Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: voidshatter on April 01, 2017, 12:48:47 am
-
Link (http://"http://photorumors.com/2017/03/31/sony-to-announce-new-100mp-and-150mp-large-format-bsi-sensors-in-2018/")
First a quick reminder - in some parts of the world it's already April 1st, so in the next 24 hours take anything you read online with a grain of salt.
There is a video interview on YouTube with Sony representatives that talks about new 100MP and 150MP medium format BSI sensors that are supposed to be introduced in 2018
Didn't watch the video carefully and seriously, but there are several sensor part numbers listed:
IMX161: we already know that this is the 44x33 50MP one in the IQ250
IMX211: likely to be the 54x40 100MP one in the IQ3100
IMX461: likely to be a 44x33 100MP BSI?
IMX411: likely to be a 54x40 150MP BSI?
-
Nice, going to have to start saving up again!!
-
Link (http://"http://photorumors.com/2017/03/31/sony-to-announce-new-100mp-and-150mp-large-format-bsi-sensors-in-2018/")
Didn't watch the video carefully and seriously, but there are several sensor part numbers listed:
IMX161: we already know that this is the 44x33 50MP one in the IQ250
IMX211: likely to be the 54x40 100MP one in the IQ3100
IMX461: likely to be a 44x33 100MP BSI?
IMX411: likely to be a 54x40 150MP BSI?
The real question is, with a Sony BSI sensor work better on a tech camera with wide angles?
-
Chris Barret might know, he works with a Sony a7RII (BSI 42MP sensor) and uses movements a lot!
-
The real question is, with a Sony BSI sensor work better on a tech camera with wide angles?
It's not the reason they make it BSI. Kodak had tech wides in the equation, but I don't think it's been there since. The mirrorless cameras has similar needs, but the wide aperture lenses in those systems still don't pose the same challanges as tech lenses. As increasing wide angle response means sacrificing some other property (like megapixel count and sensitivity) it's unlikely that we'll see sensors that have more wide angle response than the sensor designer think they need.
A small hardware improvement with improved post-processing cleanup algorithms could make a lots of difference though. Due to the overall high performance of the modern sensor you can beat up the file pretty bad with incompatible optics and still get a nice result in the end.
The greatest enemy with current technology regarding wide angle response is the ever-increasing pixel count though, as it means smaller pixel pitch which means higher stack height/pixel pitch ratio, which means worse angular response. So even if BSI improves it, the advantage can be canceled out with smaller size pixels. The a7RII BSI sensor did much, but it didn't really deliver on angular response.
A full-frame 60MP BSI CMOS sensor with current technology would probably perform rather well, but I don't see that we'll see any "low" MP count MF sensors coming out, they must provide a large MP jump from 135 to get the sales.
Anyway we need to test new sensors, and hope Capture One engineers do magic with their LCC algorithms, then it can turn out quite well even if the sensor designers never considered tech lenses. I think tech cams are legacy now, they just don't know it yet.
-
Hi,
The short answer is no. The long answer is, well, it is longer…
Chris Barret uses the A7rII with retrofocus lenses, like Hasselblad V and Canon TSE. So Chris Barret's experience tells no story about BSI sensors with symmetrical lenses.
But, the Fuji GFX does not use gapless microlenses. It may be that they have added some shielding between pixels. That would increase aliasing and reduce quantum efficiency, but it may allow for larger beam angles while reducing cross talk. That would allow for more compact lens designs and a bit more shift.
Sensor design is a compromise and sensor designs are about the most difficult analogue designs. A lot of development work going into designing a single pixel… Consider this: A CMOS sensor takes about three months to make… you start making the sensor at Christmas and it is good for delivery at Easter…
Best regards
Erik
The real question is, with a Sony BSI sensor work better on a tech camera with wide angles?
-
I think tech cams are legacy now, they just don't know it yet.
I agree with this. Schneider has already decided to pull out of this business, and I think they are clever.
Leica survived the obsoleteness of rangefinder by luxury marketing, while Nikon survived the obsoleteness of rangefinder by DSLR. Maybe Nikon will survive the obsoleteness of DSLR by mirrorless at a later time, but it remains to be seen whether Alpa will survive the obsoleteness of techcam by luxury marketing.
-
My wallet is already in pre-emptive shock over the IQ3150... :o
-
It seems to me not very economical to make two MF format sensors.
It is already a small market.
So these MF sensors will cost even more than needed.
Why not just stick to one 54 x 40 format- a clearly different format and quality than FF.
-
It seems to me not very economical to make two MF format sensors.
It is already a small market.
So these MF sensors will cost even more than needed.
Why not just stick to one 54 x 40 format- a clearly different format and quality than FF.
If it would be one format, it would be 44x33. It's much cheaper to manufacture than 54x40 thanks to the smaller area, a much larger market and manufacturers believe in it. Fuji could have made a 54x40 mirrorless if they had wanted, but why make a camera that so few can afford, when you can sell lots more of a more affordable camera. I think medium format is no longer a specialist niche, with the recent 44x33 cameras it's almost like a real mass market product.
In the longer term I think 44x33 can get a big chunk off the current high res 135 segment. High res 135 with optics is already expensive, and the step up to a little bit better is not too large. Increased volume can drive down costs further. It looks good. Medium format is becoming mainstream.
