artur5 wrote:
Sorry, but the tolerance to adapter thickness ins't related to the flange distance but to the focal lenght of the lens. An adapter skewed 0.01mm. will have the same negative effect on a Nikon 28mm. lens than on a Leica M 28mm. lens.
Floating point lenses are designed to place focus at a specific distance from the mount. They are very precise in this regard. With this type of lens, when you focus the lens, the floating element usually moves around to achieve or fine tune this. With non FLE lenses, the whole or rear group simply rack the focus plane back and forth. Consequently they are more tolerant than FLE lenses. It should be noted that some rear group focussing lenses also behave like FLE lenses.
Lastly, it behooves a photographer to shim their adapters, so as to achieve maximum lens performance. Zone focus will also work properly as well.
turnstyle wrote:
Hi all, I don't totally understand the shimming/filing thing.
Do you not have the same problem when mounting lenses to bodies that support them without adapter? I get that the adapter adds complexity -- but if lenses and bodies can be built to spec, such that it "just works" -- is there some reason to expect the adapter can't?
fwiw, I have the Voightlander adapter, and I'm trying to figure out if my side mushiness (Lux 50 on NEX7) is due to the adapter -- or if it's just what I should expect from the lens.
Thanks!
The problem lies with the system tolerances.
To achieve proper focus a system states that the lens must be seated at a specific distance from the sensor. There is usually and acceptable range, but closer to the nominal always results in optimum performance.
Let me see if I can summarize the effects of body/lens/adapter combinations.
Assume a system is originally designed for the following:
Flange distance: 50mm
Acceptable tolerance for the mount: +/- 0.01mm
Manufacturing technology: can produce +/- 0.005mm
So if we go to one end of the extreme we could have the following:
Camera mount: 50.005mm
Lens mount sits at +0.005mm (i.e. the exit pupil of the lens is too far out but still within tolerance)
This combination would result in the lens appearing at: 50.01mm, which is right at the edge of the system allowance.
Now we want to adopt a lens that ususally sits on a 60mm mount.
An adapter for this system would nominally be 10mm.
Let us say we have a lens that normally sits at the extrem of it's system as well.. so it would appear to be at 60.005mm.
If we can produce an adapter that is exactly 10.00mm thick, all would be good and this combination would result in 60.01mm. But we still have manufacturing tolerances to take in to account.
Even if we have a process that can guarantee 5X more precision we will still be off.
Our new hypothetical adapter comes out at 10.001mm.
If we add this combination up:
50.005mm + lens error 0.005mm + adapter thickness 10.001mm = 60.011mm. This is out of ideal focus by 0.011mm and out of spec by 0.001mm.
Even if we decided to make the adapter shorter, so 10mm at most and 9.998mm we will error on the other side:
Mount: 49.995 - lens error 0.005mm + 9.998 = 59.988 = 0.012mm error from ideal and 0.002mm out of spec.
This is why many manufactures create the adapter thinner than needed with the hope of shimming to proper thickness.
Hope this helps!
What is really needed is an easy way for users to adjust the adapter themselves for their camera.
phigment wrote:
What is really needed is an easy way for users to adjust the adapter themselves for their camera.
You mean like a helicoid adapter with a locking screw? Even if a photographer can adjust the adapter themselves, how would he know his result is good? Getting down to the micron precision is not easy with a typical shimming technique. The thinnest electrical tape is about 100 microns (4 mils). Typical aluminum foil thickness is about 10 to 20 microns. Common belief is that if the adapter is off by 10 microns, it would already show the defect.
I know your example is a theoretical one, but I hope there is a manufacturer out there would get the error down to 2-5 microns (in your case it's off by 1 micron). It might be unrealistic, however.
Andrew Gough wrote:
Floating point lenses are designed to place focus at a specific distance from the mount. They are very precise in this regard. With this type of lens, when you focus the lens, the floating element usually moves around to achieve or fine tune this. With non FLE lenses, the whole or rear group simply rack the focus plane back and forth. Consequently they are more tolerant than FLE lenses. It should be noted that some rear group focussing lenses also behave like FLE lenses.
Lastly, it behooves a photographer to shim their adapters, so as to achieve maximum lens performance. Zone focus will also work properly as well. ...Show more →
Yes, that's right, but this fact has nothing to do with the flange distance being shorter or longer. There're some M mount lenses with floating elements as well as there're in Nikon or Canon mounts.
Besides, I think that the sensibility of FLE designs to the absolute accuracy of distances to the focal plane is overstated.
For instance, many, if not all, modern macros have optical designs based in FLE. With those lenses, the nominal focal length is only real at infinity. As you say, when we turn the focus ring, some elements move inside to reduce the focal length, without changing the external physical dimensions of the lens. If I'm not mistaken, the Canon EF 100/2.8 macro goes down to 70mm. at the minimum focus distance of 31 cmt..
