p.7 #2 · A thinner sensor stack may be possible after all!
Steve Spencer wrote:
And to keep this on topic, if they made a thinner filter stack, then IMO this strategy of luring Canon and Nikon shooters would be much more difficult. C/N shooters lenses would not work as well on the Sony cameras giving them a good reason not pick up that extra Sony camera for the relevant particular use.
Is there evidence to support the hypothesis that traditional lenses would not work as well with a thinner stack? At least with the LensRentals test using the Sony FE 35/2.8 and 55/1.8, there was no significant difference on the thin stack vs stock stack. Perhaps with longer and very fast lenses it would matter more but with the wider lenses, it likely makes no difference.
Jan 22, 2015 at 11:51 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.7 #3 · A thinner sensor stack may be possible after all!
Tariq Gibran wrote:
Is there evidence to support the hypothesis that traditional lenses would not work as well with a thinner stack? At least with the LensRentals test using the Sony FE 35/2.8 and 55/1.8, there was no significant difference on the thin stack vs stock stack. Perhaps with longer and very fast lenses it would matter more but with the wider lenses, it likely makes no difference.
I am very sceptical that it won't make any difference remember in the lens rental first look they only measured performance at a close focussing distance. Even at close focussing distance there was a tiny bit of deterioration, but that small effect might be much larger at infinity focus. Keep in mind many wide Leica M lenses look pretty good at close focus distances and have very few issues there, but show clear problems at infinity focus. I suspect the same will be true for FE lenses at infinity focus on the modded cameras with a thin filter stack. We should know before long, but it seems like wishful thinking that lenses designed for a thicker stack will work just as well on a thinner stack.
p.7 #4 · A thinner sensor stack may be possible after all!
I agree that, if a lens is truly designed for a particular optical path, it should perform optimally in a system with that optical path. What is not clear is to what degree lenses are being designed for a particular path (i.e. filter stack) vs. having other aspects (for example, highly retrofocal designs) that just make them less sensitive to the changes to the optical path that result from changes in filter stack optical thickness.
It should be required for all commenting on this to go back and read, in detail, the early analysis by Roger and Brian on the topic. I see a lot of posts that seem to disregard the basic analysis provided in these by a highly competent optical designer. These earlier posts on the LensRentals blog are all linked to in the most-recent one.
It is also interesting to look at the variability in filter stack thickness across the DSLR lines of Nikon and Canon (which are very significant, see the summary in the aforementioned posts), which at least makes me wonder what optical path their lenses are optimized for. Similarly, one would have to wonder how, for example, Zeiss is designing the ZF and ZE series lenses. Are they actually optimized optical formulas, or just different mounts and electronics?
The analysis by Brian shows that differences between the 'designed for' optical path and the actual path in either direction will cause errors with increasing ray angle. The factory Sony stack is on the high end of the DSLR ranges. Bringing that down might actually be closer to the mid-point of the Canon DSLRs so it might even improve performance of these lenses. Or (more likely, IMO) it might not matter much because they have long enough exit pupil distances (as DSLR lenses) that the errors caused by the stack might not be that significant in any case. Although Lloyd's test results with the Leica 50mm Summicron APO were a little surprising in this regard, since many (not Lloyd, by the way, from his earlier testing) felt this was already beyond the range where it mattered.
In the end, time, testing, and real-world use will tell the story. It's fun to speculate on threads like this about technical and business strategy matters, but it's not really much more than that.
I never set out to change Sony's strategy with this project - just to have a camera that was more satisfying for the photography I do (mdemeyer.leicaimages.com for some examples) and the equipment I own. Perhaps combined with a strong technical curiosity and unwillingness to accept limitations that I think can be addressed.
Again, I welcome those who are like minded to join the fun.
p.7 #5 · A thinner sensor stack may be possible after all!
