lens as being telecentric in any sense. It's just a long lens, used with a small sensor.
--Rik
-
- Posts: 1706
- Joined: Sat Apr 14, 2018 10:26 am
Re: The Mysterious El-Zoom Nikkor
This lens would have been used exclusively on very flat subjects, right? I don't think telecentricity would have been a useful feature for it in that application.
Re: The Mysterious El-Zoom Nikkor
Rik, I genuinely appreciate the effort and time you took to explain all that. I tend to dive into things over my head (in this case, optics) and then learn from the errors I make, so a post like that is truly invaluable.
Most of what I had posted was due to a lack of understanding of what differentiates an excellent telecentric lens from one that comes close, either from lens design or some other constraint. Hence my incorrect classification. After testing everything, I knew the scale change graph wasn't like the one seen in your post On The Sensitivity of Zerene Stacker Scale Measurements. I also thought the change in scale parts per/x was a little too large compared with others' posted data. My initial impression was the lens fell somewhere between a long lens at low magnifications or a poor-performing telecentric lens. One would essentially be used in the same fashion for stacking and stitching while not being telecentric. The other would only be something like an object-side only optical design, yielding distortion some order of magnitude higher than bi-telecentric but still considerably less than a regular lens ( ill try to find that research paper.) The ~2.1-2.3 degrees undeniably makes this "not telecentric."
The 0.92 degrees came from a diagram from this site https://www.opto-e.com/resources/telece ... s-tutorial , the section between magnification constancy and low distortion. I now see what you mean by 1/0.43 times the calculated angle since the formula is for the center of the optical axis. In retrospect, there should have only been 3 sections left/right/center allowing for half of each frame & center. Yielding slightly different scale variations but, still not telecentric.
I'll leave all my mistakes in that post above, hopefully allowing someone down the road to gain a better understanding.
Guess the only thing left to figure out is aperture. Is the technique of imaging the aperture (entrance pupil as seen from the front of the lens) from a distance sufficiently far away and estimating size based on a known dimension the easiest way to go? Or is there something better? i.e., pixels in that image divided by measured lens diameter -> multiplied by the pixel dimensions of the entrance pupil.
But you've written "somewhat" telecentric, and I gather that this is based on an average value of roughly 0.92 degrees that you've calculated. Let me look at that from a couple of different standpoints.
Most of what I had posted was due to a lack of understanding of what differentiates an excellent telecentric lens from one that comes close, either from lens design or some other constraint. Hence my incorrect classification. After testing everything, I knew the scale change graph wasn't like the one seen in your post On The Sensitivity of Zerene Stacker Scale Measurements. I also thought the change in scale parts per/x was a little too large compared with others' posted data. My initial impression was the lens fell somewhere between a long lens at low magnifications or a poor-performing telecentric lens. One would essentially be used in the same fashion for stacking and stitching while not being telecentric. The other would only be something like an object-side only optical design, yielding distortion some order of magnitude higher than bi-telecentric but still considerably less than a regular lens ( ill try to find that research paper.) The ~2.1-2.3 degrees undeniably makes this "not telecentric."
The 0.92 degrees came from a diagram from this site https://www.opto-e.com/resources/telece ... s-tutorial , the section between magnification constancy and low distortion. I now see what you mean by 1/0.43 times the calculated angle since the formula is for the center of the optical axis. In retrospect, there should have only been 3 sections left/right/center allowing for half of each frame & center. Yielding slightly different scale variations but, still not telecentric.
I'll leave all my mistakes in that post above, hopefully allowing someone down the road to gain a better understanding.
Guess the only thing left to figure out is aperture. Is the technique of imaging the aperture (entrance pupil as seen from the front of the lens) from a distance sufficiently far away and estimating size based on a known dimension the easiest way to go? Or is there something better? i.e., pixels in that image divided by measured lens diameter -> multiplied by the pixel dimensions of the entrance pupil.
- rjlittlefield
- Site Admin
- Posts: 24009
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: The Mysterious El-Zoom Nikkor
I suggest to use the method described at http://www.photomacrography.net/forum/v ... 516#p61516 , in the "FAQ: What is "pupil ratio" and why would I care?" :J_Rogers wrote: ↑Sun Sep 19, 2021 2:27 pmGuess the only thing left to figure out is aperture. Is the technique of imaging the aperture (entrance pupil as seen from the front of the lens) from a distance sufficiently far away and estimating size based on a known dimension the easiest way to go? Or is there something better? i.e., pixels in that image divided by measured lens diameter -> multiplied by the pixel dimensions of the entrance pupil.
