The future of medium format

[ErikKaffehr]

Hi,

There has been a lot of talk about the demise of medium format digital. In my view it's more revarding to discuss the future of MFD.

For my part, the new Alpa FPS is inspiring and so is the Hartblei HCam. Those cameras focus on the essential. In my view both cameras require working live view.

On another thread, Stefan Steib suggested that MF vendors should sit down at round table and develop a CMOS sensor.

In my view, the future may lay with mirrorless cameras in both small and medium format. Live view at actual pixels is the optimal manual focusing method, for sure. In the studio, the cameras are often used tethered.

In my view a compact modular camera, with good live view and an optional electronic viewfinder is probably the way to go.

Best regards
Erik

[Steve Hendrix]

Hi,

There has been a lot of talk about the demise of medium format digital. In my view it's more revarding to discuss the future of MFD.

For my part, the new Alpa FPS is inspiring and so is the Hartblei HCam. Those cameras focus on the essential. In my view both cameras require working live view.

On another thread, Stefan Steib suggested that MF vendors should sit down at round table and develop a CMOS sensor.

In my view, the future may lay with mirrorless cameras in both small and medium format. Live view at actual pixels is the optimal manual focusing method, for sure. In the studio, the cameras are often used tethered.

In my view a compact modular camera, with good live view and an optional electronic viewfinder is probably the way to go.

Best regards
Erik

Thank you Erik. Along with discussions over the use of current medium format digital technology, this topic has value and is a reasonable use of one's time for anyone interested in medium format digital. Good start!

I will say that I believe the attempt to scale CMOS or a CMOS-similar technology has been underway. Obviously it is not as easy as we would hope - otherwise they'd have something by now. I'm curious whether any such sensors will begin their life as full frame 645 or start smaller in order to get the technology into the marketplace and scale up from there. That has been the trend with new sensor technology.

Steve Hendrix
Capture Integration

[FredBGG]

Really usable live view needs to be on the top of the list.
Much broader range of tilt shift lenses.
Better corner to corner focusing support, both manual and auto. Even more important than speed.

[ErikKaffehr]

Hi Fred,

With a technical camera every lens is a T&S lens, as long the camera offers tilts and shifts. The Hartblei can take any T&S lens for DSLRs, so they are in a pretty good shape.

Live view, with actual pixels, offers perfect focus corner to corner.

I'm not sure MF needs corner to corner AF, it seems users are doing fine with what they have.

I don't think MF needs to compete head on with DSLRs. I guess that DSLRs are quite optimal for the job they are intended for.

On the other hand, I really feel that a swinging mirror is also a part of old technology, weather on a Hasselblad or a Nikon. I use MLU whenever I use DSLRs on tripod.

Best regards
Erik

Really usable live view needs to be on the top of the list.
Much broader range of tilt shift lenses.
Better corner to corner focusing support, both manual and auto. Even more important than speed.

[FredBGG]

By much broader range of tilt shift lenses I mean more options, not just tech cameras and a couple for reflex cameras.
I also mean independent tilt in both vertical and horizontal as well as independent shift.
The Fuji gx680 this from 50mm to 500mm with very close focus on all lenses and in a reflex body.

Regarding better corner to corner focusing it does not come automatically with live view.
Current MF live view has no AF support.
Live view refresh rates need to be very fast to support live view focusing.
 
 

[design_freak]

Okay, so let's talk about the future
Whenever we speak, we move in the direction of DB + technical camera. I'm not surprised. Maybe we should ask ourselves what to do with MF cameras. I think it's still the weakest link. Why has not anyone made ​​a takeover ROLEI / Sinar / Leaf or Contax?

[Gel]

Hasselblad + Sony + 2 years = CMOS

But I've been championing CMOS for ages because of how far it's come.

I love shooting medium format but when I compare it to my 1DX if I had to choose only one, in general it would always be the Canon.
Although the reason, if I had to give just one is because of the ISO performance. Clean 3200 is all I need.

It's not really the price because a used H3DII 39 can be had for under £5000 now and the lenses are superb.

I'll be completely honest though, build me a full sized CMOS sensor of which I'm not locked to a body (Hi Hasselblad) and I'm there.

Because, I have the H1 and H3DII, one for film, one for digital. If any of the bodies fail I can't swap the backs around. So I'm 2000 miles from home without redundancy. Smart move Hasselblad who the hell thought that one up? Thanks for releasing the H4X btw, I'd still rather have a Hassy back on it though.

[Paul2660]

I am curious on the CMOS vs CCD issue.  Is it just not possible to get a good live view from a CCD?  Does only CMOS allow this?  The  are CCD's in the 60mp and 80mp sensors not single chips, but instead they are 8 separate chips, matted into one chip.  You can see this when you view the sensor in bright light.  You can also sometimes actually see the hard physical lines in your image  when the back has not been tuned well. The size of the separate CCD's on the IQ160 Chip appear to be about the size of a APC-C sensor.  As I recall on the older Kodak chips in the P45+ you did not see the 8 separate segments, instead it was just one large chip.  I have always assumed you have to create the chip first, then build a camera around the chip.  I am surprised that it has taken this long to get a composite style chip similar to the Dalsa CCD chips out of CMOS.  I assume it must be lack of market demand?

The Tech camera solution is more a requirement due to optical shortfalling by the MF wide angles.  I have only used the Mamiya/Phase line up, so cannot speak to Hassi or Contax.  The 35mm and 28mm both were just too soft in the corners for my work.  The ability to add tilt to a medium format wide angle lens is also a key component.  You can create a very impressive  hyperfocal distance.   All you have to do is shoot with a wide on a tech camera and see the differences in clarity/color/contrast and overall focus, you will not want to go back.   On a side note, this same issue has now appeared with the D800 and wide angle lenses, and I am sure will also appear with Canon when they bring a higher resolution solution to the market. 

