Is Sony taking over our corner of the world?

[eronald]

Actually, while we are defining things, that use of the name "CMOS" for modern sensor types is widespread but rather misses the point, since all sensors are MOS devices, and "CMOS vs n-MOS vs p-MOS" is not the significant design difference.  The far more informative description used in technical documents is Active Pixel Sensor, which is often abbreviated to "APS" – but I can see why that could be confusing, given the weird tradition of using that failed film format as an indication of sensor size.  Active pixel sensor designs are usually implemented as CMOS devices, but Panasonic has made some with n-MOS, and CCD's are built with p-MOS or n-MOS.

The key distinction of the active pixel sensor design is that the signal (the charge on a tiny capacitor, aka electron well) is read out via the voltage induced by that charge without moving the charge, allowing for amplification in the transfer (the "active" part) as well as direct transfer photosite-to-edge, and repeated reading of the charge, for noise reduction.

Also, about "interleaving": the CCD hop count is often halved by having read-out of each quadrant of the sensor to the nearest corner, with an ADC at each corner.  So on a 6000X4000 sensor, each line does up to 3000 hops to the nearest edge, and then each charge on each half line does up to 2000 hops along the edge to the nearest corner (or with 2000 and 3000 swapped.)

Thank you for this luminous explanation.

One *could* have more (parallel) readouts at the edges in a CCD, and in fact one could have on-chip A/D as well.

Quantum dot technology seems to be the new kid on the block ...
http://image-sensors-world.blogspot.fr/2015/10/invisage-to-unveil-quantumcinema.html


Edmund

[eronald]

Here are some cats to help the optical finder dSLR pigeons maintain an active fitness régime videos to inform photographers about the benefits of Sony technology







Edmund

[ddolde]

There is no "our" corner. It may be your corner but so far they don't come close the image quality I get with my Phase One IQ180.  I traded up from the IQ140 and it's a whole new dimension. Seems more than just a pixel increase, the files are thicker and richer.

[rdonson]

There is no "our" corner. It may be your corner but so far they don't come close the image quality I get with my Phase One IQ180.  I traded up from the IQ140 and it's a whole new dimension. Seems more than just a pixel increase, the files are thicker and richer.

I get that from a friend who shoots with an 8x10 view camera.   

[shadowblade]

There is no "our" corner. It may be your corner but so far they don't come close the image quality I get with my Phase One IQ180.  I traded up from the IQ140 and it's a whole new dimension. Seems more than just a pixel increase, the files are thicker and richer.

Where stitching is an option, I prefer Sony FF output to the IQ180 - at the same focal length, you get far more pixels in the same angle of view.

If they can reduce the base ISO of CMOS sensors to 25, they should far outstrip CCDs, due to greater quantum efficiency.

Of course, single-frame output is no contest at the moment.

[eronald]

Very wrong.

Yeah, I just saw your post here
http://forum.luminous-landscape.com/index.php?topic=104939.msg864442#msg864442

I understand that the Sonycams don't track their prey like you'd like them to, and of course I'd hate to fall off a helicopter while wearing one  but they still seem to be doing pretty well with the landscape shooters and I guess the image quality should suffice for studio shooters without an MF budget.

So you think Sony has a beachhead, but it's not yet an occupation

Edmund

[Chris Livsey]

So you think Sony has a beachhead, but it's not yet an occupation

Edmund

You forget how small "our" market is in the big wide world, out there the occupation is by a 'phone. In "our" world ever more are deciding good enough is good enough and 6/12 months update cycles aren't improving their output, unless the output is talking about the very latest which they have just bought, to to do precisely that!!

[shadowblade]

Yeah, I just saw your post here
http://forum.luminous-landscape.com/index.php?topic=104939.msg864442#msg864442

I understand that the Sonycams don't track their prey like you'd like them to, and of course I'd hate to fall off a helicopter while wearing one  but they still seem to be doing pretty well with the landscape shooters and I guess the image quality should suffice for studio shooters without an MF budget.

So you think Sony has a beachhead, but it's not yet an occupation

Edmund

I'd say Sony is about one generation - or a larger body - away from a mirrorless body that matches SLR bodies in all aspects of performance (note: bodies and not lenses).

The first generation A7r was essentially a manual focus machine - 5D2-level AF with native lenses, pretty much just focus confirmation with adapters. The second generation has made progress in leaps and bounds - AF with native lenses may not be quite D4s/1Dx/5D3/D810 level, but certainly matches the 6D, for instance, and other mid-range SLRs. There's no reason the third generation can't match top-end SLRs AF-wise - after, PDAF is based on the same technology in either case, whether it's on-sensor or off-sensor.