So then, three pictures: the entrance pupil, the exit pupil, and a ruler, all shot at the same scale, so as to get pupil diameters in mm. That should give us everything that's needed to make an accurate calculation of f-stop in the working configuration.... with most lenses, an even better way to take the measurement is use a shallow DOF macro setup. Focus on one pupil at any convenient distance and magnification, and take a picture. Then without changing the camera's focus, reverse the lens being tested, move it back and forth until the other pupil is in sharp focus, and take a second picture. Because both pictures were taken at the same scale, you can calculate pupil ratio using pixel counts without even needing to know what the measurement in mm is. Of course if you care about the measurement in mm, say to verify f-number, then you can photograph a ruler at the same scale also.
Normally the procedure is simple to do by placing the lens in horizontal orientation. But I recall from Enrico's description that springs inside the lens may make the EL Zoom Nikkor spontaneously reconfigure itself if tipped horizontally. Given that, I suggest to get a small ordinary mirror and place that under your lens, mirror tipped at 45 degrees so that you can measure the reflection of the entrance pupil.
--Rik
Re: The Mysterious El-Zoom Nikkor
Measured the pupil diameters.
Maximum Zoom
Entrance pupil
-->fully stopped down: 11.855mm
-->wide open: 31.252mm
Exit pupil
-->fully stopped down: 08.013mm
-->wide open: 21.138mm
Minimum Zoom
Entrance pupil
-->fully stopped down: 04.299mm
-->wide open: 11.034mm
Exit pupil
-->fully stopped down: 08.981mm
-->wide open: 23.099mm
f/# = focal length / entrance pupil diameter
Yields: 230mm --> f/7.35 - f/19.40 & 99mm --> f/8.97 - f/23.08
Pupil magnification = Diameter (exit pupil) / Diameter (entrance pupil)
Averaging the final number from wide open calculation with fully stopped down calculation yields: 0.676 for maximum zoom and 2.091 for minimum zoom.
Maximum Zoom
Entrance pupil
-->fully stopped down: 11.855mm
-->wide open: 31.252mm
Exit pupil
-->fully stopped down: 08.013mm
-->wide open: 21.138mm
Minimum Zoom
Entrance pupil
-->fully stopped down: 04.299mm
-->wide open: 11.034mm
Exit pupil
-->fully stopped down: 08.981mm
-->wide open: 23.099mm
f/# = focal length / entrance pupil diameter
Yields: 230mm --> f/7.35 - f/19.40 & 99mm --> f/8.97 - f/23.08
Pupil magnification = Diameter (exit pupil) / Diameter (entrance pupil)
Averaging the final number from wide open calculation with fully stopped down calculation yields: 0.676 for maximum zoom and 2.091 for minimum zoom.
- rjlittlefield
- Site Admin
- Posts: 24009
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: The Mysterious El-Zoom Nikkor
Thanks for the numbers.
So then, using the formulas at "FAQ: What is "pupil ratio" and why would I care?", and plugging in the magnifications 0.473 and 0.216, I get effective (working) F-numbers, wide open, of f_e = 12.51 at maximum zoom, and f_e = 9.90 at minimum zoom.
Another way of looking at the lens spec is to calculate what the nominal F-number of a lens with pupil ratio = 1 would have to be, to give those same working F-numbers. I don't know a standard name for that number, but let's call it the "acts like" F-number, f_actsLike.
We know that for p=1, the formula is f_e = f_r * (m+1), so we can calculate f_actsLike = f_e/(m+1)
Plugging in the numbers, we get that, wide open, the EL-Zoom Nikkor is f_actsLike = 8.49 at maximum zoom, and f_actsLike = 8.14 at minimum zoom.
Sounds to me like calling that lens "F/8" is a pretty good label!
--Rik
So then, using the formulas at "FAQ: What is "pupil ratio" and why would I care?", and plugging in the magnifications 0.473 and 0.216, I get effective (working) F-numbers, wide open, of f_e = 12.51 at maximum zoom, and f_e = 9.90 at minimum zoom.
Another way of looking at the lens spec is to calculate what the nominal F-number of a lens with pupil ratio = 1 would have to be, to give those same working F-numbers. I don't know a standard name for that number, but let's call it the "acts like" F-number, f_actsLike.
We know that for p=1, the formula is f_e = f_r * (m+1), so we can calculate f_actsLike = f_e/(m+1)
Plugging in the numbers, we get that, wide open, the EL-Zoom Nikkor is f_actsLike = 8.49 at maximum zoom, and f_actsLike = 8.14 at minimum zoom.