The Tech camera route, leads to a need for a better Live view, so hopefully Phase will follow up in 2013 with a new back with Live view.  I have to commend Phase for having put into place an excellent software solution in Capture One that can accommodate the LCC process so well.

Paul 

[Gel]

What surprises me is they haven't developed a form of piggy back system.

When I shoot film I always use the 5D3 as my exposure meter for the histogram and relate it to the film exposure.

If it's such an issue why can't they create a bolt on accessory or even an iphone app that simulates exposure, tilt, shift with built in focus peaking (simulated) based on feedback from the camera (that sits in the hotshoe).

[Hywel]

I currently use a Hassleblad H3Dii-31 and a Canon 7D, shooting for my own website which I'd characterise as "fetish fashion" (http://www.restrainedelegance.com if anyone is interested - not safe for work...)

I use the Hasselblad wherever possible because of its clarity, colour rendition and sharpness. I like the big bright optical viewfinder, the delay on the mirror flip (which makes a perceptible difference in critical sharpness even with a very short delay), the fact that the camera compensates for focus shift at different apertures, and the leaf shutters which further reduce vibration and allow me to work with short shutter speeds with studio flash.The lack of optical low pass filter is a mixed blessing- increases perceived sharpness and micro-contrast, but is the devil's own job to shoot a model wearing stockings and suspenders.

Nothing stops the 35mm dSLR crowd from matching the salient points here to achieve similar levels of sharpness and perceived image quality.

However, I don't believe the dSLRs are there yet. My technique has probably got a bit sloppy but I'm spoilt by not having to worry too much about camera shake. When I have to fall back to the Canon and shoot at 1/160th of a second, with OLPF, and lenses which don't match the quality of the Hasselblad ones, I certainly notice the difference big-time at 100% zoom (let alone the lack of megapixels).

I don't shoot tethered and I dislike live view as a method of shooting. If I was a fan, I'd probably be trialling using my RED Scarlet for stills as well as video.

Until electronic viewfinders have resolution similar to the Mk I eyeball plus a Hassy lens and viewfinder, I want to stick to fast, unencumbered shooting with direct zero-lag vision of what's going on to pick the moment to press the shutter release.

The key improvement that would sell me on an MF upgrade is light sensitivity. The main reason I fall back to the Canon is available light shooting. Even though the H3Dii-31 has microlenses I find it hugely light-hungry, shooting ETTR I rate it somewhere around ISO 64. Indirect lighting indoors? Forget it.

There's always room to improve autofocus. I don't need more points- I have the Canon set to only use the centre one to focus and recompose anyway. But faster performance, especially in low light, would be welcome, and every little bit of accuracy helps - Hassy's TrueFocus sounds worthwhile. As various web reviewers are discovering, there's more to getting an accurately-focussed sharp 30+ megapixel image than slapping a finer grained sensor in a Nikon body- you need to pay attention to all these little physical niggles like focus shift, shift on recompose, etc.

A base ISO 400-800 sensor at 645 size with 40-ish megapixels and all the precision of MF? Sold.

  Cheers, Hywel.

[ErikKaffehr]

Hi,

The 7D has a double disadvantage of sensor size. A larger sensor is mostly beneficial. Regarding OLP filtering Canon could remove it if they didn't regard it necessary.

A CMOS sensor would probably offer a significant improvement in ISO capability, as they seem to have much lower noise levels compared to CCD.

Live view is a feature mostly beneficial to users of MF on technical cameras or users who need exact focusing.

A small pixel sensor would be helpful with Moiré, and so would stopping down to smallish apertures.

Best regards
Erik

I currently use a Hassleblad H3Dii-31 and a Canon 7D, shooting for my own website which I'd characterise as "fetish fashion" (http://www.restrainedelegance.com if anyone is interested - not safe for work...)

I use the Hasselblad wherever possible because of its clarity, colour rendition and sharpness. I like the big bright optical viewfinder, the delay on the mirror flip (which makes a perceptible difference in critical sharpness even with a very short delay), the fact that the camera compensates for focus shift at different apertures, and the leaf shutters which further reduce vibration and allow me to work with short shutter speeds with studio flash.The lack of optical low pass filter is a mixed blessing- increases perceived sharpness and micro-contrast, but is the devil's own job to shoot a model wearing stockings and suspenders.

Nothing stops the 35mm dSLR crowd from matching the salient points here to achieve similar levels of sharpness and perceived image quality.

However, I don't believe the dSLRs are there yet. My technique has probably got a bit sloppy but I'm spoilt by not having to worry too much about camera shake. When I have to fall back to the Canon and shoot at 1/160th of a second, with OLPF, and lenses which don't match the quality of the Hasselblad ones, I certainly notice the difference big-time at 100% zoom (let alone the lack of megapixels).

I don't shoot tethered and I dislike live view as a method of shooting. If I was a fan, I'd probably be trialling using my RED Scarlet for stills as well as video.

Until electronic viewfinders have resolution similar to the Mk I eyeball plus a Hassy lens and viewfinder, I want to stick to fast, unencumbered shooting with direct zero-lag vision of what's going on to pick the moment to press the shutter release.

The key improvement that would sell me on an MF upgrade is light sensitivity. The main reason I fall back to the Canon is available light shooting. Even though the H3Dii-31 has microlenses I find it hugely light-hungry, shooting ETTR I rate it somewhere around ISO 64. