douglasf13 wrote:
I'm not saying that Sony has the thinnest stack. The very old 1Dsii, and the Leica cameras, have much thinner stacks. I'm just saying that the Sony stack isn't ridiculously thick. I guess I should have said a little thicker than average.
Sony has the second-thickest stack, second only to the very tele-centric MFT camera system. This would dictate that Sony should also make very tele-centric lens designs to get good performance, which is, frankly, ridiculous. Why make the camera so small, and the register distance so short and then mandate compromises in lens design of this nature? Makes absolutely no sense. They should have gone for a much thinner sensor topping to reap the benefit of the short register distance.
carstenw wrote:
Sony has the second-thickest stack, second only to the very tele-centric MFT camera system. This would dictate that Sony should also make very tele-centric lens designs to get good performance, which is, frankly, ridiculous. Why make the camera so small, and the register distance so short and then mandate compromises in lens design of this nature? Makes absolutely no sense. They should have gone for a much thinner sensor topping to reap the benefit of the short register distance.
That 3mm is a bit of a guess on the part of Roger's team, because it is difficult to measure. Either way, it's just a bit thicker than most of the other cameras, and I don't think the difference would have been huge had it been only 2mm. The important thing is that the lenses are designed for whatever the stack is. Making a super thin stack to accommodate Leica lenses may have caused issues with Sony legacy DSLR and e-mount lenses.
I still think Sony should have made the FE mount larger, which would have allowed more telecentric designs that could have likely been shorter, albeit larger in diameter. As it stands now, as theSuede has mentioned, any lenses faster than around f2 will essentially be the same size as the DSLR equivalent with an adapter attached, although, like we're seeing with the 55/1.8, the lack of a mirror box still allows for more interesting designs at that size.
Well, I don't know a lot about the theoretical part of the sensor stack designs, but I do know that my D800 and A7R give (as far as I can tell) the same optical results using the same lenses. Other than the A7R being slightly sharper, lenses that have smearing off in the corners on my D800 have exactly the same behavior on the A7R AND lenses that are sharp into the corners on my D800 are very much the same on the A7R.
Also, I've found that the sensor reflections are identical (as near as I can reproduce) between the two cameras. Now the A7 is a different story as the sensor reflections are much worse compared to the A7r/D800 - one of the primary reasons I bought the A7r over the A7.
So whatever toppings the A7r has over the sensor it seems to act very much identical to the D800.
Also it seems that the exit pupil distance number alone does not correlate to how the lens does off in the corners. My 16/3.5 AI fisheye is excellent across the frame (even wide open) on the D800 and A7r and it has an exit pupil distance of 51.5mm (it's pretty much the sharpest lens I own). My 45/2.8P has a slightly larger exit pupil distance (52mm) and it's smeary in the borders/corners unless stopped down quite a bit (and even then it never quite goes away at the very far corners). My 20/2.8D (54mm exit pupil) is also pretty smeary off in the borders/corners
So it seems like it's a combination of focal length, exit pupil distance, and other lens design attributes that indicate if a lens will do well on a full-frame sensor.
I will sit back and listen again to those much smarter to me on this topic . . . . Great conversation so far.
Concerning the angst over the 35 & 50 Loxia lenses being possibly 'repackaged' ZMs - perhaps the modifications required to improve these two ZM lenses to whatever specification Zeiss has for Loxia were the least complex? Therefore easiest/fastest to bring to market?
Thinking about this briefly - are there any currently available Zeiss lenses that are sub-par performers? Considering that Loxia is aimed at a niche within a niche (manual focus prime lens, mirrorless lens mount, likely at a price premium), they've got to know they'll be dealing with very discerning and informed photographers, many using a camera with the highest resolution FF sensor currently available. A sensor that will reveal all sorts of defects. And sensor resolution will certainly increase during the lifetime of these lenses. Expectations will be high. Given that Zeiss published information some time ago (see CLB41 - From the series of articles on lens names: Distagon, Biogon and Hologon by H. H. Nasse - Dec. 2011) explaining the drawbacks of using certain of their ZM lenses on digital, they obviously know what the problems are, and the required solutions.
Bottom line for me is I would be surprised if these lenses are significantly compromised. Do I expect perfection? No, but I expect above average performance - better than Sony's offerings and better than most vintage SLR glass. Price point will also dictate the expected level of performance.