That said, most of these macros are often used with extension tubes or bellows to increase the magnification. It goes without saying that, in these cases, the distance from the glass elements to the focal plane changes a lot, not a hundred of a mm. like when we have a slightly defective adapter ring. In spite of this, the results obtained with a modern macro and bellows or tubes are often excellent.
I'm not saying that your statement isn't theoretically right, but I have yet to see a test where a lens with floating elements shows noticeable image degradation when used with an adapter ring slightly out of specifications ( not skewed, just a bit thinner or thicker than required )
The easiest thing would be to ensure the adapter is as close to ideal as possible. This way it won't make things worse than a native lens mounted on the camera. You should be able to measure this to some degree with calipers down to around 0.02mm.
But I guess this won't help counteract camera mount innaccuracies.
I would be happy with ensuring my lenses focus properly at infinity.
artur5 wrote:
Sorry, but the tolerance to adapter thickness ins't related to the flange distance
? Can you elaborate this? were you saying 0.1mm difference in flange distance will have the same effect on Leica M and Leica R as long as their focal length are the same? So, based on this, can I assume all the lens will have same design no matter what flange distance as long as focal length the same?
Anyway, My point is pretty clear on this: flange distance change is bad for lens designed/optimized for certain system. Further, the shorter the original flange distance, the bigger the impact on performance, for its relative error compare to original.
Flange distance change, or focal length change, or magnification change, or focus distance change, whatever you want equivalent this effect, the ending results is lens perform under different condition than it was designed for.
phigment wrote:
The easiest thing would be to ensure the adapter is as close to ideal as possible. This way it won't make things worse than a native lens mounted on the camera. You should be able to measure this to some degree with calipers down to around 0.02mm.
But I guess this won't help counteract camera mount innaccuracies.
I would be happy with ensuring my lenses focus properly at infinity.
I'm glad to hear this from an adapter maker. Do you have any plan to offer adapters for other mounts (Canon EF, Nikon F, etc.) to E-mount?
p.8 #10 · Sony A7r some full resolution test shots
zhangyue wrote: ? Can you elaborate this? were you saying 0.1mm difference in flange distance will have the same effect on Leica M and Leica R as long as their focal length are the same? So, based on this, can I assume all the lens will have same design no matter what flange distance as long as focal length the same?
well the R 90AA and m 90AA have the same design and very different flange distances.
artur's point is that if the lens design is the same the shorter flange difference won't make the lens perform worse due to adapter intolerances. in fact, basic trig will show that angular displacement close to the sensor (short adapter) is better than angular displacement far from sensor (long adapter) because the center of the image circle will be closer to the center of the sensor (assuming angular displacement is the same for both adapters). either way both adapters will have the same difference in focus between opposite corners.
zhangyue wrote:
Anyway, My point is pretty clear on this: flange distance change is bad for lens designed/optimized for certain system.
this point is fine.
zhangyue wrote:
Further, the shorter the original flange distance, the bigger the impact on performance, for its relative error compare to original.
this point does not seem obvious to me at all and will require a technical explanation for me to believe.
p.8 #11 · Sony A7r some full resolution test shots
phigment wrote:
I would be happy with ensuring my lenses focus properly at infinity.
Is there a recommended way to do that? Star? Trouble is, if the lens doesn't go past infinity, it seems hard to know if you're exactly where you want to be when you're at infinity, if you see what I mean? (eg, with the hard stop at infinity)
I gather you have a M to E-mount adapter? Cool beans! It passes the lens name, but not (I assume?) the aperture used?
p.8 #12 · Sony A7r some full resolution test shots
zhangyue wrote: ? Can you elaborate this? were you saying 0.1mm difference in flange distance will have the same effect on Leica M and Leica R as long as their focal length are the same? So, based on this, can I assume all the lens will have same design no matter what flange distance as long as focal length the same?
Anyway, My point is pretty clear on this: flange distance change is bad for lens designed/optimized for certain system. Further, the shorter the original flange distance, the bigger the impact on performance, for its relative error compare to original.
Flange distance change, or focal length change, or magnification change, or focus distance change, whatever you want equivalent this effect, the ending results is lens perform under different condition than it was designed for. ...Show more →
You can think of it this way, do you need to shift the glass on a DSLR 24mm lens more to focus it on 5m from infinity than a 24mm rangefinder lens? If not, then the focus error produced by uneven thickness of an adapter will be the same regardless of the flange distance.
p.8 #13 · Sony A7r some full resolution test shots
artur5 wrote:
Yes, that's right, but this fact has nothing to do with the flange distance being shorter or longer. There're some M mount lenses with floating elements as well as there're in Nikon or Canon mounts.