I think that the talk about that the larger format gives a distinctive look tied to the size itself has almost disappeared. It's about as relevant as the CMOS vs CCD wars, that is almost noone cares any longer. It's just about how many sharp pixels you get, and sure the 54x40 paired with great optics is a significant step up, but well, while the 44x33 systems are cheap enough to present a reasonable attractive price/performance equation, the 54x40 systems are for those where that equation is not important, so we should be thankful that Sony cares to make them, as I'm not sure it makes much business sense for such a large corporation.
-
One can dream!
-
it appears those that know more than I do don't believe BSI wouldn't help much with tech lenses and lens cast, but what about cross talk? LCC's perform pretty well with lens cast, not as effective trying to resolve cross talk problems. Is there still a "well" with sides so light can leak through to neighboring pixels?
-
The same info is official from Sony semiconductors:
http://www.sony-semicon.co.jp/products_en/news/detail/170301.pdf
-
it appears those that know more than I do don't believe BSI wouldn't help much with tech lenses and lens cast, but what about cross talk? LCC's perform pretty well with lens cast, not as effective trying to resolve cross talk problems. Is there still a "well" with sides so light can leak through to neighboring pixels?
It's not a well in the same sense, but there's quite a buildup of the color filter and micro lens. See this image:
(http://www.embedded-vision.com/sites/default/files/technical-articles/EDNImageSensors/Figure2c.jpg)
the slanted light rays shown in the image are at a much higher angle (closer to perpendicular) than a tech lens presents at the edges. I don't have the exact numbers, but I think it can be as low as 30 degrees.
If light comes in at such a low angle there can be crosstalk already in the microlens/color filter layer. The solution to the problem is to make the pixels larger (so the color filter/microlens becomes relatively thinner) and reduce the aperture of the photo diode (=add light shields), think of laying a thin opaque layer on top of the pixels (but below the color filter) and making a small hole for each pixel. Kodak did that, and removed the microlens completely (which has the problem that it may do more harm than good at low angles). This does not come without cost though.
Smaller aperture of the pixel reduces quantum efficiency, it's basically the same thing as taking a 135 format sensor and spreading out the pixels and not register the light in-between, and as there's gaps between pixels it produces more aliasing. I don't think people would complain about aliasing, somehow medium format users never do, but the reduced quantum efficiency would show up as reduced dynamic range (and worse ISO performance), and that people would notice in their crazy shadow push tests.
With BSI technology the disadvantages could be made much smaller than it was for the Kodak sensors though, so it would still perform very well. I don't think it's necessary to remove the micro lenses for example, but you would probably make a less aggressive design, and the light shield could be narrower thanks to BSI.
Sony won't design a sensor for legacy lenses though. In theory new mirrorless systems would gain from using symmetrical wides as they can be light and compact and at the same time very high performance (and color cast could be corrected in firmware in those systems), but the problem is that these lenses also by necessity have huge vignetting and small aperture, which is okay for technical type of photography, but not for the all-around photography consumers would expect to get from a modern camera.
Angular response is not something that just happens to become what it becomes, it's designed to meet a specific target. To meet that target you need to compromise other performance aspects. When BSI allows you to get more angular response "for free", the likely approach is to increase other aspects of sensor performance, rather than increasing angular response, unless there is a strong demand to do so. Tech cam lens manufactures are however far better at discontinuing stuff than putting pressure on sensor manufacturers, so I don't think it looks very good.
-
Smaller aperture of the pixel reduces quantum efficiency, it's basically the same thing as taking a 135 format sensor and spreading out the pixels and not register the light in-between, and as there's gaps between pixels it produces more aliasing.
Very interesting, thank you Torger. I can see why a smaller pixel aperture could reduce effective QE hence SNR but I am curious about the second part of the comment above:
why would a smaller aperture with pixels at the same pitch result in more aliasing? Pitch stays the same therefore so should Nyquist.
In addition the low pass action provided by the smaller aperture is reduced so higher frequencies are less attenuated - all good news for the frequency domain, no?
Jack
-
Very interesting, thank you Torger. I can see why a smaller pixel aperture could reduce effective QE hence SNR but I am curious about the second part of the comment above:
why would a smaller aperture with pixels at the same pitch result in more aliasing? Pitch stays the same therefore so should Nyquist.
In addition the low pass action provided by the smaller aperture is reduced so higher frequencies are less attenuated - all good news for the frequency domain, no?
Jack
The reason is that there are gaps between pixels that are not recorded. When looking at the basic math around aliasing it's assumed that there are no gaps, which is the "normal" way to sample things like audio etc. The thing here is that we have <sample> <gap> <sample> <gap> <sample>, and where the gap is there could be a different color, so we get a different kind of aliasing. With gapless pixels that color would be averaged in with the rest so you get less aliasing, but possibly also less "crispy" image as it's more averaged, so I would not be surprised if many users would consider the effect to be an advantage.
-
With respect to colour caste, when I bought the A7R2 (I was using the A7R), I tested as many wide RF lenses I could get my hands on at the time, and documented info from a few other posts. Not one of them showed any corner colour caste. Unfortunately as we all know the BSI sensor did not improve the corner smearing from the lenses. I checked the following lenses;
12 and 15mm VC (version 1), I also checked the 15mm VC III which is optically excellent in the corners as it was redesigned, 18mm ZM, 24mm Leica Elmar, 28/1.9 VC, 28/2 VC, 35/2 Canon LTM
and got info from others from these lenses;
18mm Leica SEM, 28/2.8 Leica Asph
-
The reason is that there are gaps between pixels that are not recorded. When looking at the basic math around aliasing it's assumed that there are no gaps, which is the "normal" way to sample things like audio etc. The thing here is that we have <sample> <gap> <sample> <gap> <sample>, and where the gap is there could be a different color, so we get a different kind of aliasing. With gapless pixels that color would be averaged in with the rest so you get less aliasing, but possibly also less "crispy" image as it's more averaged, so I would not be surprised if many users would consider the effect to be an advantage.