Apparently there will be an official announcement early next week...
jhinkey wrote:
Well, I don't know a lot about the theoretical part of the sensor stack designs, but I do know that my D800 and A7R give (as far as I can tell) the same optical results using the same lenses. Other than the A7R being slightly sharper, lenses that have smearing off in the corners on my D800 have exactly the same behavior on the A7R AND lenses that are sharp into the corners on my D800 are very much the same on the A7R.
Also, I've found that the sensor reflections are identical (as near as I can reproduce) between the two cameras. Now the A7 is a different story as the sensor reflections are much worse compared to the A7r/D800 - one of the primary reasons I bought the A7r over the A7.
So whatever toppings the A7r has over the sensor it seems to act very much identical to the D800.
Also it seems that the exit pupil distance number alone does not correlate to how the lens does off in the corners. My 16/3.5 AI fisheye is excellent across the frame (even wide open) on the D800 and A7r and it has an exit pupil distance of 51.5mm (it's pretty much the sharpest lens I own). My 45/2.8P has a slightly larger exit pupil distance (52mm) and it's smeary in the borders/corners unless stopped down quite a bit (and even then it never quite goes away at the very far corners). My 20/2.8D (54mm exit pupil) is also pretty smeary off in the borders/corners
So it seems like it's a combination of focal length, exit pupil distance, and other lens design attributes that indicate if a lens will do well on a full-frame sensor.
I will sit back and listen again to those much smarter to me on this topic . . . . Great conversation so far.
i'm no expert either (very far from it), but when you're using lenses from other mounts adapted to the e-mount on the A7, the adapter is increasing the flange distance and decreasing the ray-angle into the corners.
the concerns with the ratio between flange distance to sensor stack, or simply the ray angle into the corners, relate to lenses and their native mount measurements.
A thin stack wouldn't have caused any issues with legacy DSLR lenses because most were designed for film, and curiously, even the ones designed for digital don't seem to have taken the sensor stack in consideration (look at the FE zooms, and even the FE 35 corner performance). As far as I'm concerned, Sony's position regarding the sensor cover is undefendable and incomprehensible.
Roger and Brian Caldwell did a practical experiment to show that lenses are (must be) optimized for their intended receptor. It's on the lensrentals blog somewhere.
But it's quite important to remember that there's no real DIFFERENCE between optimizing for a thick stack or a thin stack. You can - as long as you don't have reflection problems between stack elements, like the A7 camera - use whatever you want. The only thing that changes is how easy it is to use lenses from one manufacturer (optimization) on another manufacturer's body.
The reason Leica did what they did is not because of "quality". It was purely to keep compatibility with older lenses. They actually lost quality because of it, but they seem to be learning how to get around the difficulties you get from having a thinner stack (insufficient filtering, angle-dependent UVIR energy, hence weird color responses).
ONT: I'm not to keen on the fact that they (Zeiss) used a 6/4 construction for the planar 50. We have much better performing base constructions nowadays, extended ultrons and other variations on the planar. Optimizing a pure planar for LoCA is impossible outside the 1:1 magnification domain, optimizing it for the other important aberrations together, at the same time, at multiple target distances, is also very hard.
Take a simple and cheap lens, like the Nikon 50/1.8, or the Canon or other corresponding lens. They're already bordering on "really good", and they're simple 6/7 constructions. Then put some more construction effort and production cost into optimizing it. At double the production cost, you COULD get something truly remarkable when you have the shorter RFD of the E-mount. Now, they can never match the production series lengths (and hence neither the production cost optimizations...) of the cheaper lenses - but the "cheap" lenses can retail for about 120-150USD...
Sony themselves opted for a base construction that gives a totally different character than a planar in their 55/1.8, which I for one definitely applaud. IMO Sonnar>Distagon>Planar>Tessar in image usability and pleasantness. I don't like the edgyness you get from constructions with abrupt aberration line modulations, even though it can give superbly sharp renditions of what is actually IN focus when stopped down a bit. Some find it "interesting", I find it "distracting" and "ugly".
jhinkey wrote:
Also, I've found that the sensor reflections are identical (as near as I can reproduce) between the two cameras. Now the A7 is a different story as the sensor reflections are much worse compared to the A7r/D800 - one of the primary reasons I bought the A7r over the A7.