Sounds to me like calling that lens "F/8" is a pretty good label!
--Rik
- rjlittlefield
- Site Admin
- Posts: 24009
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: The Mysterious El-Zoom Nikkor
Now, going back to this observation:
So, it makes perfect sense that you would be able to go 0.25 mm either away from center without seeing any difference at all. That would only be 1/12 lambda wavefront error, which is not visually distinguishable from perfect focus. On the other hand, at 2.5 mm away from perfect focus, you'd be at 0.8 lambda error, which does indeed look entirely out-of-focus. If you visit my thread at viewtopic.php?f=8&t=23751&p=146835, download the .zip that it links to, and just drag the .zip into Zerene Stacker, you can look at image degradation for various wavefront errors in that vicinity.
At this point, I think you have the lens pretty well characterized.
Effective f/12.5 is what Zerene Stacker's DOF calculator considers to be "near optimum" for your camera's sensor, by which it means that the diffraction-limited DOF is nearly the same as a classic circle-of-confusion DOF with COC = 3 pixels wide.
But no, at effective f/12.5 the lens is not significantly out-resolving your D3300 at 6000 pixels in 23.5 mm. Nyquist cutoff for the sensor is at 127.7 cycles/mm, and diffraction cutoff for the lens is only a little higher at 145.5 cycles/mm. This is very close to Nikon's rule that an "optimum" match has those two values equal.
--Rik
Plugging the magnification and calculated effective aperture into the standard equations, I get 1/4 lambda wavefront error DOF = 1.539 mm at the subject.J_Rogers wrote: ↑Sat Sep 18, 2021 10:29 amAnother interesting situation arose when adjusting the focus of the image for testing. Either the field of view is (too large?) for a small crop sensor, or the lens is completley out-resolving my camera. From "perfect focus" it takes a step size of 250 microns in either direction, up or down, until there is any noticeable difference in image. I'm not talking about a live view preview either. There is about 500 microns where you can take any raw image into photoshop, zoom in 1600% or 3200% and search the entire image, of a flat target, for a change in pixels. Other than camera noise there is no change between any image whatsoever. At 25 micron limits of either side of that 500 micron range the only noticeable change is the introduction of the faintest chromatic aberrations in -A- corner of the image, again seen only looking at pixels at 1000%+ zoom in photoshop. So that is how I decided what was "perfect focus" and what wasn't. Split the difference. And then use 250 micron steps in either direction. I am inclined to say it is out-resolving due to the fact it only took 10 steps at 250 microns in either direction past the perfect focus limit to reach an image that was entirely out-of-focus.
As to the test itself: aperture was wide open, lens was set on maximum zoom.
So, it makes perfect sense that you would be able to go 0.25 mm either away from center without seeing any difference at all. That would only be 1/12 lambda wavefront error, which is not visually distinguishable from perfect focus. On the other hand, at 2.5 mm away from perfect focus, you'd be at 0.8 lambda error, which does indeed look entirely out-of-focus. If you visit my thread at viewtopic.php?f=8&t=23751&p=146835, download the .zip that it links to, and just drag the .zip into Zerene Stacker, you can look at image degradation for various wavefront errors in that vicinity.
At this point, I think you have the lens pretty well characterized.
Effective f/12.5 is what Zerene Stacker's DOF calculator considers to be "near optimum" for your camera's sensor, by which it means that the diffraction-limited DOF is nearly the same as a classic circle-of-confusion DOF with COC = 3 pixels wide.
But no, at effective f/12.5 the lens is not significantly out-resolving your D3300 at 6000 pixels in 23.5 mm. Nyquist cutoff for the sensor is at 127.7 cycles/mm, and diffraction cutoff for the lens is only a little higher at 145.5 cycles/mm. This is very close to Nikon's rule that an "optimum" match has those two values equal.
--Rik
Re: The Mysterious El-Zoom Nikkor
While not an official test target/image representing resolving power, these should be useful in putting some of the math above into perspective. The center fiducial lines are 200 microns thick. Both images are 100% crops. Light falloff is because I didn't have a large enough backlight.
You can see the larger uncropped images here: https://www.easyzoom.com/albumaccess/30 ... 22037aad55
Minimum Zoom - wide open aperture Maximum Zoom - wide open aperture
You can see the larger uncropped images here: https://www.easyzoom.com/albumaccess/30 ... 22037aad55
Minimum Zoom - wide open aperture Maximum Zoom - wide open aperture