Besides, I think that the sensibility of FLE designs to the absolute accuracy of distances to the focal plane is overstated.
For instance, many, if not all, modern macros have optical designs based in FLE. With those lenses, the nominal focal length is only real at infinity. As you say, when we turn the focus ring, some elements move inside to reduce the focal length, without changing the external physical dimensions of the lens. If I'm not mistaken, the Canon EF 100/2.8 macro goes down to 70mm. at the minimum focus distance of 31 cmt..
That said, most of these macros are often used with extension tubes or bellows to increase the magnification. It goes without saying that, in these cases, the distance from the glass elements to the focal plane changes a lot, not a hundred of a mm. like when we have a slightly defective adapter ring. In spite of this, the results obtained with a modern macro and bellows or tubes are often excellent.
I'm not saying that your statement isn't theoretically right, but I have yet to see a test where a lens with floating elements shows noticeable image degradation when used with an adapter ring slightly out of specifications ( not skewed, just a bit thinner or thicker than required )...Show more →
I think that we are mixing terms here. When we say flange distance, I am including the adapter in that distance. If we are talking manufacturer flange distance to the native mount then it is upto the lens designer to achieve optimal performance. If we add the adapter into the equation, then we introduce a lot variability into the flange (exit pupil) to sensor distance. SInce this tolerance is very small for FLE lenses, it is easy to see how they could move outside of their optimum zone.
I have shot the Lux on various M cameras and NEX's, and from practical experience, I can tell you that the adapter makes a big difference. I think that Sony micro lens architecture is not optimized for RF lenses, and that is the primary reason for poor performance. The Lux is very good on the M. What is surprising, is the poor performance of the APO 90mm R lens. This lens is also a very good lens on many full frame DSLR's unto and including the D800.
Anyway, it was not my intention to drive this thread away from its initial intent. I will wait and see what raw images look like from the A7r.
p.8 #14 · Sony A7r some full resolution test shots
Andrew Gough wrote:
I think that we are mixing terms here. When we say flange distance, I am including the adapter in that distance. If we are talking manufacturer flange distance to the native mount then it is upto the lens designer to achieve optimal performance. If we add the adapter into the equation, then we introduce a lot variability into the flange (exit pupil) to sensor distance. SInce this tolerance is very small for FLE lenses, it is easy to see how they could move outside of their optimum zone.
I have shot the Lux on various M cameras and NEX's, and from practical experience, I can tell you that the adapter makes a big difference. I think that Sony micro lens architecture is not optimized for RF lenses, and that is the primary reason for poor performance. The Lux is very good on the M. What is surprising, is the poor performance of the APO 90mm R lens. This lens is also a very good lens on many full frame DSLR's unto and including the D800.
Anyway, it was not my intention to drive this thread away from its initial intent. I will wait and see what raw images look like from the A7r....Show more →
Did I miss something? Who tested and reported poor performance of the Leica R 90mm Apo lens?
p.8 #15 · Sony A7r some full resolution test shots
sebboh wrote:
well the R 90AA and m 90AA have the same design and very different flange distances.
This is the one I aware of, 80lux R and 75lux M also share similar design. But that is doesn’t make my point invalid, I need investigate if both 90AA have the same optical group to sensor distance. if they are the same, I agree the tolerance impact will be the same for both case.
sebboh wrote:
in fact, basic trig will show that angular displacement close to the sensor (short adapter) is better than angular displacement far from sensor (long adapter) because the center of the image circle will be closer to the center of the sensor (assuming angular displacement is the same for both adapters). either way both adapters will have the same difference in focus between opposite corners.
I have to do some research on this, I may get nothing in the end, but it seems to me the short the flange distance, the bigger in focus plane change for its relative angular error. So, in the end, in focused image will get more blur, because in this portion of image (mostly corner), it focus more closer or further compare to long flange distance.
sebboh wrote:
this point does not seem obvious to me at all and will require a technical explanation for me to believe.
I agree, It is very intuitive thinking. I got the idea from above reply to your angular error. Without known some fundamental lens design knowledge, I may have no way to prove it. But just learn a new trick from CarstenW: you might welcome to disprove it
Nanh wrote:
You can think of it this way, do you need to shift the glass on a DSLR 24mm lens more to focus it on 5m from infinity than a 24mm rangefinder lens? If not, then the focus error produced by uneven thickness of an adapter will be the same regardless of the flange distance.
I can’t Because based on lens size difference I have seen between Canikon and Leica, obviously I will move less distance for rangefinder lens. So this made my point.