So whatever toppings the A7r has over the sensor it seems to act very much identical to the D800.
- John
Great to hear that experience of lenses with the A7r vs D800.
theSuede wrote:
Roger and Brian Caldwell did a practical experiment to show that lenses are (must be) optimized for their intended receptor. It's on the lensrentals blog somewhere.
jhinkey wrote:
Well, I don't know a lot about the theoretical part of the sensor stack designs, but I do know that my D800 and A7R give (as far as I can tell) the same optical results using the same lenses. Other than the A7R being slightly sharper, lenses that have smearing off in the corners on my D800 have exactly the same behavior on the A7R AND lenses that are sharp into the corners on my D800 are very much the same on the A7R.
Also, I've found that the sensor reflections are identical (as near as I can reproduce) between the two cameras. Now the A7 is a different story as the sensor reflections are much worse compared to the A7r/D800 - one of the primary reasons I bought the A7r over the A7.
So whatever toppings the A7r has over the sensor it seems to act very much identical to the D800.
Also it seems that the exit pupil distance number alone does not correlate to how the lens does off in the corners. My 16/3.5 AI fisheye is excellent across the frame (even wide open) on the D800 and A7r and it has an exit pupil distance of 51.5mm (it's pretty much the sharpest lens I own). My 45/2.8P has a slightly larger exit pupil distance (52mm) and it's smeary in the borders/corners unless stopped down quite a bit (and even then it never quite goes away at the very far corners). My 20/2.8D (54mm exit pupil) is also pretty smeary off in the borders/corners
So it seems like it's a combination of focal length, exit pupil distance, and other lens design attributes that indicate if a lens will do well on a full-frame sensor.
I will sit back and listen again to those much smarter to me on this topic . . . . Great conversation so far.
Fisheye lenses, being not rectilinearly corrected, have absolutely no corner smearing problems, neither on film, nor on digital. The price you pay of course, is a fish vision
ecarlino wrote:
i'm no expert either (very far from it), but when you're using lenses from other mounts adapted to the e-mount on the A7, the adapter is increasing the flange distance and decreasing the ray-angle into the corners.
the concerns with the ratio between flange distance to sensor stack, or simply the ray angle into the corners, relate to lenses and their native mount measurements.
All I'm simply saying is that the D800 and A7r sensors when used with identical F-mount lenses show the same results - so at least for these lenses the toppings, if different, don't show any differences.
Now you are likely correct that when the ray angle becomes more extreme (say when you use a native E-mount lens or and adapted M-mount) the difference may manifest themselves more greatly - i.e, become much more noticeable.
theSuede wrote:
ONT: I'm not to keen on the fact that they (Zeiss) used a 6/4 construction for the planar 50. We have much better performing base constructions nowadays, extended ultrons and other variations on the planar. Optimizing a pure planar for LoCA is impossible outside the 1:1 magnification domain, optimizing it for the other important aberrations together, at the same time, at multiple target distances, is also very hard.
Take a simple and cheap lens, like the Nikon 50/1.8, or the Canon or other corresponding lens. They're already bordering on "really good", and they're simple 6/7 constructions. Then put some more construction effort and production cost into optimizing it. At double the production cost, you COULD get something truly remarkable when you have the shorter RFD of the E-mount. Now, they can never match the production series lengths (and hence neither the production cost optimizations...) of the cheaper lenses - but the "cheap" lenses can retail for about 120-150USD...
Sony themselves opted for a base construction that gives a totally different character than a planar in their 55/1.8, which I for one definitely applaud. IMO Sonnar>Distagon>Planar>Tessar in image usability and pleasantness. I don't like the edgyness you get from constructions with abrupt aberration line modulations, even though it can give superbly sharp renditions of what is actually IN focus when stopped down a bit. Some find it "interesting", I find it "distracting" and "ugly"....Show more →
Comparing photos from the Contax G45 and SEl55/1.8, the Contax definitely has more issues with LoCA at wide apertures. However, I'm an absolute fan of the "edginess", and it's part of the intrinsic character of that lens I love (for certain subjects). "Edginess" can go too far (like what I see in some Foveon images), but Planars strike the right balance for my tastes. I also find that extra microcontrast makes manual focusing on an EVF much easier, which would be important for this lens.
