Is it phyiscally possible for a NEX bright pancake prime?

[ProfHankD]

As I see it, it is actually fairly easy to make a faster-than-f/2 fixed-focal-length pancake for the NEX in the 20-35mm range. Basically, take a classic double-Gauss "fast fifty" design and scale it down -- it should work fine for NEX (both the coverage and flange distance scale down: FF-> APS-C and 44.5-> 18mm).

Basically, I believe the reason the 16mm f/2.8 is f/2.8 is to make room for the autofocus and aperture motors and couplings. It also is a little shorter focal length than is easy (so as to not overlap the kit zoom?) and yet it is a cheap lens.

Sales volume on the 16mm was probably Sony's way of checking if the market would pay for developing pancake primes. If so, I think the answer Sony got was no, despite what people say. I think they are reconsidering that now, but it might be an easier product for Samyang -- they do fancy glass in manual mounts all the time....

 

[blue_skies]

As I see it, it is actually fairly easy to make a faster-than-f/2 fixed-focal-length pancake for the NEX in the 20-35mm range. Basically, take a classic double-Gauss "fast fifty" design and scale it down -- it should work fine for NEX (both the coverage and flange distance scale down: FF-> APS-C and 44.5-> 18mm).

Aren't those the Biogons?

They are very compact, but 'pancakes' -as discussed here- they are not.

Basically, I believe the reason the 16mm f/2.8 is f/2.8 is to make room for the autofocus and aperture motors and couplings. It also is a little shorter focal length than is easy (so as to not overlap the kit zoom?) and yet it is a cheap lens.

Probably f/2.8 for image quality issues. Most people complain, but then they compare with wide angles that start at f/4.0

Sales volume on the 16mm was probably Sony's way of checking if the market would pay for developing pancake primes. If so, I think the answer Sony got was no, despite what people say. I think they are reconsidering that now, but it might be an easier product for Samyang -- they do fancy glass in manual mounts all the time....

Doubtful, the 16mm could be a pancake because of its focal length. A 30mm would not be a pancake. I think that having a wide, relatively fast, pancake lens was laudable. It not overlapping with the kit zoom makes sense, and the two adapters for ultra-wide and fish-eye is very clever.

BTW - anyone try the conventional wide or tele (front) objectives? - They don't work on the E-mount lenses.

--
Cheers,
Henry

 

[blue_skies]

Just one caution: the Nex sensor falls off (vignets) signifcantly at the outer edges when the rear lens element is moved close. This shows up in the Biogon-style RF lenses (20-40mm). Nex-5 and Nex-7 are worse, Nex-5N is somewhat better.

Because of this 'angle of light inclination' light fall-off, I presume that all E-mount lenses are retrofocus to some degree. All SLR lenses (e.g. Distagons) are retrofocus and do not exhibit this vignetting (well, very wide vignets, but for other reasons).

Having to start with a retro-focus design constraint, the space for the lens elements is pushed outwards from the sensor - causing lenses to get larger. This may be an underlying reason why the E-mount lenses are the sizes that they are.

• But as I just read in another posts in this thread - the X-1 Pro lenses are essentially the same size as the E mount lenses. So, perhaps it is pure a design issue.
• Also - lenses are always a little longer, the casing always extends past the foremost lens element, except in ultra-wides (that don't accept filters, etc.)

Leica and Fuji compensated for this by pusing the micromirros out of phase with the pixes near the edges.

M43 did have a milder problem with the 4x3 ratio, and I believe that they solve this in software.

We will see more software done in all (mirrorless) cameras in the future, that is clear.

The difference is that one is an afterthought (Nex) and the other is a design principle (M43).

Ultimately we care about image quality. And the lens+ correction on MFT seems to work extremely well.

If these "optically correct" lenses from NEX were really so great, I could accept that argument. But they're not.

Whether these M43 users will want a Nex, is unclear.

It's not a war for consumers. For consumers, it' opportunity and choice.

I am shopping for a camera. I am not an MFT or NEX user. I want to buy. Being frustrated at Sony for not producing innovative glass that is PROPORTIONAL to their sensor size, is not because I hate NEX... but because I'd love to buy a NEX with smaller carrying size (lens+camera)

Other thread, they were telling me the shorter register distance is a design ADVANTAGE for making small primes:

http://forums.dpreview.com/forums/readflat.asp?forum=1041&message=40416806

Sensor size dictates lens size. Longest axis.