Steve Spencer wrote:
I am very sceptical that it won't make any difference remember in the lens rental first look they only measured performance at a close focussing distance. Even at close focussing distance there was a tiny bit of deterioration, but that small effect might be much larger at infinity focus. Keep in mind many wide Leica M lenses look pretty good at close focus distances and have very few issues there, but show clear problems at infinity focus. I suspect the same will be true for FE lenses at infinity focus on the modded cameras with a thin filter stack. We should know before long, but it seems like wishful thinking that lenses designed for a thicker stack will work just as well on a thinner stack....Show more →
I'm wondering what the principle behind your skepticism is. With film lenses on cameras with thick sensor stacks, the loss of quality was the result of the fact the exit pupil was so short that the sharp angle that the light went through the glass caused problems. Lenses with longer exist pupil distances solved that problem for sensor stacks. In principle such lenses would be significantly less affected by a thin stack than short exit pupil lenses would be on a thick stack.
Jan 22, 2015 at 04:03 PM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.7 #6 · A thinner sensor stack may be possible after all!
MAubrey wrote:
I'm wondering what the principle behind your skepticism is. With film lenses on cameras with thick sensor stacks, the loss of quality was the result of the fact the exit pupil was so short that the sharp angle that the light went through the glass caused problems. Lenses with longer exist pupil distances solved that problem for sensor stacks. In principle such lenses would be significantly less affected by a thin stack than short exit pupil lenses would be on a thick stack.
Except Roger and Brian in their analysis showed that the Zeiss Otus 55 f/1.4 was adversely affected when no glass was in optical path vs. when 2mm thick glass was in the optical path. This pretty clearly demonstrates to me that a lens with a long exit pupil (but wide aperture) can be adversely affected by thinner than designed cover glass (in this case no cover glass). Remember exit pupil and max aperture both come into play in causing problems for the filter stack, so fast lenses even with pretty long exit pupils like the Otus can cause problems.
p.7 #7 · A thinner sensor stack may be possible after all!
Steve Spencer wrote:
And to keep this on topic, if they made a thinner filter stack, then IMO this strategy of luring Canon and Nikon shooters would be much more difficult. C/N shooters lenses would not work as well on the Sony cameras giving them a good reason not pick up that extra Sony camera for the relevant particular use.
My impression is just the opposite Steve. The A7 stacks are thicker than any Canon Stacks, and the 1740, as shown by Fred a year ago, is not as good on the A7r as on the 5D.
The Canikon lenses do not perform as well on the Sony A7 series as on the Native bodies, the best of which have very thin stacks indeed. Even the Nex 5 had a thinner stack.
The "Lure" route would have been to actually test those lenses and tune the stack for them, which would have been very very smart.
The Otus is a very recent design, and an outlier lens in every way.
Jan 22, 2015 at 04:42 PM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.7 #8 · A thinner sensor stack may be possible after all!
uhoh7 wrote:
My impression is just the opposite Steve. The A7 stacks are thicker than any Canon Stacks, and the 1740, as shown by Fred a year ago, is not as good on the A7r as on the 5D.
The Canikon lenses do not perform as well on the Sony A7 series as on the Native bodies, the best of which have very thin stacks indeed. Even the Nex 5 had a thinner stack.
The "Lure" route would have been to actually test those lenses and tune the stack for them, which would have been very very smart.
The Otus is a very recent design, and an outlier lens in every way....Show more →
Hi Charlie,
The A7 stack is not thicker than any Canon stack. For example, according to the lens rental data base the A7 measures 1.85mm and the 6D measures 2.0mm. They are actually very close in physical measurement, but in these very recent cameras the Canon has the thicker sensor stack. What really matters is optical measurement and we have almost no data at all on the optical measurement of sensor stacks.
With regard to the 17-40 on the A7r, I didn't see that analysis, but I am beginning to do some of my own and one of the issues with Canon lenses on the Sony cameras is that the metabones adapters are almost all too thin. I think that may have been the issue for Fred as well as I know he got better results with Canon lenses when he got a different adapter.
I think by far the biggest problem with Canon lenses on the Sony right now is the adapter and that is far more important than the sensor stack thickness, which is quite similar between Sony and Canon.
p.7 #9 · A thinner sensor stack may be possible after all!
It is worthy to note that Lloyd has just posted his test results with 21 SEM and A7rM versus the Stock.
Very interesting results showing that the 21 SEM is an excellent lens on the A7rM. I am disappointed that I have just sold my 21 SEM !! The truth is having the WATE is a better option as it is excellent on all A7's. I do hope there may be further improvements with the WATE. Generally the WATE performs best at the 16mm with M series, so this may also translate to the A7rM.