But if they are the same design, same lens group to sensor distance I agree, but it doesn’t contradict my point.
p.8 #17 · Sony A7r some full resolution test shots
zhangyue wrote:
I can’t Because based on lens size difference I have seen between Canikon and Leica, obviously I will move less distance for rangefinder lens. So this made my point.
But if they are the same design, same lens group to sensor distance I agree, but it doesn’t contradict my point.
you are incorrect here, if there is no floating element the lens group moves the same distance for a given focal length independent of lens design or flange distance. you can measure yourself to verify - the elements on your R 35 cron move the same distance to focus from 1m to infinity as the elements on your m 35 cron v1.
p.8 #18 · Sony A7r some full resolution test shots
Andrew Gough wrote:
Phillip,
Thank you for taking the time to post these images - you might have saved me a lot of money! I am thinking of cancelling my order for the Sony. Given that the Minolta 100mm F/2.4 shows better corner performance than both Leica 90mm APO lenses, I am going to guess that it is a sensor issue as well. The second 90mm APO, is that the R version?
On modern floating element lenses, adapter accuracy is everything. I have struggled with poor adapters, and through shimming, I have achieved very good performance with FLE lenses. As Zhangyue pointed out, you cannot even guarantee centre performance on a poor adapter. Even if your adapter is perfect in every measurement, the sensor's micro lens architecture might be incompatible with a RF lens. It is becoming obvious, that Sony do not want us using other off brand small lenses. They want to sell you their own branded lenses, as that is where they are probably making most of their money.
Regarding the Leica Lux asph, it is sharp on my 24mp M all the way into the corners, so I highly doubt that it is the lens. My Summilux 50mm Pre-asph is super sharp on the M too.
I do not see any mention of Phillip conducting a test on a Minolta 100mm f2.4 lens. A 100mm f2.5 Minolta lens yes. Also, I do not see Phillip having done any tests on either the 90mm M Apo or the 90mm R Apo lenses. Am I missing something or where are you getting the information that the Minolta performs better for corner performance than the 2 90mm Apo lenses?
p.8 #19 · Sony A7r some full resolution test shots
sebboh wrote:
you are incorrect here, if there is no floating element the lens group moves the same distance for a given focal length independent of lens design or flange distance. you can measure yourself to verify - the elements on your R 35 cron move the same distance to focus from 1m to infinity as the elements on your m 35 cron v1.
I have used Lens group here for my reply, because I am not exact sure which surface I should use to measure this distance. the last element?, the aperture?
My assumption is the short flange distance can scale everything the same way make design smaller such as lens element size and distance. Hence more sensitive to tolerance.
I might be wrong if the truth is for given sensor size, and focal length, something might have been fixed.
I am not very convinced that is the truth: is it really fixed regardless of the flange distance? Must not be for given so many design difference on number, size, group of elements.....
I realize this become very off-topic, though it is not totally irrelevant for the matter discussed here in Alt forum. sorry about that, Philip.
p.8 #20 · Sony A7r some full resolution test shots
zhangyue wrote:
I have used Lens group here for my reply, because I am not exact sure which surface I should use to measure this distance. the last element?, the aperture?
doesn't matter they all move together as one group. the easiest way to measure is simply to measure how much the lens barrel moves if the lens barrel moves with the front element (i think both those lenses do). second easiest to measure would be the rear element movement, but it doesn't matter which, you measure because they all move together. you could also compare the R 50 lux to your cron rigid btw.
zhangyue wrote:
My assumption is the short flange distance can scale everything the same way make design smaller such as lens element size and distance. Hence more sensitive to tolerance.
I might be wrong if the truth is for given sensor size, and focal length, something might have been fixed.
I am not very convinced that is the truth: is it really fixed regardless of the flange distance? Must not be for given so many design difference on number, size, group of elements.....
you are, in fact, wrong on this. iris size for example, must be the same for a given focal length and aperture independent even of sensor size. the same is true for distance travelled to change from one focus point to another – it depends on focal length independent of sensor size for any lens design (assuming all lens elements move together). there is probably some subtle variation between lenses with different designs versus an idealized single element lens, but the differences are not big enough to make a difference that you would noticed if you switched helicoids on any two lenses.
zhangyue wrote:
I realize this become very off-topic, though it is not totally irrelevant for the matter discussed here in Alt forum. sorry about that, Philip.
? Can you elaborate this? were you saying 0.1mm difference in flange distance will have the same effect on Leica M and Leica R as long as their focal length are the same? So, based on this, can I assume all the lens will have same design no matter what flange distance as long as focal length the same? 
Because based on lens size difference I have seen between Canikon and Leica, obviously I will move less distance for rangefinder lens. So this made my point.