Sensor size and lens dimensions scale linearly. 25% longer long axis of sensor should force you at worst to have a 25% larger lens in length and width.

Sony's are nowhere close to the lens miniaturization we see in MFT.

The lens dimensions should only increase linearly with the diagonal of the sensor. So we should only have a 25% longer and fatter Sony lens for the same optical properties. Not something close to double.

Of course len+camera carrying size matters the most. But to make an optically identical setup, at worst, sony lens designs should be as big as Samsungs + 6.5 mm of length. Instead, the sony NEX + lens is much bigger. I was pointing out that the WORST, Sony could do is add 6.5 mm of tube to what Samsung makes and end up with optically identical lenses to Samsung's, only 6.5 mm longer. Obviously a "ground-up design" would not involve dead tube.

Sorry, I didnt' make my point clearer.

It stems from the earlier thread I referenced regarding lens design in general. Obviously from a physics standpoint, if you scale a sensor up by a factor Y, an optically equivalent lens can be made that has linear dimensions Y times the original lens.

My question whether removing the considerable minimum register distance of an SLR (removing mirror box) lessened design constraints on lens size (since you're free to putthe closest element closer or the same as you did before), and the answer was:

"YES, because a lens of focal length shorter than the register distance requires a retrofocal design which is larger. So at least in the case of short focal length lenses, the solution set for whatever optic problem you're trying to solve, now allows for all the possible designs that are closer to the sensor AS WELL AS designs that are farther from the sensor)."

It was not clear to anybody in the thread, whether short register distance allows for smaller solutions for other types of lenses (big zooms, macro, for example

I offered that I recognize such thinking could be paranoia. but certainly its common in the business world. Camera companies frequently leave out sotware features in their "consumer model" cameras that are in their "expert model", that could easily be added with a firmware adjustment, in order to make people buy the more expensive "expert model". Hell sometimes they just turn off the GPS chip in the lower model of phone to differentiate the product from teh moer expensive one.

You can have both.

Why not purchase M43 in your case? I am sure that Sony gets that message.

Actually I'm still in the market. And I'm open to NEX still. I'm open to MFT.

I look at MFT, and they trade off sensor quality for size. This is INTRINSIC to being smaller than APS-C, and the question will be "by how much?".

Sony has an APS-C sensor and trade off size for sensor quality. But the INTRINSIC limitation (given similar lens innovation) should be linear to the ration of APS-C sensor size and the MFT sensor size (and actually, maybe BETTER than that because the short register distance allows for more degrees of freedom on lens design, at least for short focal length lenses). So at worst, I want Sony's lens+camera body to be 25% thicker (and the lens to be 25% larger diameter) than the MFT lens.

Again, not because I hate NEX... but because I'd love to buy a NEX with smaller carrying size (lens+camera)

--
Cheers,
Henry

 

[blue_skies]

I believe that this is already common place in most compact (zoom) single-lens cameras today.

An aside:

You've convinced me that part of this is design philosophy of the lenses (maybe an outdated one), maybe because the idea of aggressively correcting PREDICTABLE AND SYSTEMATIC BIAS OR DISTORTION in information collection was not something that was "mainstream" in camera design at the time of the mirrorless revolution.

But it seems that it could be a very interesting discussion today, whether aggressively taking advantage of mathematical corrections allows even more freedom in lens design, If barrel and vignetting, etc. are well corrected with software and less constraining issues, and only sharpness, uniform planar focus, and contrast & size become top points of emphasis, future lens designs can be FAR smaller and maybe far better. We do that type of thing all the time in other fields (telescopy, astrophysics, not just optical, but other EM ), where maximum energy collection and resolving power are most important... and spatial fidelity is secondary (and easily corrected with software).

Maybe MFT "got lucky" and chose the right path regarding portable lens design, and the future is all about designing lenses that lend themselves to best correction, and soon NEX and others will have to get on that bandwagon.

The future may very well be all lens designs doing some significant mathematical correction.

Perhaps even custom calibrating the correction for each lens and storing the info in firmware. If they don't do that already to some degree.