Hi Torger,
1) <sample> <gap> <sample> <gap> <sample>
is called sparse sampling and applies for instance to any situation when fill factor is not 100% (or to the single channel of a Bayer sensor). It should not affect Nyquist as long as sampling (pixel) pitch is held constant. In other words
2) <----sample----> <----sample-----> <----sam....
has the same sampling pitch/frequency and aliasing potential as 1). In fact, because 1) integrates over a smaller area it is able to retain more higher frequencies. Initially counterintuitive until one looks at the math, huh? Some background here (http://www.strollswithmydog.com/resolution-model-digital-cameras-iii/).
Jack
-
Hi Torger,
1) <sample> <gap> <sample> <gap> <sample>
is called sparse sampling and applies for instance to any situation when fill factor is not 100% (or to the single channel of a Bayer sensor). It should not affect Nyquist as long as sampling (pixel) pitch is held constant. In other words
2) <----sample----> <----sample-----> <----sam....
has the same sampling pitch/frequency and aliasing potential as 1). In fact, because 1) integrates over a smaller area it is able to retain more higher frequencies. Initially counterintuitive until one looks at the math, huh? Some background here (http://www.strollswithmydog.com/resolution-model-digital-cameras-iii/).
Jack
You know the math better than me. Anyway here's an illutstrative example from this page:
https://recordingsofnature.wordpress.com/2014/08/26/effect-of-pixel-fill-factor-on-aliasing-and-moire/
100% fill factor (reference image)
(https://recordingsofnature.files.wordpress.com/2014/08/fill08result_no_bayer77555.png)
50% fill factor
(https://recordingsofnature.files.wordpress.com/2014/08/fill06result_no_bayer3441.png)
which looks more crisp but also jaggied (aliased).
The difference is not very large though, so probably we shouldn't worry that much about that aspect. The loss of sensitivity is a bigger issue that people will actually notice in their tests. I doubt that it matters much in real-world photography though because DR is already very good, but real-world photography is not what sells cameras ;-)
-
Ah, I see. Nyquist stays the same at 0.5 cycles/pixel in both cases but in the second case the higher frequencies are less attenuated because of the smaller pixel aperture. The result for the lower fill factor pixel is therefore better resolution at MTF50 but also more energy in the higher frequencies above Nyquist - therefore more aliasing potential. That appears to be a simulation of just pixel aperture and as such an extreme example (in reality lens blur, diffraction etc. would make the effect much less obvious) but I see what you mean now.
Jack
-
Hi,
The corner smearing comes from the cover glass. Light passes trough the outlet pupil and each pixel is essentially illuminated by the entire outlet pupil. So if the cone of light coming from at a large set of angles it will shifted when passing trough the glass in many different ways, so this smear the image and do it differently in different directions.
That is the reason Leica tries to keep the cover glass thin. Ideally, the cover glass should be a part of the optical calculation. It is possible to replace the cover glass on the A7xx sensors with a thinner one, but that would induce errors in lenses that are calculated for the 2.5 mm (or so) cover glass the A7xx usually has.
Best regards
Erik
With respect to colour caste, when I bought the A7R2 (I was using the A7R), I tested as many wide RF lenses I could get my hands on at the time, and documented info from a few other posts. Not one of them showed any corner colour caste. Unfortunately as we all know the BSI sensor did not improve the corner smearing from the lenses. I checked the following lenses;
12 and 15mm VC (version 1), I also checked the 15mm VC III which is optically excellent in the corners as it was redesigned, 18mm ZM, 24mm Leica Elmar, 28/1.9 VC, 28/2 VC, 35/2 Canon LTM
and got info from others from these lenses;
18mm Leica SEM, 28/2.8 Leica Asph
-
Hi,
The corner smearing comes from the cover glass. Light passes trough the outlet pupil and each pixel is essentially illuminated by the entire outlet pupil. So if the cone of light coming from at a large set of angles it will shifted when passing trough the glass in many different ways, so this smear the image and do it differently in different directions.
That is the reason Leica tries to keep the cover glass thin. Ideally, the cover glass should be a part of the optical calculation. It is possible to replace the cover glass on the A7xx sensors with a thinner one, but that would induce errors in lenses that are calculated for the 2.5 mm (or so) cover glass the A7xx usually has.
Best regards
Erik
Yes, that's why the BSI had no effect on the smearing. Kolari offer this service and are now testing out a new version that looks very promising but as you say it will likely effect very good native WA lenses like the 21mm Loxia or 25mm Batis for example. The other option is to add an element to the front of the lens to compensate. Not as good as a lens design for the coverglass but some people are getting good results.
http://www.fredmiranda.com/forum/topic/1453834
-
Ah, I see. Nyquist stays the same at 0.5 cycles/pixel in both cases but in the second case the higher frequencies are less attenuated because of the smaller pixel aperture. The result for the lower fill factor pixel is therefore better resolution at MTF50 but also more energy in the higher frequencies above Nyquist - therefore more aliasing potential. That appears to be a simulation of just pixel aperture and as such an extreme example (in reality lens blur, diffraction etc. would make the effect much less obvious) but I see what you mean now.
