Has the IQ160 microlenses or not

[eronald]

Doug,

 I guess the real question is "can back XX be shifted" and in fact there's a variety of design issues which intervene here; paradoxically, I would guess microlenses will mitigate not only the sin4 law but also cell design assymetries, and also interact with crosstalk issues,  and so the answer is not trivial, it may even vary with the shift orientation, and in fact I would expect the P30 and P45 to behave differently with respect to centerfold issues at a shift of ... 0. My naive expectation would be to see less vignetting on the P30.

 At some point, I guess someone should make a big table of all back models and their known features; with the proliferation of small editions of backs, it's practically impossible for most of us to remember these trivia. I am already geriatric, but of course nothing is forgotten down in the subbasements of the netherworld  


Edmund

That's because I tend to focus on what a person is actually asking, rather than the way they've phrased it.

He's asking why he's seeing better ISO performance on the IQ160 than P40+.

That leaves two (potentially overlapping) possibilities:
- He's an engineering-oriented nerd* and primarily is intellectually curious about the disparate ISO results and wants to understand the physics of microlens designs and underlying electronics that explain the effect he is seeing
- He's a results-oriented photographer mostly wanting to know why his new camera works better and is grasping at microlenses as one possibility, but in reality isn't that interested in the physical architecture of the silicon at the heart of his camera

I tend to assume even nerdy photographers, at the end of the day, are interested in technical minutia as a means to an end, and not an intellectual end unto itself. In other words they want to know if a camera is USB2 or USB3 because they are interested in a camera which tethers fast. In answering questions about USB3 on an IQ back I'm likely to explain that because P1 has a better processing pipeline, ultra fast processors and coprocessors, and excellent raw compression, and was built with tethering as a core function rather than an add-on feature, it's able to tether with consistently fast speed despite having resolutions much higher than a dSLR. The fact that USB3 is a factor in this is intellectually interesting, but at the end of the day the person is asking "how fast does it tether?" rather than "does it adhere to an international data transfer protocol known as USB3?".

Anyway, in direct response... Most modern-era digital backs have some form of lens in front of each pixel. Generally I refer to backs with "strong microlens" meaning that the microlens significantly increases the effective ISO over similar sensors and reduces the real-world ability to use the back on a tech camera vs similar sensors. For instance the P30+ was a "strong microlens" back in that it had a sensor which was generationally similar to the P45+ but had an aggressive microlens design which severely limited it's use on a tech camera and increased it's native ISO to ISO100 (in comparison the P45+ was a base ISO50 back). In this regard the IQ160 is not a "strong microlens" back. But yes it does technically have microlenses in front of the sensor well. More to the point the presence/type of microlens on the sensors of the IQ160 and P40+ is not the reason why the OP is noticing a difference in high ISO performance between these two backs.

Again the above answer is informed by my assumption that most people who want to know about microlenses are moreso asking about practical things like "can I use this on a tech camera, and if so with what limitations?" or "how good is this camera at ISO X" vs "Can you inform me about the last 10 years of sensor architecture, including varying generations of the types, shapes, and material used to create a microlens which gathers light and handles inbound angles of light with varying strategies and levels of success."

*no slight intended - I'm proudly a nerd

[ErikKaffehr]

Hi,

It seems that the IQ-160 and the IQ-260 work well with shifts (according to my reading) although they have microlenses.

My take is that microlenses are probably a good thing, mostly. They increase quantum efficiency, reduce aliasing (by sampling larger area) and also probably reduce cross talk. But they probably don't work well with shifts on sensors with small pitch. Obviously not all microlenses are created equal either.

Best regards
Erik

Doug,

 I guess the real question is "can back XX be shifted" and in fact there's a variety of design issues which intervene here; paradoxically, I would guess microlenses will mitigate not only the sin4 law but also cell design assymetries, and also interact with crosstalk issues,  and so the answer is not trivial, it may even vary with the shift orientation, and in fact I would expect the P30 and P45 to behave differently with respect to centerfold issues at a shift of ... 0. My naive expectation would be to see less vignetting on the P30.

 At some point, I guess someone should make a big table of all back models and their known features; with the proliferation of small editions of backs, it's practically impossible for most of us to remember these trivia. I am already geriatric, but of course nothing is forgotten down in the subbasements of the netherworld  


Edmund

[torger]

Yes they work well with shifts, assuming you have the Digaron retrofocus lenses. The don't work well in shift with SK28 and SK35, you can still use them if you don't shift that much though. The IQ180/IQ280 with it's smaller pixel size get a little bit larger issues, and will not produce that good results extreme-shifted even on Digaron-W wides.