jhinkey wrote:
All I'm simply saying is that the D800 and A7r sensors when used with identical F-mount lenses show the same results - so at least for these lenses the toppings, if different, don't show any differences.
given that Sony makes Nikon's sensors, the D800 probably has a similar (I'm not suggesting identical) stack, so once you've adapted a Nikkor onto the A7r - it should have nearly identical performance because the flange distance will be the same and the stack is similar.
edwardkaraa wrote:
Fisheye lenses, being not rectilinearly corrected, have absolutely no corner smearing problems, neither on film, nor on digital. The price you pay of course, is a fish vision
I've never been quite clear on the difference between "smearing" and plain old blurry-ness from not being sharp.
For instance, my 16/3.5 AI is very sharp off in the corners while the 16/2.8D is not all that sharp off int he corners (at any aperture). So is this lack of sharpness of the 16/2.8D due to the lens design (fundamentally not sharp) or to the interaction of the sensor (smearing)? Typical film just can't show the sharpness that a 36MP FX sensor can, so perhaps hid this lack of sharpness or smearing that is evident now with high MP FX sensors.
ecarlino wrote:
given that Sony makes Nikon's sensors, the D800 probably has a similar (I'm not suggesting identical) stack, so once you've adapted a Nikkor onto the A7r - it should have nearly identical performance because the flange distance will be the same and the stack is similar.
Yes. This was questioned earlier though which is why hearing some real world confirmation is helpful.
ecarlino wrote:
given that Sony makes Nikon's sensors, the D800 probably has a similar (I'm not suggesting identical) stack, so once you've adapted a Nikkor onto the A7r - it should have nearly identical performance because the flange distance will be the same and the stack is similar.
But the sensors and the cover glass/stack are two different things that are separable. So the underlying sensor might be the same, but that doesn't mean the glass on top is (AFAIK). Someone else smarter than me will chime in on this.
All I know is that the sensor + glass covering appear to behave very much identical for the F-mount lenses that I have on the D800/A7R.
What would be very interesting would be to take the D800 and A7R sensors and put lenses on them that were designed for much shorter registration distances and see if they still behave the same. Of course that would require dis-assembly (and likely destruction if I did it) of a perfectly fine D800 . . .
edwardkaraa wrote:
A thin stack wouldn't have caused any issues with legacy DSLR lenses because most were designed for film, and curiously, even the ones designed for digital don't seem to have taken the sensor stack in consideration (look at the FE zooms, and even the FE 35 corner performance). As far as I'm concerned, Sony's position regarding the sensor cover is undefendable and incomprehensible.
I know that theSuede has mentioned that our old A900s had a lower quality AA, and that seemed to cause smearing compared to the D3X at the time, but I think that was independent of the actual sensor stack thickness that we're talking about here. A thicker stack isn't much of an issue, if the materials are good and the lenses are designed for it. The issue is that we all seem to be interested in the performance of lenses not designed for the camera, and that's optimistic when thinner sensor stacks can produce other issues.
I think, ultimately, that the concept of adapting legacy lenses to digital cameras is a tough proposition, now that we're all getting so technical about things like corner performance. Unlike film, where we had a relatively consistent medium for decades, digital sensor sizes and types are not standardized, so keeping a lens for twenty years and expecting it to perform brilliantly on all cameras is going the way of the dodo. I no longer think of any lens that I buy as an option for anything other than the current camera system I'm using, and new advances in digital imagery, like curved sensors, will make things even tougher.
As far as Loxia, I'm worried that we just have ZM lenses in a new shell, which would be an odd decision.
jhinkey wrote:
But the sensors and the cover glass/stack are two different things that are separable. So the underlying sensor might be the same, but that doesn't mean the glass on top is (AFAIK).
All I know is that the sensor + glass covering appear to behave very much identical for the F-mount lenses that I have on the D800/A7R.
when i had the D800, i called NikonRepair and asked if i could just change the sensor glass to make it a D800E (after all the concern about moire passed) - they told me the stack and sensor were one and could not be separated.
Roger (via the link i posted above) said that the A7r's stack is glued to the sensor.
The fact that you observe similar performance between the 2 cameras with the same lens also suggests they very similar.