Jack
Now we understand why the 22mp Kodak backs had a rep for sharpness AND aliasing
-
Ah, I see. Nyquist stays the same at 0.5 cycles/pixel in both cases but in the second case the higher frequencies are less attenuated because of the smaller pixel aperture. The result for the lower fill factor pixel is therefore better resolution at MTF50 but also more energy in the higher frequencies above Nyquist - therefore more aliasing potential. That appears to be a simulation of just pixel aperture and as such an extreme example (in reality lens blur, diffraction etc. would make the effect much less obvious) but I see what you mean now.
Jack
The aliasing that photographers note is due to the Bayer filter. It is quite noticeable, because color stripes are obvious to us humans.
The aliasing that this study is talking about is different and would appear on a monochrome sensor. Its artefacts are rarely noted by photographers.
Please note that a pixel fill of more than 100% is also possible (on multishot shifting sensors).
-
Now we understand why the 22mp Kodak backs had a rep for sharpness AND aliasing
Also those backs were common during the era of Capture One 3 and other early raw processing software. In the decade+ since those days Capture One has evolved tremendously. The same raw files that showed ghastly aliasing now often show little or no aliasing.
In so many discussions, tests, research papers etc the element of raw processing is simply forgotten. That's a shame since raw processing algorithms is a huge component of image quality and not all raw processors produce equal quality.
-
I think we came to the conclusion that the fill factor effect does have an effect on aliasing, but it's a small one, maybe even negligible in practical photography. The advantage manufacturers mention with large fill factor is better sensitivity, that is better ISO.
If Sony would make a light-shielded version of their sensors to meet the need of tech-style wide angle designs, the disadvantages would be 1) probably less megapixels (think 60MP fullframe), 2) somewhat less dynamic range, 3) lower sensitivity (that is worse ISO performance).
Considering the base performance of the current sensor there certainly is plenty of margin to make such a thing. However, it's a great risk to take even a small step down in those performance numbers, it's not what consumers want to see, especially when the only gain is being able to use niche optics that doesn't exactly look like they have their future in front of them. So it won't happen.
The only chance for these already-or-soon-to-be legacy wide lenses is that there is some major technology shift that gives wide angular response for free so it's just there without having to design for it, and BSI in its current form is not it, unfortunately.
-
I'm surprised we haven't yet seen spectacular shift lenses in the medium format world as we are moving away from tech cam friendly sensors. Leica promised a wide PC lens a while back and it never materialized for the S system. Hasselblad has the HTS...but crop factor takes away the true wide angle bit. Especially for cameras like the GXF or even the Pentax. It would only make sense to have an excellent shift lens. I would think they would be decent sellers as well.
-
The only chance for these already-or-soon-to-be legacy wide lenses is that there is some major technology shift that gives wide angular response for free so it's just there without having to design for it, and BSI in its current form is not it, unfortunately.
Rodenstock has been asleep at the wheel for years it seems, but I would think that if the tech camera lens business still means something for them they may be somehow in the prcess of designing better wides able to limit downsides with the CMOS sensors?
Zeiss, Sigma, Nikon, Fuji,... have all been demonstrating the tremendous progress made in lenses design these past 3-4 years, to the extend that tech camera lenses are falling behind although they are much more costly and have huge intrinsic architecture advantages, I would hope that the engineers at Rodenstock have some pride left.
Cheers,
Bernard
-
The only chance for these already-or-soon-to-be legacy wide lenses is that there is some major technology shift that gives wide angular response for free so it's just there without having to design for it, and BSI in its current form is not it, unfortunately.
I was told by an engineer in Sony's sensor department that the new 150MP BSI sensor (IMX411) will likely have less color cast and less heat dissipation than the existing 100MP FSI sensor (IMX211) does, despite the 50% increased pixel count.
Rodenstock has been asleep at the wheel for years it seems, but I would think that if the tech camera lens business still means something for them they may be somehow in the prcess of designing better wides able to limit downsides with the CMOS sensors?
Zeiss, Sigma, Nikon, Fuji,... have all been demonstrating the tremendous progress made in lenses design these past 3-4 years, to the extend that tech camera lenses are falling behind although they are much more costly and have huge intrinsic architecture advantages, I would hope that the engineers at Rodenstock have some pride left.
I believe the main challenge is the increasing weight which is not good for Copal 0 shutters. The 32HR is already known to be delicate enough. Maybe they should just drop Copal 0 shutters and only install aperture control, for upcoming digital backs where electronic shutter is standard. They must have some very close relationship with Sony or whoever is the next sensor giant to collaborate the design, otherwise the revenue may not justify the cost of new designs.
-
Rodenstock has been asleep at the wheel for years . . .
Zeiss, Sigma, Nikon, Fuji, have all been demonstrating the tremendous progress made in lenses design these past 3-4 years . . .
tech camera lenses are falling behind although they are much more costly and have huge intrinsic architecture advantages
This epitomizes the challenges for formats larger than 36x24mm: the intrinsic advantages of a larger format are battling the far greater resources and faster technological progress driven by the far higher sales volume and revenues for the smaller formats of the "mainstream" DSLR system from Canon, Nikon and Sony. More so when lenses and some other technology also get shared with even smaller formats like "APS-C", further increasing the revenue potential of an investment in improved designs. In addition to those lens comments, note that the latest innovations in Sony sensors have arrived in 36x24mm format but not yet in the larger formats. It is not as bad though as when the "plus-sized formats" were stuck with CCDs!