--
Cheers,
Henry

 

[ProfHankD]

As I see it, it is actually fairly easy to make a faster-than-f/2 fixed-focal-length pancake for the NEX in the 20-35mm range. Basically, take a classic double-Gauss "fast fifty" design and scale it down -- it should work fine for NEX (both the coverage and flange distance scale down: FF-> APS-C and 44.5-> 18mm).

...

They are very compact, but 'pancakes' -as discussed here- they are not.

They are when scaled down 25% to 50%. A typical double-Gauss fast fifty (which is most fast fifties that are not f/1.2) is about 1.5" long -- I think less than 1" long qualifies as a pancake, don't you?

FWIW, I agree the Sony 16mm f/2.8 is a very good pancake, especially for the price.

 

[ProfHankD]

Because of this 'angle of light inclination' light fall-off, I presume that all E-mount lenses are retrofocus to some degree. All SLR lenses (e.g. Distagons) are retrofocus and do not exhibit this vignetting (well, very wide vignets, but for other reasons).

Let's use the right word: this is called cos4 "natural vignetting." The light falloff is proportional to the 4th power of the cosine of the ray angle as the light hits the sensor. Microlenses can worsen this effect by changing the incident angle and/or misdirecting rays away from the sensing part of a pixel, which is also how transverse CA can be made worse by microlenses.

The catch is that the ray angle is not directly a function of the position of the rear element of the lens -- it has more to do with the exit pupil, which is different. Lenses designed for longer flange distances tend to have less problematic projection angles, so the cos4 vignetting is minimal. However, the real fix is to design the lens to be telecentric in its rear projection -- with all rays going straight back to the sensor. The E-mount is large enough to permit this for APS-C; it would be very tight for full frame. Other possible fixes include shaded filters, analog gain adjustments, or digital correction.

Long focal length lenses don't have to be retrofocus to produce good incident angles. However, I think it is possible to create a desirable exit pupil even for a wide-angle lens without being retrofocus. It certainly is easier to just use any of the many existing lens designs that were intended for longer mount flange distances....

 

[blue_skies]

Wouldn't work as you envision: in scaling down, assuming the same exit pupil, the light bending has to increase. This changes the lens characteristics. If the exit angles are to be preserved upon scaling, the lens grows outwards.

This is why wide legacy SLR lenses work better than wide RF (non retrofocus) lenses.

Your scaling solution has same problem as wide RF lenses.

Shrink the sensor and it works, e.g. in M43.

As I see it, it is actually fairly easy to make a faster-than-f/2 fixed-focal-length pancake for the NEX in the 20-35mm range. Basically, take a classic double-Gauss "fast fifty" design and scale it down -- it should work fine for NEX (both the coverage and flange distance scale down: FF-> APS-C and 44.5-> 18mm).

...

They are very compact, but 'pancakes' -as discussed here- they are not.

They are when scaled down 25% to 50%. A typical double-Gauss fast fifty (which is most fast fifties that are not f/1.2) is about 1.5" long -- I think less than 1" long qualifies as a pancake, don't you?

FWIW, I agree the Sony 16mm f/2.8 is a very good pancake, especially for the price.

--
Cheers,
Henry

 

[Dennis]

Probably f/2.8 for image quality issues. Most people complain, but then they compare with wide angles that start at f/4.0

David Kilpatrick had written something a while back about the 16, saying that for its specs, it was a pretty impressive little lens. I'm no optics expert, so trust that it was a challenging lens to build.

Doubtful, the 16mm could be a pancake because of its focal length. A 30mm would not be a pancake.

The Samsung 30/2 is compact enough that whether you call it a pancake or not, I don't much care Much longer than 30 or 35mm and I think you have to just decide whether you're shooting for "compact" (as compact as reasonably possible) or throwing in the towel for the sake of IQ. I think there's room for both in a lineup - a handful of compact lenses, plus a number where size is not the main goal. We see compact (pancake like) lenses from 40mm up to 70mm (Pentax 70/2.4) but those are for SLR mounts with significantly longer flange distances. It would be unreasonable to expect longer pancakes for these reduced registration distance ILCs. Maybe 3 pancakes at most, (thinking Sony's 16, a 30, and maybe one more in between) would be the best we could hope for.

