Matt Grum wrote:
My understanding based on reading everything that's been posted on the subject here and elsewhere is:
1. The filter stack imparts a small amount of spherical aberration everywhere in the image (this was mentioned in discussions about the MB speed booster, which Brian Caldwell claims corrects for this aberration).
2. The filter stack causes "smearing" in the corners when rays strike the sensor at an oblique angle.
3. The design sensor itself causes colour shift in the corners when rays strike the sensor an an oblique angle due to colour cross talk.
Lens designers can counter the first by leaving the lens overcorrected for SA, and the second by increasing the exit pupil distance (the third can be handled in software).
So if you remove the filter or make it much thinner, then the increased exit pupil distance is of no consequence but you will end up slightly overcorrected for SA.
This might go some way to explain why manufacturers are reluctant to change filter stacks. Changes to the registration distance / infinity focus as a result of thinning the stack can surely be solved by moving the sensor.
...and it seems like the issue with wides on the A7's - including most SLR wides - is that you are inherently going to get those oblique light rays off axis. Not as much as is the case with more symmetrical rangefinder designs of course but enough to cause a performanc hit at the borders and corners.
I suspect (others would know I'm sure. Maybe theSuede answered this with a diagram a while back?) that there is only so much a lens designer can do when dealing with the contraints of the Sony E-Mount used for FF with wide angle lenses (and very large aperture/ fast lenses as well). For instance, is it even possible to design a wide angle with an extreme telecentric design (resulting in practically parallel light rays off axis and at the borders) within the constraints of the E-mount used FF?
Tariq Gibran wrote:
I suspect (others would know I'm sure. Maybe theSuede answered this with a diagram a while back?) that there is only so much a lens designer can do when dealing with the contraints of the Sony E-Mount used for FF with wide angle lenses (and very large aperture/ fast lenses as well). For instance, is it even possible to design a wide angle with an extreme telecentric design (resulting in practically parallel light rays off axis and at the borders) within the constraints of the E-mount used FF?
I remember that thread but I couldn't quite understand the arguments about the physical mount limitations. Surely if the inner diameter of the mount is greater than the sensor diagonal and the rays are parallel to the sensor the image will fit through the mount just fine.
The main problem I imagine is that the resulting lens would be physically large which goes against the goals of the system.
In pursuit of the ultimate in compact wide angle image quality it might be necessary to pair the A7R with a fixed lens curved sensor body!
Matt Grum wrote:
2. The filter stack causes "smearing" in the corners when rays strike the sensor at an oblique angle.
Lens designers can counter the first by leaving the lens overcorrected for SA, and the second by increasing the exit pupil distance (the third can be handled in software).
the smearing seems to be a type of astigmatism caused by rays at different angles being refracted by different amounts relative to each other by the cover glass. from zeiss's description of the issue (i've lost the link for this), it will exist for all lenses unless all rays hit at 90 degrees to the sensor. of course the more telecentric a lens is the smaller the effect. roger has indicated that his optics consultants say that there is another way to correct for this besides excessive telecentricity. lens designers can induce an astigmatism in the opposite direction in the lens design so that the effect of the sensor stack is canceled out.
yes, they do. the FE 55 was designed for the a7* focus stack, while the 50 lux asph was designed for film. a comparison of the 50 lux asph on the m240 versus the same lens on the a7* shows how the differing filter stacks effect it's performance as well.
Steve Spencer wrote:
Edward, I think you are right about some of the dumb choice Sony has made (e.g., the translucent mirror in the DSLR "upgrade"), but I don't think the sensor stack is quite as bad as you make it out to be. As I understand the issue from Roger Cicala's excellent analysis with Brian Caldwell and several others, the issues is not as simple as thin stacks are good and thick stacks are bad. That is basically true if we are talking about lenses designed from film cameras. One important point to keep in mind is that it is worse the shorter the exit pupil. So for film lenses, the problems of the sensor stack are reduced if the sensor stack is thin and the exit pupil is long.
Leica which has to deal with a lot of legacy lenses designed for film, when they started making digital cameras smartly made the sensor stack as thin as they could. This was especially important for them because many of these legacy lenses had very short exit pupils. One thing that Leica has also done, however, is that as they have come out with new lenses since going digital it is pretty clear they have lengthened the exit pupil.
Now if I understand Roger and Brian's analysis, however, if you design the lenses for the sensor stack a thick stack does not need to cause problems. I suspect the 55FE is a little under designed to handle the sensor stack fully. I expect, however, these new Loxia lenses to be fully designed to handle the stack, which incidentally won't make them bad candidates for any other possible mirror less formats. Almost all FF cameras and APS-C cameras too (other than Leica), have about a 2mm sensor stack according to Roger's tear down, and the one clear exception is an interesting one (earlier Canon 1 series cameras that had to deal with a lot more lenses designed for film). So, I don't think the sensor stack will completely ruin these cameras. They will work fine for lenses designed for the sensor and although the lens won't quite react optimally even older lenses with a long enough exit pupil will still do quite well especially at narrower apertures. The problems are worse at wider apertures as well as one might expect....Show more →
As I mentioned before the FE 55 is just superb on the A7, but still, there is one feature of the A7 sensor that the lens isn't immune to. Not sure if it's astigmatism or something else, but the sensor seems to stretch bokeh, mostly visible in highlights and blur discs, in the image periphery. The stretch is similar to a UWA rectilinear effect. It also creates halos along the same lines. And that is the best lens that I rightfully nicknamed the mini-Otus. All other lenses I tried were actually worse.
sebboh wrote:
yes, they do. the FE 55 was designed for the a7* focus stack, while the 50 lux asph was designed for film. a comparison of the 50 lux asph on the m240 versus the same lens on the a7* shows how the differing filter stacks effect it's performance as well.
I don't think it's a fair comparison. I wish he would have compared the ZM 50 Planar instead. The 50 Lux is a special effects lens known for its horrible field curvature.
Tariq Gibran wrote:
I think the question is more intersting put this way - what are the benefits/ advantages of having a thick sensor stack? We know the disadvantages - and I tend to agree with Edward on those. One possibility is simply cost, maybe versatility of application. The other, which Leica has shown in the past, is proper IR filtering. Others?
According to Sony themselves, the thick filter stack allows for better ultrasonic dust shake off, as well as making the dust spots less visible in the image. They went even further with the A900 and placed the filter stack further away from the sensor. This resulted in very weak corners with perfectly superb lenses like the ZA 24-70 and 16-36, as well as the ZA 24. I tried these lenses on film and they were absolutely tack sharp in the corners.
edwardkaraa wrote:
I don't think it's a fair comparison. I wish he would have compared the ZM 50 Planar instead. The 50 Lux is a special effects lens known for its horrible field curvature.
from the samples I've seen the FE 55 is sharper in the corners at infinity from wide open to f/5.6 than the zm 50/2 on the m240. not sure about f/8 or smaller or how things go in the center. they have very different looks though which are much larger differences. personally I much prefer the bokeh from 55 at most distances, I expect you to feel the opposite though given previous discussions of sissy bokeh.
adamdewilde wrote:
Seems odd that the aperture ring moves in the opposite direction of existing Zeiss lens rings. I call shenanigans!
Yes, it seems this would upset everyone equally since the focus direction is what many would consider the "correct" way but then the aperture direction is the opposite!
sebboh wrote:
from the samples I've seen the FE 55 is sharper in the corners at infinity from wide open to f/5.6 than the zm 50/2 on the m240. not sure about f/8 or smaller or how things go in the center. they have very different looks though which are much larger differences. personally I much prefer the bokeh from 55 at most distances, I expect you to feel the opposite though given previous discussions of sissy bokeh.
Hahaha, touche! No but sincerely I really liked the bokeh of the FE 55 and I don't dispute its smoothness. The Planar has a more regular field curvature than the Lux, and even though it looses some resolution in the corners , the results are still outstanding, but for sure not at the mini-Otus levels. My problem with the FE 55 was that it did not have the Zeiss look, and mostly because the A7 sensor managed to alter its flawless performance. Add to that the lossy raw compression that caused the concentric rings in the sky, and i couldn't keep on working with this camera.
It does look much shorter than the ZM planar and the planar + adapter becomes very long on the Sony. This is definitely not the ZM design + longer mount.
edwardkaraa wrote:
As I mentioned before the FE 55 is just superb on the A7, but still, there is one feature of the A7 sensor that the lens isn't immune to. Not sure if it's astigmatism or something else, but the sensor seems to stretch bokeh, mostly visible in highlights and blur discs, in the image periphery. The stretch is similar to a UWA rectilinear effect. It also creates halos along the same lines. And that is the best lens that I rightfully nicknamed the mini-Otus. All other lenses I tried were actually worse.
Edward, I think the oof shape you note on the FE 55 is specific to that particular lens design. It bothers me as well - and is why I don't really think I will buy that lens. Other 50's don't do that universally on the A7's so I'm not so sure it's due to the sensor (or some design factor) stretching bokeh per se.
Tariq Gibran wrote:
Yeah, it's going to be really intersting to see the design of that "Biogon" 35/2. It can't be the exact design of the ZM version.
the zm 35/2 isn't that bad on the a7, so I can imagine it working with some minor reformulating for the sensor stack. hopefully they did more than that, I'm really not a fan of its bokeh at portrait distance.
, is proper IR filtering. Others?