This is why I wonder if the inherent performance advantage of the modest size advantage of 44x33mm over 36x24mm (about "half to one a stop") is enough. But at worst, if 44x33 fails then I expect the larger formats like 54x40 to offer a clear IQ advantage over 36x24 and so a market niche will persist for at least one plus-sized format.
-
Rodenstock has been asleep at the wheel for years it seems, but I would think that if the tech camera lens business still means something for them they may be somehow in the prcess of designing better wides able to limit downsides with the CMOS sensors?
Disagree. The best lenses in MF digital right now are Rodenstock ones.
Nearly everybody agrees the 90mm HRSW is the finest MF lens out there, bar none. Large image circle (100mm, good for a 4 way stitch) and sharp beyond anything you have seen in MF. I hear good things about the latest design wide ones, though they are not of interest to me.
Remember production numbers for MF are very low, and if an Otus for FF Canon/Nikon is $4500, then for the reduced production quantity for MF - larger element size, smaller numbers and you're talking about a $20,000 lens... and for what? - the 90mm is astonishingly good, even the old HR series (60f4, 100f4, 35f4, etc) are still excellent, (but no movements).
I'd like to see Zeiss come back into MF too, but it's simply not going to happen as production numbers are too low. The only hope is that the new Fuji takes off, and others (Sony, Phase) come into the MF mirrorless, and dramatically increase the market place for such lenses.
-
Excuse the dumb question, but what happens if one uses eg the Canon TSE wides on MF? Is the shift ability negligible?
Edmund
Disagree. The best lenses in MF digital right now are Rodenstock ones.
Nearly everybody agrees the 90mm HRSW is the finest MF lens out there, bar none. Large image circle (100mm, good for a 4 way stitch) and sharp beyond anything you have seen in MF. I hear good things about the latest design wide ones, though they are not of interest to me.
Remember production numbers for MF are very low, and if an Otus for FF Canon/Nikon is $4500, then for the reduced production quantity for MF - larger element size, smaller numbers and you're talking about a $20,000 lens... and for what? - the 90mm is astonishingly good, even the old HR series (60f4, 100f4, 35f4, etc) are still excellent, (but no movements).
I'd like to see Zeiss come back into MF too, but it's simply not going to happen as production numbers are too low. The only hope is that the new Fuji takes off, and others (Sony, Phase) come into the MF mirrorless, and dramatically increase the market place for such lenses.
-
I believe the main challenge is the increasing weight which is not good for Copal 0 shutters. The 32HR is already known to be delicate enough.
This always baffled me that they decided to design the lens for a Copal 0. A Copal 1 is only slightly larger (could probably fit on most MF lens boards) and the max speed is 1/400 instead of 1/500. For all intents and purposes, using the stronger Copal 1 would have been the better decision given the weight of the lens.
-
Official Sony document on this: http://www.sony-semicon.co.jp/products_en/news/detail/170301.pdf
"Type 3.6" is 44x33mm; "Type 4.2" is 54x40mm. (Don't ask!)
P. S. (1) Sorry for the repeat of an earlier link. (2) Note that the advertised applications do not include any reference to medium format cameras, but are instead the same ones that Kodak and Dalsa primarily made their big CCD's for: "Aerial Photography", "Large Area Surveillance", "Cultural Heritage (meaning extremely recordings of paintings for conservation work and such), and "Inspection" (machine vision on production lines?). It seems that camera formats larger than 36x24mm continue to rely on sharing technology with other "technical" usages. (Except the Leica S). This is probably good news: it somewhat protects DMF against the worst economy-of-scale problems in its competition against 36x24mm.
-
It seems to me not very economical to make two MF format sensors.
It is already a small market.
So these MF sensors will cost even more than needed.
Why not just stick to one 54 x 40 format- a clearly different format and quality than FF.
As the Sony brochure http://www.sony-semicon.co.jp/products_en/news/detail/170301.pdf indicates, the main markets are other "technical uses" like surveillance and machine vision. The size difference probably translates into a sufficiently large cost difference that some customers will need the higher IQ while others will need the less expensive option.
-
They list two monochrome sensors ... can't any sensor be monochrome if you leave off the filter array? Just curious what target market this would be for.
-
Hi,
Some report very good results and I have seen a few awful results.
There is a lot of IQ-variation in the TSE-line. I guess it depends on getting a very good sample and what expectations are made on image corners. Many images don't have a great amount pf detail in the extreme corners.
Best regards
Erik
Excuse the dumb question, but what happens if one uses eg the Canon TSE wides on MF? Is the shift ability negligible?
Edmund
-
Hi,
A 54x40 sensor is several times the cost to make compared to 44x33 and the market is probably much larger for the smaller format, especially now that we have the X1D and the GFX that both seem to sell extremely well.
We can assume that the GFX format (if we call it so) goes to 100 MP at 2019, probably with a hike in price but it will probable be still affordable with enough competition around to keep vendors honest.
So 100 MP coming to affordable MFD is not making things more expensive, rather cheaper. Also, the new lenses probably keep up with 100 MP sensors. Probably not all of them and probably not all over the field.
Best regards
Erik
It seems to me not very economical to make two MF format sensors.
It is already a small market.
So these MF sensors will cost even more than needed.
Why not just stick to one 54 x 40 format- a clearly different format and quality than FF.
-
I'm guessing they would have a GFX50S and a GFX100S as I don't think they want to risk their status as the cheapest MF camera.
Hi,
A 54x40 sensor is several times the cost to make compared to 44x33 and the market is probably much larger for the smaller format, especially now that we have the X1D and the GFX that both seem to sell extremely well.
We can assume that the GFX format (if we call it so) goes to 100 MP at 2019, probably with a hike in price but it will probable be still affordable with enough competition around to keep vendors honest.
So 100 MP coming to affordable MFD is not making things more expensive, rather cheaper. Also, the new lenses probably keep up with 100 MP sensors. Probably not all of them and probably not all over the field.
Best regards
Erik
-
They list two monochrome sensors ... can't any sensor be monochrome if you leave off the filter array? Just curious what target market this would be for.
See my post above: the target markets include things like surveillance and machine vision (and perhaps also astronomy) where there is significant demand for the greater sensitivity of a monochrome sensor, and the option to detect UV and IR with the use of suitable filters.
-
Fab cost premium depends on process yield - it can be insignificant to several x, my guess next year 54x40 will be sub 1K in price, especially since Sony is now sole supplier and can afford to invest in production equipment and R&D and according to japanese pricing philosophy will price low to avoid new market entries. The next frontiers so to speak are going to be 54x54 and in a few years 6x9 and 4x5. I doubt I will live to see commercial cheap 8x10.
Edmund
Hi,
A 54x40 sensor is several times the cost to make compared to 44x33 and the market is probably much larger for the smaller format, especially now that we have the X1D and the GFX that both seem to sell extremely well.
We can assume that the GFX format (if we call it so) goes to 100 MP at 2019, probably with a hike in price but it will probable be still affordable with enough competition around to keep vendors honest.
So 100 MP coming to affordable MFD is not making things more expensive, rather cheaper. Also, the new lenses probably keep up with 100 MP sensors. Probably not all of them and probably not all over the field.
Best regards
Erik
-
General question – where does this all end? If Fuji and Sony move to 100mp are enough people really going to follow? Set aside the question as to who really needs 100mp. I’m thinking about all the other costs associated costs of moving up in file size. When I moved up to the XF100 the other costs added up over time. A new desk top computer, a new laptop, new monitors (why have anything but EIZO;), more storage (at least that’s cheap), new lenses and better software (had to have C1 to get the most out of the XF files). Once you get obsessed with quality you are always trying to solve the weak link in the chain! Now my latest problem – my image files when processed often reach 2Gs+. These large files can’t be stored in LR and Bridge – at least to my knowledge. I get that many working pros may need the best available but it does come with a cost. Maybe there enough people like me who can’t help themselves, have enough discretionary income and print large! Moving from 24mp to 50mp is not that big of a deal but moving to 100mp+ really stresses the weak link not to mention that to really get the quality you seek you will likely be shooting on a tripod. Did I mention how heavy my pack is now:)
-
General question – where does this all end? If Fuji and Sony move to 100mp are enough people really going to follow? Set aside the question as to who really needs 100mp. I’m thinking about all the other costs associated costs of moving up in file size. When I moved up to the XF100 the other costs added up over time. A new desk top computer, a new laptop, new monitors (why have anything but EIZO;), more storage (at least that’s cheap), new lenses and better software (had to have C1 to get the most out of the XF files). Once you get obsessed with quality you are always trying to solve the weak link in the chain! Now my latest problem – my image files when processed often reach 2Gs+. These large files can’t be stored in LR and Bridge – at least to my knowledge. I get that many working pros may need the best available but it does come with a cost. Maybe there enough people like me who can’t help themselves, have enough discretionary income and print large! Moving from 24mp to 50mp is not that big of a deal but moving to 100mp+ really stresses the weak link not to mention that to really get the quality you seek you will likely be shooting on a tripod. Did I mention how heavy my pack is now:)
Up until recently I was working with a P45+, which by today's standards is a low MP count, even compared to 35mm. Also, the noise in anything above 100 ISO was not great.
I never heard any complaints from a client. I sent over images that I considered noisy, waiting for negative responses that never came. When clients asked the resolution and I said 39, no one ever said we need more.
I soon figured out clients don't care as much as we do. I only upgraded to an IQ 260 for the better screen and interface, and connectivity to the XF platform. Noise is better, but I tell you what, some of the noisiest pictures I brought back from Cuba are the ones people like the most, and I never hear any comments about the noise.
As far as I am concerned, the only reason I could possibly consider these newer sensors is if lens cast is better on it then on the IQ260.
The never ending race for more and more resolution makes little sense to me and almost every other pro I talk to about it.
-
I agree with you Joe. I make all my money from photography and switched to digital in 1996 with a 1,4 MP Kodak DCS 420. The pixel race was on for us pro's. I think that's over now. Actually I know it's over. No one asks for pixel count anymore. That stopped ages ago. My back is packed away and hasn't seen the light of day this year. Business is pretty good at the moment. Has been all year. I am picking up new clients and have bookings until July when I take a month off to go to Tibet. I'm pretty much paying for that break with a 7 day corporate event job at various safari lodges around South Africa and must produce photo books for the attendees at the end of it. What on earth would I do with my back and Cambo on a shoot like that. No one needs it or wants it. Lots of work like that these days.
It seems mostly that it's what I would call high end enthusiasts chasing MP now. The pro market doesn't really have a big need for it but the enthusiasts are after something else. Good for them. Why not if they can afford it and have the enthusiasm for it.