We need to drop the idea that "no microlenses = good for tech cams", "with microlenses = bad for tech cams", as it's more complex than that. Each model needs to be tested what it can do with what lenses. Microlenses today do not necessarily look and work the same way as Kodak/Dalsa microlenses did.

I don't think we're going to see any sensors without microlenses anytime soon, but we can still get good tech wide angle performance. Probably the Sony CMOS would have been quite good if it just had light shields between pixels but it was not designed with that. Anyway the microlenses on the Sony CMOS don't produce any ripple artifacts from what I have seen, and I think they don't worsen crosstalk really, the crosstalk issue is more about lack of shielding between pixels. If you don't need light shields you rather don't manufacture with them as it makes the manufacturing more complex and you might reduce photo diode coverage, so there's no surprise the Sony CMOS don't have any specific shielding, it would only be tech cams that would benefit.

Backside illuminated sensors (which are getting larger and larger) put the photo diode much nearer the surface and increases sensitivity, and as a side effect greatly increases the angular response and reduces the need of pixel shielding. I would guess that the introduction of BSI techniques to MF sensors will be the first time we'll see an improvement again of tech wide compatibility. It's probably two or three generations away though.

[eronald]

there's no surprise the Sony CMOS don't have any specific shielding, it would only be tech cams that would benefit.

Why so? compact cameras are mirrorless, it's useful to be able to place a back element closer to the sensor instead of a projector further away, as Leica shows this allows nice wides, and also probably more compact camera designs. What is so good about retrofocus inherently?

Edmund

[torger]

Why so? compact cameras are mirrorless, it's useful to be able to place a back element closer to the sensor instead of a projector further away, as Leica shows this allows nice wides, and also probably more compact camera designs. What is so good about retrofocus inherently?

I'm not optical design expert by I think retrofocus makes it possible to have less vignetting and wider aperture. Symmetrical wides get small aperture and requires center filters, which is okay in tripod-mounted deep dof photography but not for many other styles, and is thus not so suitable for a camera designed for general purpose photography.

Additionally the shifting property of technical cameras get the wide angle pulled into an extreme. Most (all?) of the tech camera lenses work well already today with the MF Sony CMOS sensor in the center position, so one could probably make much less retrofocus wide angles to say the A7r today if someone would like to.

[eronald]

I'm not optical design expert by I think retrofocus makes it possible to have less vignetting and wider aperture. Symmetrical wides get small aperture and requires center filters, which is okay in tripod-mounted deep dof photography but not for many other styles, and is thus not so suitable for a camera designed for general purpose photography.

Additionally the shifting property of technical cameras get the wide angle pulled into an extreme. Most (all?) of the tech camera lenses work well already today with the MF Sony CMOS sensor in the center position, so one could probably make much less retrofocus wide angles to say the A7r today if someone would like to.

Leica seems to prove one can do nice efficient and *discrete* 35mm non-retrofocus designs. I haven't heard their lenses described as second rate yet, although I'm sure some Phase One cheerleader is going to pop-up in a minute . There was a very nice Hasselblad compact SWC, and Rollei TLRs. And then there were the beautiful compact Mamiya 7 series, the Xpan etc.

I do understand that big, heavy cameras are "professional". But I'm getting old for weight-lifing, and wonder why we cannot have some of the advantages of old designs reincarnated in modern digital form.

Edmund

[ErikKaffehr]

Hi,

A good point, but keep in mind that Leica has gone to some extreme measures to get i working

- They developed sensor with shallower well design
- They use very thin IR filter ( 1 mm ), most DSLRs use 2mm and 4/3 4mm, MFD, I don't know
- They use stretchcoded information from the lenses to do LLC in firmware

MFD vendors could also do this, of course.

Best regards
Erik

Leica seems to prove one can do nice efficient and *discrete* 35mm non-retrofocus designs. I haven't heard their lenses described as second rate yet, although I'm sure some Phase One cheerleader is going to pop-up in a minute . There was a very nice Hasselblad compact SWC, and Rollei TLRs. And then there were the beautiful compact Mamiya 7 series, the Xpan etc.

I do understand that big, heavy cameras are "professional". But I'm getting old for weight-lifing, and wonder why we cannot have some of the advantages of old designs reincarnated in modern digital form.

Edmund

[torger]

Cool, I did not know about that Leica stuff. Hopefully it will spread and even Sony realizes that there's gain from wide angular response. Pixel shielding is just a "hack fix" you get vignetting instead of crosstalk and with huge DR you can take it, but the real solution is to make a shallow pixel well of course.

Symmetrical lenses are indeed simpler designs so you could make cheaper lighter lenses (in the case of Leica just lighter of course). Possibly folding compact cameras could gain from this type of design, and possibly some already are as the BSI sensors exists in the small sizes.