I think that having a wide, relatively fast, pancake lens was laudable. It not overlapping with the kit zoom makes sense, and the two adapters for ultra-wide and fish-eye is very clever.

It's a great little lens. (Well, not optically as good as a non-pancake might be, but a very handy addition to the lineup). I've posted before that I just don't like Sony's prioritization.

• Dennis

--
Gallery at http://kingofthebeasts.smugmug.com

 

[blue_skies]

Because of this 'angle of light inclination' light fall-off, I presume that all E-mount lenses are retrofocus to some degree. All SLR lenses (e.g. Distagons) are retrofocus and do not exhibit this vignetting (well, very wide vignets, but for other reasons).

Let's use the right word: this is called cos4 "natural vignetting." The light falloff is proportional to the 4th power of the cosine of the ray angle as the light hits the sensor. Microlenses can worsen this effect by changing the incident angle and/or misdirecting rays away from the sensing part of a pixel, which is also how transverse CA can be made worse by microlenses.

The catch is that the ray angle is not directly a function of the position of the rear element of the lens -- it has more to do with the exit pupil, which is different. Lenses designed for longer flange distances tend to have less problematic projection angles, so the cos4 vignetting is minimal. However, the real fix is to design the lens to be telecentric in its rear projection -- with all rays going straight back to the sensor. The E-mount is large enough to permit this for APS-C; it would be very tight for full frame. Other possible fixes include shaded filters, analog gain adjustments, or digital correction.

Long focal length lenses don't have to be retrofocus to produce good incident angles. However, I think it is possible to create a desirable exit pupil even for a wide-angle lens without being retrofocus. It certainly is easier to just use any of the many existing lens designs that were intended for longer mount flange distances....

The RF cameras emerged in film days, and film likes to be agitated under an angle. Often, center softness was a problem. CMOS sensors don't like this angle, hence the microlenses.

The exit pupil position determines best 'optical' tradeoffs. It appears that wide angle RF lenses are too close for comfort, and thereby it also appears that any 'fast, longer FL' pancake design is compromised, as this exit pupil would have to move too close to the sensor for the lens to be deemed a pancake.

Older SLR designs haven this exit pupil past the critical point, going (much) further away from the sensor. But the lenses can become huge. Long focal lens is ok, but fast wide angle SLR lenses become huge, and expensive.

To stress this again, in a non retrofocus design, the exit pupil to sensor distance is the focal length. Therefore a 35mm lens would already protrude 35 -18 = 17mm. Then add the one inch for the lens groups and the protrusion is already 17 + 24 = 41mm. Add in mounting rings, etc, and we see a 45mm protrusion, not atypical for some of the (non retrofocus) RF lenses at 35mm.

I do not consider those pancakes.

To make a pancake, you have to move the exit pupil much closer to the sensor. That leads to vignetting, distortion and more, and must be remedied. Retrofocus works for a longer lens, as it is telecentric. A short lens is wide, and retrofocus will cause all sorts of distortion - even fisheye, or severe barrel-like. Let alone extreme vignetting.

For all those that keep crying 'give me a pancake, please!' I keep thinking "just get a camera with a smaller sensor".

Technology is advancing rapidly, and software can do a lot, but some appreciation of the limits, tradeoffs and choices made, go a long way, imo.

--
Cheers,
Henry

 

[RussellInCincinnati]

Kwisatch...Is it phyiscally possible for a NEX bright pancake prime?...My guess is Sony limited their lens size by using the APS-C sensor size. So maybe it is futile to keep waiting for a pancake prime like the M43 20mm 1.7 lens for NEX cameras?

Of course it is futile to wait for a lens that is exactly the same equivalent focal length as a smaller-sensor micro four thirds lens, and also has exactly the same price, size, weight, features, image quality. Because when you make a lens for a larger sensor, the lens is going to have to be bigger or more burdensome or sacrifice some desirable aspect (except maybe have more room on the larger lens for controls) in order to create a larger but equally tele-centric (i.e. as nearly perpendicular) image circle.