The big issue is processor and battery power - processor power to allow for fast AF and lag-free viewfinders (common in professional camcorders, but not used in small mirrorless cameras due to power requirements), and battery power to drive all of this (heavy lenses take a lot of power to move at high speed, and you can only AF as fast as you can move the glass). They could do it now if they made a larger body - say, a D810-sized body to accommodate larger batteries and a more powerful processor. For mirrorless to be a true competitor to SLR systems (outside certain areas such as studio and landscape work where the modus operandi is 'IQ über alles) rather than a lesser option, they'll have to do so at some stage. After all, mirrorless systems have largely supplanted mirrrored (translucent mirror) systems in cinematography - but only after mirrorless camcorders increased in both size and performance to match mirrored ones.

[Tony Jay]

I'd say Sony is about one generation - or a larger body - away from a mirrorless body that matches SLR bodies in all aspects of performance (note: bodies and not lenses).

The first generation A7r was essentially a manual focus machine - 5D2-level AF with native lenses, pretty much just focus confirmation with adapters. The second generation has made progress in leaps and bounds - AF with native lenses may not be quite D4s/1Dx/5D3/D810 level, but certainly matches the 6D, for instance, and other mid-range SLRs. There's no reason the third generation can't match top-end SLRs AF-wise - after, PDAF is based on the same technology in either case, whether it's on-sensor or off-sensor.

The big issue is processor and battery power - processor power to allow for fast AF and lag-free viewfinders (common in professional camcorders, but not used in small mirrorless cameras due to power requirements), and battery power to drive all of this (heavy lenses take a lot of power to move at high speed, and you can only AF as fast as you can move the glass). They could do it now if they made a larger body - say, a D810-sized body to accommodate larger batteries and a more powerful processor. For mirrorless to be a true competitor to SLR systems (outside certain areas such as studio and landscape work where the modus operandi is 'IQ über alles) rather than a lesser option, they'll have to do so at some stage. After all, mirrorless systems have largely supplanted mirrrored (translucent mirror) systems in cinematography - but only after mirrorless camcorders increased in both size and performance to match mirrored ones.

Interesting thoughts.

Tony Jay

[eronald]

If the body needs to be large for heat dissipation, I wonder if it wouldn't be easier in a way to move the sensor rather than the lens elements - at least for small movements.
Go from 5 axis to 6 axis stabilisation, and also turn all that nice new Zeiss MF glass into AF glass

Edmund

I'd say Sony is about one generation - or a larger body - away from a mirrorless body that matches SLR bodies in all aspects of performance (note: bodies and not lenses).

The first generation A7r was essentially a manual focus machine - 5D2-level AF with native lenses, pretty much just focus confirmation with adapters. The second generation has made progress in leaps and bounds - AF with native lenses may not be quite D4s/1Dx/5D3/D810 level, but certainly matches the 6D, for instance, and other mid-range SLRs. There's no reason the third generation can't match top-end SLRs AF-wise - after, PDAF is based on the same technology in either case, whether it's on-sensor or off-sensor.

The big issue is processor and battery power - processor power to allow for fast AF and lag-free viewfinders (common in professional camcorders, but not used in small mirrorless cameras due to power requirements), and battery power to drive all of this (heavy lenses take a lot of power to move at high speed, and you can only AF as fast as you can move the glass). They could do it now if they made a larger body - say, a D810-sized body to accommodate larger batteries and a more powerful processor. For mirrorless to be a true competitor to SLR systems (outside certain areas such as studio and landscape work where the modus operandi is 'IQ über alles) rather than a lesser option, they'll have to do so at some stage. After all, mirrorless systems have largely supplanted mirrrored (translucent mirror) systems in cinematography - but only after mirrorless camcorders increased in both size and performance to match mirrored ones.

[shadowblade]

If the body needs to be large for heat dissipation, I wonder if it wouldn't be easier in a way to move the sensor rather than the lens elements - at least for small movements.
Go from 5 axis to 6 axis stabilisation, and also turn all that nice new Zeiss MF glass into AF glass

Edmund

Not for heat dissipation, but in order to hold a powerful enough processor to both AF quickly and drive a lag-free viewfinder, and a battery powerful enough to run the processor.

Moving the sensor certainly makes sense as a tool for AF, though.

[hjulenissen]

Not for heat dissipation, but in order to hold a powerful enough processor to both AF quickly and drive a lag-free viewfinder, and a battery powerful enough to run the processor.