I love shooting for myself and I'm generally out once or twice a week on personal projects. I have no interest in the the backs anymore. Sorry but I now equate it with work. I have owned six different backs, three different brands mounted on 5 different cameras. It was fun at the time.
-
Someone posted about this on FB this morning and I said I'd rather have a 60mp 4x5 back.
Word.
-
As far as I am concerned, the only reason I could possibly consider these newer sensors is if lens cast is better on it then on the IQ260.
The never ending race for more and more resolution makes little sense to me and almost every other pro I talk to about it.
For post production I need big files. I prefer to draw masks at 100% than at 300% magnification.
However, my P40+ is enough for all my work, 60 MP would be the maximum. More would slow down the retouch process.
(But you already know, Joe, I spent my money for a fast race bike. So there is anyway nothing left for a 60 MP back. ;))
Lens cast is a pain. I really like the SK 28 XL and the other Schneiders from 24 to 72, but the lens cast can be very time consuming
and the CF in backlight can destroy the picture.
I also have the Rodie 23 and the 28 HR but they have other weaknesses.
I really wish to have a back without cast.
And by the way, the screens on all the backs are embarrassing, :(
Regards,
Ben
-
Edmund,
Firstly, I would love to read the basis for your predictions about the costs of large sensors; they always seem vastly more optimistic than what I read elsewhere.
But mostly:
The next frontiers so to speak are going to be 54x54 and in a few years 6x9 and 4x5.
This seems utterly delusional. The medium format camera world has completely consolidated on formats no larger than 56x42mm ("645"), now that the slow, painful demise of the Rollei system is complete. Why would anyone expect new, larger sensors to be developed for formats that no current MF system supports?
-
Someone posted about this on FB this morning and I said I'd rather have a 60mp 4x5 back.
Word.
I would like that as well. Was just looking at the largesense 8x10 cameras which have reversed LCD display technology and turned them into sensors. I wonder if the new retina type screens will enable them to increase their resolution? They claim to have a 4x5 version coming...
http://largesense.com/files/4714/9073/3595/LS911-Side-View.jpg
-
I would like that as well. Was just looking at the largesense 8x10 cameras which have reversed LCD display technology and turned them into sensors. I wonder if the new retina type screens will enable them to increase their resolution? They claim to have a 4x5 version coming...
http://largesense.com/files/4714/9073/3595/LS911-Side-View.jpg
They should raise the price 16k and make it an even 10 grand a Megapixel.
-
Someone posted about this on FB this morning and I said I'd rather have a 60mp 4x5 back.
Word.
Not sure if I could warrant buying one, but surely would rent it. It would be nice to put my 4x5 to good use again, as opposed to it being an ornament in my office.
-
Edmund,
Firstly, I would love to read the basis for your predictions about the costs of large sensors; they always seem vastly more optimistic than what I read elsewhere.
But mostly:This seems utterly delusional. The medium format camera world has completely consolidated on formats no larger than 56x42mm ("645"), now that the slow, painful demise of the Rollei system is complete. Why would anyone expect new, larger sensors to be developed for formats that no current MF system supports?
A look at some older large format images may answer the why question.
There seem to be some companies doing research on low cost low rez large format sensors.
http://www.isorg.fr
Edmund
-
A look at some older large format images may answer the why question.
There seem to be some companies doing research on low cost low rez large format sensors.
http://www.isorg.fr
It is easy to answer why some forum pundits wish for it; my question instead was about why anyone should expect it to actually happen. That is, can you explain why you expect such sensors to find an economically viable outlet?
That won't be achieved by nostalgic rich hipster nonsense like the $104,000 LargeSense (http://largesense.com) 11"x9" camera, which with only 12MB and 14-bit depth is mathematically doomed to be worse for the combination of detail and fine tonal gradations than some current Sony sensors used behind today's best lenses, no matter what wonderful old LF lens is put in front of it. (Any potential dynamic range virtue of its huge photosite area is nullified by its 14-bit ADC — at a guess LargeSense does not bother with a higher bit-depth ADC because the DR of the signal arriving from those photosites is not high enough to make it worthwhile.)
-
It is easy to answer why some forum pundits wish for it; my question instead was about why anyone should expect it to actually happen. That is, can you explain why you expect such sensors to find an economically viable outlet?
That won't be achieved by nostalgic rich hipster nonsense like the $104,000 LargeSense (http://largesense.com) 11"x9" camera, which with only 12MB and 14-bit depth is mathematically doomed to be worse for the combination of detail and fine tonal gradations than some current Sony sensors used behind today's best lenses, no matter what wonderful old LF lens is put in front of it. (Any potential dynamic range virtue of its huge photosite area is nullified by its 14-bit ADC — at a guess LargeSense does not bother with a higher bit-depth ADC because the DR of the signal arriving from those photosites is not high enough to make it worthwhile.)
I think the large sensors in question will be based on cheap tech made for non photo purposes, repurposed. There is a market out there for stuff like no-moving part scanners etc.
-
I think the large sensors in question will be based on cheap tech made for non photo purposes, repurposed. There is a market out there for stuff like no-moving part scanners etc.
Indeed: read on.
Was just looking at the largesense 8x10 cameras which have reversed LCD display technology and turned them into sensors. I wonder if the new retina type screens will enable them to increase their resolution?