It might be meaningful to ask a much more specific question like: is it physically possible to make a Nex pancake prime that sells profitably for less than $300, has a speed of F2.0 or faster, has the flare resistance of lens X, the sharpness at various apertures of lens Y, the features including autofocus and stabilization of lenses Z and A, the durability of lens B, the fit and finish of such-and-so lenses, the longitudinal chromatic aberrations in out of focus regions of lens C, etc etc etc.

But such a meaningful, answerable question is not fun to read and answer. And in that very tediousness shows the difficulty of making simple judgements about the extremely complex corporate decisions made by huge, sophisticated entities like Sony. Sony doesn't make a 27mm prime like the micro four thirds 20/1.7 "pancake" because they haven't wanted to yet. Sony's closest answer to the 100-gram, optically rather nice but high-distortion circa $400 dollar Panasonic/u43 20/F1.7 lens is the thousand dollar, over-200-gram, much lower distortion 24mm/F1.8 lens.

There are undoubtedly dozens of reasons why Sony did not make a lens exactly like the Panasonic pancake at some exactly analogous price for the Nex, but am quite sure that one of those reasons is not that Sony is just too dumb to see how smart those micro four thirds lens makers are.

 

[ProfHankD]

Shrink the sensor and it works, e.g. in M43.

We are shrinking the sensor: from full frame to APS-C.

 

[Rehabdoc]

Kwisatch...Is it phyiscally possible for a NEX bright pancake prime?...My guess is Sony limited their lens size by using the APS-C sensor size. So maybe it is futile to keep waiting for a pancake prime like the M43 20mm 1.7 lens for NEX cameras?

I was thinking more along the lines of Samsung's 30 mm f2.0 for it's NX, which last I heard, sported an APS-C sensor size.

Andbecause the NEX is an ILC, you end up with a choice to buy or not buy that lens.

Of course it is futile to wait for a lens that is exactly the same equivalent focal length as a smaller-sensor micro four thirds lens, and also has exactly the same price, size, weight, features, image quality. Because when you make a lens for a larger sensor, the lens is going to have to be bigger or more burdensome or sacrifice some desirable aspect (except maybe have more room on the larger lens for controls) in order to create a larger but equally tele-centric (i.e. as nearly perpendicular) image circle.

Nobody asked for that. We've asked for something with EQUIVALENT characteristics and PROPORTIONAL in size. An APS-C sensor is around 25% larger than an MFT sensor. So it would be nice to have a camera+lens combo that is at most, 25% deeper than the MFT kit of the same specs.

Since BASIC GEOMETRY shows that you CAN simply scale an MFT straight up by 25% and get equivalent optics putting light on an APS-C Sensor.

The APS-C kit doesn't even have to be 25% larger though... because you only have to scale up the lens and the space between the lens and the sensor by 25% (i.e. the sensor is 25% wider and taller, the lens is 25% wider and taller, and the entire length from tip of lens to sensor is 25% longer (But the rest of the camera doesn't have to be scaled up. The materials for the shell of the camera, the thickness of the swivel screen, the handle, the EVF, etc. do NOT have to be scaled up at all. They can remain the same size)

The fact is that the NEX lenses with comparable specs are not only 25% thicker and wider, they're much, much thicker and wider.

It might be meaningful to ask a much more specific question like: is it physically possible to make a Nex pancake prime that sells profitably for less than $300, has a speed of F2.0 or faster, has the flare resistance of lens X, the sharpness at various apertures of lens Y, the features including autofocus and stabilization of lenses Z and A, the durability of lens B, the fit and finish of such-and-so lenses, the longitudinal chromatic aberrations in out of focus regions of lens C, etc etc etc.

It might be more meaningful to ask if it would be physically possible to make lenses that result in lens-body combinations that are as shallow as the Samsung NX cameras? Yes, of course.

Proportionate to MFT cameras? Yes, of course.

But such a meaningful, answerable question is not fun to read and answer. And in that very tediousness shows the difficulty of making simple judgements about the extremely complex corporate decisions made by huge, sophisticated entities like Sony. Sony doesn't make a 27mm prime like the micro four thirds 20/1.7 "pancake" because they haven't wanted to yet. Sony's closest answer to the 100-gram, optically rather nice but high-distortion circa $400 dollar Panasonic/u43 20/F1.7 lens is the thousand dollar, over-200-gram, much lower distortion 24mm/F1.8 lens.