Processors are in all likelihood not the problem. Ample processing power for crunching digital numbers is available at low cost and low energy. If more is needed, one (at least Canon and Sony) can design an ASIC that does even more for less.

Physics are a problem. Whenever you do anything involving mechanics, light,... stuff like that does not progress at the rate predicted by Moores law. In order to get great AF performance you need to move heavy glass fast and accurately and read a physical, analog image sensor many times a second. Without overheating. Without killing battery. Similar for liveview.

-h

[shadowblade]

Processors are in all likelihood not the problem. Ample processing power for crunching digital numbers is available at low cost and low energy. If more is needed, one (at least Canon and Sony) can design an ASIC that does even more for less.

You need a processor fast enough to take a constant stream of data from the sensor and convert it into an image to be displayed on an LCD so fast as to be imperceptible. Plenty of such processors around, yes - but generally in things such as notebook computers and camcorders plugged into mains power, which are much bigger than a camera and may not have t rely on a battery.

Physics are a problem. Whenever you do anything involving mechanics, light,... stuff like that does not progress at the rate predicted by Moores law. In order to get great AF performance you need to move heavy glass fast and accurately

Same as in any other camera. Mirrorless is not unique in having to do this.

But moving heavy glass takes lots of energy. So you need a more powerful battery, which means either a larger camera, a more advanced, denser power source or shorter battery life.

and read a physical, analog image sensor many times a second.

Not a problem. Top-level camcorders already do this. That's why they have imperceptible lag times when filming with them.

Also, 3D circuitry allows for pixel-parallel readouts and A/D conversion, instead of the current column-parallel approach. That is, independent readout and A/D conversion behind each pixel, rather than at the end of each column. This, naturally, is much faster on a physical level.

[BJL]

Recent generations of ARM-based processors in phones are doing some very fancy processing for camera phones, so in far bigger devices like an ILC with lens, especially in 36x24mm format, I doubt that the weight and cost of a powerful processor and the battery to run it are much of an issue.  (And for the EVF vs OVF comparison, note that the weight and space saved by eliminating the mirror and pentaprism allow for a lot of additional battery capacity within them same "size and weight envelope"!)

BTW, the idea of adding auto-focus to a manual focus lens by moving the sensor was tried in the Contax AX, back when the sensor was a chemical emulsion on a polymer film.  It made for a very bulky camera body.

[Chuck Fan]

Recent generations of ARM-based processors in phones are doing some very fancy processing for camera phones, so in far bigger devices like an ILC with lens, especially in 36x24mm format, I doubt that the weight and cost of a powerful processor and the battery to run it are much of an issue.  (And for the EVF vs OVF comparison, note that the weight and space saved by eliminating the mirror and pentaprism allow for a lot of additional battery capacity within them same "size and weight envelope"!)

BTW, the idea of adding auto-focus to a manual focus lens by moving the sensor was tried in the Contax AX, back when the sensor was a chemical emulsion on a polymer film.  It made for a very bulky camera body.

You have to move a lot more stuff, including film take up spool, film canister, pressure plate, and mirror, the PDAF sensor, and also the pentaprism, etc, with the focal plane when you move the focal plane of a film camera.   That's part of the reason Contax AX was so bulky.   It is literally a full sized 35mm film camera inside an outer body, with the outer body big enough to give plenty of room for the inner camera to move around.

This reason doesn't apply as much to a Digital.  In a Digital, all you have to move is the sensor, the shutter, and the wire ribbon leading to it.  That's why you can have in-body stabilization with DSLR, and almost certainty not with film camera.   
In addition, If you use mirrorless and PDAF and EVF, you don't need to move the mirror or the pentaprism.   You might even get away with not moving the shutter, or even not having a shutter, if you can get your sensor to flush fast enough. 
   
The other part of the reason why moving focal planes inside the body makes for bulky and thick bodies is with some lenses, the focal plane has to move quite a lot to cover the entire focus range of the lens.  If you don't give the sensor enough range of front-back movement inside the body, then it will never find a good focus on its own.   So the body would have to be much thicker than the maximum range of focus movement of any lens you plan to use with it in AF mode.   Otherwise you have to manually prefocus before the autofocus will work.

So I think with a digital camera, if you do something creative, such as make the camera body deep but slim, like a compact camcorder, instead of the DSLR shape or the current SONY mirrorless shape, which is wide, tall and thin front and back, you can produce a good digital full frame camera with in-body focusing and in-body stabilization, that never needs AF lenses.
 