The magenta is still present but from my observations with all f stops the magenta cast is less with the A7rM than with the A7r stock.
It will be easy to incorporate with the Lr workflow corrections at different f stops without having to resort to C1 Pro with LCC corrections. The WB disc shots at different f stops only has to be done once, and files then kept within Lr.
p.7 #10 · A thinner sensor stack may be possible after all!
Steve Spencer wrote:
Except Roger and Brian in their analysis showed that the Zeiss Otus 55 f/1.4 was adversely affected when no glass was in optical path vs. when 2mm thick glass was in the optical path. This pretty clearly demonstrates to me that a lens with a long exit pupil (but wide aperture) can be adversely affected by thinner than designed cover glass (in this case no cover glass). Remember exit pupil and max aperture both come into play in causing problems for the filter stack, so fast lenses even with pretty long exit pupils like the Otus can cause problems.
I'm assuming this is the analysis you're referring to?
This is the difference between 0mm and 2mm. So the difference between between the Sony A7 stack and whatever they replaced it with is necessarily less than a 2mm difference. A 1mm difference? I would expect that would be roughly the sort minimal difference describe for the native lenses.
I'm not saying you shouldn't be skeptical. I think that's a wise position to take this early on. Considering the very tentative nature of this project, I'd say having a devil's advocate is a very good thing.
Jan 28, 2015 at 08:43 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.7 #11 · A thinner sensor stack may be possible after all!
MAubrey wrote:
I'm assuming this is the analysis you're referring to?
This is the difference between 0mm and 2mm. So the difference between between the Sony A7 stack and whatever they replaced it with is necessarily less than a 2mm difference. A 1mm difference? I would expect that would be roughly the sort minimal difference describe for the native lenses.
I'm not saying you shouldn't be skeptical. I think that's a wise position to take this early on. Considering the very tentative nature of this project, I'd say having a devil's advocate is a very good thing.
Yes, that is the graph I was thinking of, but there is still some ambiguity about how thick the A7 and particularly the A7r cover glass is. In the lens rental database they list the A7 as having a physical 2mm cover glass, but they list the A7r as having 2.85mm in optical measurement which they say is what is critical. So, if the optical measurement is 2.85mm for the A7r (and maybe all A7 series camera), then the replacement cover glass could be more than 1.5mm and maybe as much as 2mm. So, there is so much that we still don't know. Hopefully Roger will do some testing of the FE lenses on his optical bench (if he can, I seem to remember there were some issues with that, but at least he could test the Loxia lenses) and we will get a better idea of how the lenses will perform at infinity focus.
p.7 #12 · A thinner sensor stack may be possible after all!
The finding that the 35 ASPH Summicron with 2mm CG: 'Overall I'd say it wasn't better or worse, just different.'
Might be a burgeoning market for APS-C sensor users by fitting extra cover glass, the data are much better out to IH 15mm before outer frame weirdness sets in. Leica does Zeiss flat curves in a wide Summ-M. ;-) Interesting days indeed.
p.7 #13 · A thinner sensor stack may be possible after all!
The manual of my ancient Minolta 250mm stated quite clearly that one of their (glass)filters had to be at the end of that lens in order to give satisfsctory performance (on analogue film). Other optics may well behave otherwise, but since this long one had that caveat, I am led to believe that refracion and dispersion charachteristics of the final glass does matter.
p.7 #14 · A thinner sensor stack may be possible after all!
paulhofseth wrote:
The manual of my ancient Minolta 250mm stated quite clearly that one of their (glass)filters had to be at the end of that lens in order to give satisfsctory performance (on analogue film). Other optics may well behave otherwise, but since this long one had that caveat, I am led to believe that refracion and dispersion charachteristics of the final glass does matter.
An optical flat behind the lens gives a focus shift, which the user would certainly notice if the filter is removed. It also introduces spherical aberration, all over the field, which may be noticeable with large-aperture lenses. Finally it causes astigmatism, smearing the corners when the lens design has the exit pupil close to the sensor.
p.7 #16 · A thinner sensor stack may be possible after all!
Charlie, I look forward to your testing with the modified A7
Mine will be a short while longer as it needs to be sent via FedEx to Australia.