You started with sketchy argument that it wasn't possible to make equivalently small lenses, and now you're just saying "they don't want to yet".

I can buy that. But I think there's a more sensible theory.

There are undoubtedly dozens of reasons why Sony did not make a lens exactly like the Panasonic pancake at some exactly analogous price for the Nex, but am quite sure that one of those reasons is not that Sony is just too dumb to see how smart those micro four thirds lens makers are.

Actually, I think "too dumb" is overly harsh.

I think its more likely that

(1) their marketing (and therefore design philosophy) is built around the idea that "significantly smaller than DSLR is GOOD ENOUGH", and so their lens designs have not been moving in that direction.

(2) philosophic emphasis on "optical purity" over mathematically correction, pointed away from small designs possible with such a short register distance, which is shorter than even MFT.

(3) What's becoming clear, that even though NEX is a huge success, it could still be even more popular if they produced even just a couple of lenses that allowed for what is clearly the "SEXY" carry size of the Panny GX1X, Canon G1X, etc.

People are buying those cameras, at HIGHER prices than the NEX 5N kit... with LOWER image quality and/or performance... simply because Sony doesn't have a "SEXY" lens for carry with it.

Today Pentax launched K01, they know what's up.... they launched a Mirrorless ILC that has a big fat mirror box sized gap inside it... and the first lens they come out with is a razor thin pancake.

NEX, because it's an ILC, can be the exact same system it currently is... AND BLOW UP the MFT market at the same time, by producing just a couple of those kinds of lenses, at which point, the only reason for even considering MFT would be focus speed, since the carry size GAP would be closed by a LOT.

It would certainly easily tip me over to buying a NEX 7 as my main system right now.

 

[RussellInCincinnati]

RussellInCincinnati:Of course it is futile to wait for a lens that is exactly the same equivalent focal length as a smaller-sensor micro four thirds lens...when you make a lens for a larger sensor, the lens is going to have to be bigger or more burdensome or sacrifice some desirable aspect...Sony doesn't make a 27mm prime like the micro four thirds 20/1.7 "pancake" because they haven't wanted to yet.

Rehabdoc:You started with sketchy argument that it wasn't possible to make equivalently small lenses,

I see, you are calling it "sketchy" to argue that an identical-features-performance-price lens for a bigger sensor has to be bigger.

Rehabdoc: and now you're just saying "they don't want to yet".

A larger sensor demands a larger, more expensive lens, if the larger sensor lens is going to have identical features, performance, materials and build quality .

So if Sony's decision-makers dictate that a Nex lens must be exactly as good and cheap and featured as a u43 lens, then physics dictates that the Nex lens has to be bigger .

But what if Sony decision-makers said that the Nex lenses didn't have to be as good and cheap and as featured as a u43 lens? Then despite the physics of the larger sensor, if you let quality/features/price go out the window, heck Sony could have decided to make Nex lenses smaller than u43.

There are so many factors besides size that go into making a valuable, competitive lens, it feels meaningless to pick out one factor like size and talk about it in isolation.

Why not start a thread asking why Sony e-mount lenses aren't made out of the great plastic that so many micro four thirds lenses are made of? After all that great plastic saves weight and is easier to crank out in volume. We don't start such a simplistic thread because we know there are a ton of things that must go into deciding the case materials of a line of lenses, and it's not something we can armchair second guess. Same goes for lens size.

Little 110mm single-coated Tessar enlarging lens at F8 or F11, with great lens hood.

 

[MoreorLess]

Body size affects lens size as well. The Fuji X Pro 1 is, according to DPReview, "quite substantially larger than the Sony Alpha NEX-7". That means it has the luxury of being able to use shorter lenses.

No, flange distance on the fuji is 17.7mm vs 18mm on NEX. Fuji is thicker from LCD and maybe they wanted a beefier grip. Who knows.]

However, a larger aperture pancake on NEX might not be very easy to do. Samsung has a flange distance of more like 25mm and so they can make a small 30mm lens fairly easy. NEX might do the same lens and it might be an extra 5 or 10mm thick or so do to the difference in flange distance. However, most would probably consider such a lens small enough. So I think we will see small lenses for NEX, but maybe not many pancakes other than the 16mm or other ultra wides.