[BJL]

You have to move a lot more stuff, including film take up spool, film canister, pressure plate, etc, with the focal plane when you move the focal plane of a film camera.   That's why Contax AX was so bulky.

With a DSLR, all you have to move is the sensor, the shutter, and the wire ribbon leading to it.   You might even get away with not moving the shutter, or even not having a shutter, if you can get your sensor to flush fast enough.  That's why you can have in-body stabilization with DSLR, and almost certainty not with film camera.

True, it might be less bad, but not by much (see below!)

But always the first question to ask is:

It is a simple idea, that has even been publicly tried before, so why are none of the multiple competing camera companies doing this now?

Usually the answer is the same as when outsider political candidates – free from any record or failure due to being free from any record at all – offer simple one-line solutions to persistent problems: when you look at the details, it would not actually work very well.

So it is mostly pointless to try to refute such unproven ideas; it make more sense to challenge the proposers to explain the details of how it could work.  But I will try anyway!  One likely problem is that a very large amount of movement of the focal plane (the sensor) relative to the lens would be needed for longer focal lengths.  Close focusing down to a modest 1:4 magnification requires a lens extension (or focal plane movement) of about (focal length)/4, so for a focal length of 100mm, already 25mm or one inch of sensor movement would be needed.

And on the other hand, modern internal focusing designs only require moving a few relatively light intermediate lens elements, not the big heavy ones at the front.

[Chuck Fan]

I would say no one has tried it before because no one is yet ready to break completely away from the traditional camera form factor for a "serious" camera.   Although some smartphone camera attachment has already moved in this direction

A really workable in-body focusing camera will have to be shaped like a camcorder, long in the lens axis, and trim in the other two axis.  It might even look like a short monocular telescope, or a detached rifle scope,  with the user looking into a EVF eye piece that is directly behind and in line with the interchangible objective lens.   Traditional serious camera is the other way around, skinny in the lens axis, and long in the other two axis, broadly like a flat plate bolted perpendicularly to the lens.  That form doesn't work very well with a focal plane that has to move quite a bit.    I think it will take some doing to move serious users accustomed to the traditional form factor to this new one.   The first models won't sell too well.   That's why no one wants to be the first to try, unless they have nothing to lose.

[ErikKaffehr]

Hi,

Camcorders have small sensors, so they have very large DoF. Professional video cameras don't have AF. Have you seen any PL-mount lens with AF. They use a "focus puller" an extra guy doing just the focusing…

In video, fast AF is not wanted because they don't want jumpy AF. Sony actually has a slow AF mode just for video. Video is different, so leave it out.

Modern lenses use internal focusing, just moving a single group of elements, that reduces the mass to focus. Modern AF lenses to be used with CDAF need to have fast movements, that is the reason Sony uses linear motors on the new lenses.

Very clearly, Sony cameras may not be able to match fast focusing Canons and Nikons, and I think this may to have a bit to do with limitations in on sensor PDAF.

I cannot comment really from  my own experience, as I don't have a Canon camera and mostly use Sony A-mount lenses on my A7rII. The 90/2.8G has very fast AF but I had no opportunity to compare with classic PDAF.

Canon has a lot of new focusing technology in the 5D3 and the 1DX.

Best regards
Erik

You need a processor fast enough to take a constant stream of data from the sensor and convert it into an image to be displayed on an LCD so fast as to be imperceptible. Plenty of such processors around, yes - but generally in things such as notebook computers and camcorders plugged into mains power, which are much bigger than a camera and may not have t rely on a battery.

Same as in any other camera. Mirrorless is not unique in having to do this.

But moving heavy glass takes lots of energy. So you need a more powerful battery, which means either a larger camera, a more advanced, denser power source or shorter battery life.

Not a problem. Top-level camcorders already do this. That's why they have imperceptible lag times when filming with them.

Also, 3D circuitry allows for pixel-parallel readouts and A/D conversion, instead of the current column-parallel approach. That is, independent readout and A/D conversion behind each pixel, rather than at the end of each column. This, naturally, is much faster on a physical level.

[BernardLanguillier]

What's interesting is that by the time Sony releases an A9 with a correctly sized battery,... in a package similar to that of the Leica SL, the size disadvantage of what a D820/5Dx fitted with a EVF instead of an OVF would be would be mostly gone...

Cheers,
Bernard

[Theodoros]

I have started a related to FF mirrorless future forum in the MF forum... http://forum.luminous-landscape.com/index.php?topic=105069.msg864580#new...

It aims to discuss the FF mirrorless + view camera applications related to the future of imaging...  The usual winners of "not even a bent match to light a cigar" are kindly requested NOT to participate...