From all the real test results with A7r modified with Roger/Brian and Lloyd, I will very happy if the M mount glass performs close to what I have seen. Of course there maybe other issues with other lenses, but until I purchase these lenses this may not be an issue. By then I will have acquired the A7rII/A9.
p.7 #17 · A thinner sensor stack may be possible after all!
Charles I forgot to add yo have me very excited with your reports from Lloyd. If the SEM21 is now good, that will be some huge progress, because, as you know, unmodified, that lens is a pale shadow of it's true, best 21 in the world, self
p.7 #19 · A thinner sensor stack may be possible after all!
charles.K wrote:
It is worthy to note that Lloyd has just posted his test results with 21 SEM and A7rM versus the Stock.
Very interesting results showing that the 21 SEM is an excellent lens on the A7rM. I am disappointed that I have just sold my 21 SEM !! The truth is having the WATE is a better option as it is excellent on all A7's. I do hope there may be further improvements with the WATE. Generally the WATE performs best at the 16mm with M series, so this may also translate to the A7rM.
The magenta is still present but from my observations with all f stops the magenta cast is less with the A7rM than with the A7r stock.
It will be easy to incorporate with the Lr workflow corrections at different f stops without having to resort to C1 Pro with LCC corrections. The WB disc shots at different f stops only has to be done once, and files then kept within Lr.
I am now looking forward to getting my A7rM back ...Show more →
Hi Charles,
You have confused me a little as you are going back and forth in your discussion of the 21mm f3.4 SEM and the WATE. Are you saying that Lloyd's testing showed that with the 21mm SEM that the magenta cast was still present at all f stops but less visible with the A7rM than the A7r stock? Or are you indicating this to be the case with the WATE?
Also, you indicated that on an M body that the WATE performs better at 16mm than at 21mm is that correct? I believe that on a stock A7r that the WATE is supposed to perform better at 21mm or am I incorrect? I have only used mine at 21mm so far and not nearly as much as you have used yours. Also, I have normally been shooting my WATE at f8 due to the circumstances and the desire to maintain DOF. In what conditions have you been getting the Magenta cast in the corners for the WATE? And how are you eliminating the cast?
Also, regardless, if the 21mm SEM now will perform outstandingly well on the A7rM the WATE would still be a more flexible lens since it offers 3 focal lengths and everything in between. I would have to consider the purchase of a 21mm SEM if that were the case providing I modified my A7r.
I will have to await everyone's tests with their modified cameras before I even consider making the modification largely due to the usage of my Leica R lenses (particularly my 28mm f2.8 Elmarit V2, 100mm f2.8 Apo-Macro-Elmarit, 280mm f4 Apo Telyt, my 180mm f3.4 Apo Telyt, and my R zoom lenses). I know that Derek has suggested that the modification should have little or no affect on the performance of the R lenses since they were designed for film, but I would want confirmation of this.
p.7 #20 · A thinner sensor stack may be possible after all!
Hi Rich,
I am sorry I confused the discussion with the WATE. Lloyd's test results with the 21 SEM performed really well. There is a vast difference between the A7rM and stock version with the 21 SEM. Yes there is some magenta cast, but with the A7rM it appears to be less than for the A7r stock at a given f stop. This is easily compensated within the Lr workflow, by taking shots at the different f stop with a WB discs and keeping this on file within Lr. This only has to be done once. Also you can use C1Pro which has the LCC for profile corrections. I still think the simplest is using Lr.
From test results I have seen it has been suggested that the WATE best performs at 16mm with the M240. I cannot confirm this as I had already sold my M240. This does make sense, and I suspect the wider angle at 16mm may be causing more issues with the A7r and less at 21mm. I am hoping this will improve with the A7rM and perform better at all lengths. With regards to the magenta cast, I have had no need to compensate to date, so should improve with the A7rM. The only time I see a very slight cast is with bright side lighting. This I also had with the M240 and M9.
I have no intent acquiring the 21 SEM again, as I have just sold it! The WATE is so much more versatile, and I can use it on the A7II, A7s and A7rM.
The other lens I am hoping that will work really well is the 50 Lux Asph. This a great lens, and hopefully Lloyd's test results with the 50 Cron APO will translate to the 50 Lux Asph.