All I think Sony needs to do is hurry up and get some lenses available. They have mostly consumer grade lenses, and they go release a pro grade camera like the NEX 7 with only 1 or 2 pretty good lenses.

Is it as simple as just adding on the difference in flange distance to lens size? I'm no expert but the impression I had was that Sony's problems with large lenses steemed from the combination of a small flange distance, no in camera distorition correction and the AA filter meaning that light would be hitting the sensor at too steep an angle.

 

[ProfHankD]

The RF cameras emerged in film days, and film likes to be agitated under an angle. Often, center softness was a problem. CMOS sensors don't like this angle, hence the microlenses.

I thought the microlenses are there primarily to enhance fill factor. Only a fraction of the area of a pixel is actually light sensitive, and the microlens is supposed to direct light from the entire pixel area there. Because metal tends to go in top layers of an IC (casting shadows), grinding down the substrate and having the sensor rear lit increases fill factor, but it's still not very close to 100%. Of course, the Foveon stacked pixels are deeper, so they'll have more angle-shading trouble.

Microlenses generally are designed assuming light comes straight toward the sensor. At angles, they misdirect light so it doesn't go to the sensitive spot, thus causing increases in vignetting and CA.

The thicker emulsions (that light can bounce around in), log sensitivity, and various other factors do make many films less blatantly show incident angle problems, but I don't think the differences between film and sensor are the major issue here. In fact, Leica's M8 improved sensor performance quite a lot by positioning the microlenses to correct for the expected ray angles. Correcting the microlenses, or getting close to 100% fill factor (which film has), will make this problem go away.

Unfortunately, cos4 vignetting is a lens thing, and it happens with film too -- as many lenses make painfully obvious with slide film. A cos2 factor comes from inverse square loss for a longer ray, another (approximately) cos factor comes from angular distortion of the apparent exit pupil shape, and the last cos factor comes from spreading of each ray over a larger area at the sensor. Thus, cos4 loss will not go away, but the impact can be reduced by analog and/or digital means.

...

For all those that keep crying 'give me a pancake, please!' I keep thinking "just get a camera with a smaller sensor".

Well, that certainly works. However, all the problems have to do with angles, and angle-wise a 30mm on APS-C is a lot like a 50mm on a full frame sensor. I think a very fast 30mm thin enough to call a pancake is absolutely doable with corner performance as good as a "fast fifty" on full frame.

Technology is advancing rapidly, and software can do a lot, but some appreciation of the limits, tradeoffs and choices made, go a long way, imo.

Also agreed. However, it looks to me like the NEX-5N may be doing some analog gain tweaks to fix vignetting (after all, the vignetting fixes show-up in RAW, and Sony started doing analog noise reduction in their sensors years ago). That alone changes the tradeoffs significantly....

 

[Promit]

I think the comparisons to m4/3 are disingenuous at best. I have yet to see why a version of the Samsung NX 30mm pancake lens couldn't be created for NEX by simply cloning the optical formula and adding the flange distance.

 

[blue_skies]

I think the comparisons to m4/3 are disingenuous at best. I have yet to see why a version of the Samsung NX 30mm pancake lens couldn't be created for NEX by simply cloning the optical formula and adding the flange distance.

And adding necessary software correction? But you are correct - this should be feasible.

The Samsung NX flange distance is also higher (see kit zoom pic below), and, I am guessing, when mounted, the EX-S30NB extends a little further out than the SEL16F28 does on the Sony Nex.

I'd like to see this lens mounted on NX side by side to the SEL30M35 on the Sony and measure the exact difference. I think that we are talking about something like only 0.5-1" perhaps.

Photozone likes the 30mm lens quality:

• http://www.photozone.de/samsungnx/696_samsungnx30f2

It is Samsung NX' smallest lens, and is in fact smaller than their 20mm lens:



Fwiw, the Samsung 1855 kit zoom lens for the NX is in fact LARGER than the Sony Nex 1855 kit zoom. This probably hints of the fact that both lenses are ok to use WITHOUT optical software correction. It is also larger than the Panasonic GX1 with the 14-42mm zoom lens.







--
Cheers,
Henry