More precisely, they manufacture the CMOS sensor using "large format" steppers designed for making LCD screens. I have read of phone screens pushing 800dpi (beyond any practical value but anyway ...), which would give about 50MP in 10"x8", 12MP in 5"x4", 4MP in 6x7.
They claim to have a 4x5 version coming...
That is a very different idea, and a "long range plan", to be polite. Instead of 12MP in 11"x9", the description is 200MP in 4"x4", so photosites 100 times smaller in area.
-
"ISORG is the pioneering company in Organic and Printed Electronics devices for large-area photodetectors and image sensors with a paradigm shift in the industry.
ISORG converts plastic and glass surfaces into smart surfaces.
ISORG offers a new generation of high performance imagers with 3D product integration capability recognizing any shapes and form factors."
Indeed: read on.More precisely, they manufacture the CMOS sensor using "large format" steppers designed for making LCD screens. I have read of phone screens pushing 800dpi (beyond any practical value but anyway ...), which would give about 50MP in 10"x8", 12MP in 5"x4", 4MP in 6x7.That is a very different idea, and a "long range plan", to be polite. Instead of 12MP in 11"x9", the description is 200MP in 4"x4", so photosites 100 times smaller in area.
-
"ISORG is the pioneering company in Organic and Printed Electronics devices for large-area photodetectors and image sensors with a paradigm shift in the industry.
ISORG converts plastic and glass surfaces into smart surfaces.
ISORG offers a new generation of high performance imagers with 3D product integration capability recognizing any shapes and form factors."
Lots of marketing hype, but what does ISORG tell us about key specs like the resolution and dynamic range that these jumbo sensors are capable of? The only spec I found is 500dpi for a fingerprint sensor (something with very low DR needs), which is in line with what can be made with steppers designed for LCD fab and such. That makes sense, since ISORG is surely using fab equipment from the familiar suppliers (AMSL, Nikon Precision, Canon, etc.), not making its own, and 500dpi is roughly in line with the resolutions of phone screens and such, though a bit less than my optimistic 800dpi in an earlier post.
That 500dpi converts to 1.25MP in 6x7 [56x70mm] format for example, so I ask again
why we should expect this technology to bring us usable 6x7 sensors for medium format cameras?
Even if the 22MP of early DMF sensors is enough in a 6x7 format camera, that needs about 2000ppi; I do not see any display screens needing anywhere close to much resolution, so why would fab equipment for making display screens go that fine?
-
4x5 and 8x10 revival?
Edmund
Lots of marketing hype, but what does ISORG tell us about key specs like the resolution and dynamic range that these jumbo sensors are capable of? The only spec I found is 500dpi for a fingerprint sensor (something with very low DR needs), which is in line with what can be made with steppers designed for LCD fab and such. That makes sense, since ISORG is surely using fab equipment from the familiar suppliers (AMSL, Nikon Precision, Canon, etc.), not making its own, and 500dpi is roughly in line with the resolutions of phone screens and such, though a bit less than my optimistic 800dpi in an earlier post.
That 500dpi converts to 1.25MP in 6x7 [56x70mm] format for example, so I ask again
why we should expect this technology to bring us usable 6x7 sensors for medium format cameras?
Even if the 22MP of early DMF sensors is enough in a 6x7 format camera, that needs about 2000ppi; I do not see any display screens needing anywhere close to much resolution, so why would fab equipment for making display screens go that fine?
-
Not sure if I could warrant buying one, but surely would rent it. It would be nice to put my 4x5 to good use again, as opposed to it being an ornament in my office.
I sold all my film cameras a decade ago but I wish I had saved a few, like a 500c/m and a view camera. So sad.
My equipment shelf in 2004...
-
Hi,
The nice thing with "Blads" is that used "Blads" are abundant and cheap, mostly. Superachromats and 40/4 IF are still pricey, tough.
Best regards
Erik
I sold all my film cameras a decade ago btu I wish I had saved a few, like a 500c/m and a view camera. So sad.
My equipment shelf in 2004...
-
4x5 and 8x10 revival?
I was responding initially to your talk of reviving 6x6 and 6x7, where we now seem agreed that these sensors based on lower-cost "LCD fab" are not very appealing. Move that 500dpi up to 5"x4" and you get 2500x2000 = 5MP. Do you see much appeal in that?
-
I was responding initially to your talk of reviving 6x6 and 6x7, where we now seem agreed that these sensors based on lower-cost "LCD fab" are not very appealing. Move that 500dpi up to 5"x4" and you get 2500x2000 = 5MP. Do you see much appeal in that?
actually, if i had a cheap chassis that would revive my Lancaster Instantograph I would love it :)
there's a lot of old large format equipment floating around, and the images from old large format gear look great, I used a lot of vintage plate gear when I was a kid, it was all I could get where I lived, and I could develop and contact print with just dishes and a contact frame. It's easy to write of that old tech now, but like the Leica and Rolleiflex it was pretty effective at what people wanted to do, especially landscape, city architecture studies and studio portrait. 500 dpi and 20 Mp sounds like a decent contact print from an 8x10.
And then there's always the nuclear option of loading a large-format film chassis once the preview looks good. I think 8x10 mono film eg Hpx outresolves most 35mm cameras even nowadays and competes well in DR if you scan it and can live with some noise :)
Interestingly, I see the same phenomenon in watercolor equipment, where some 100 year old palette designs remain highly functional and sought-after objects, and are even reproduced by craftsmen for new/old designs.
Edmund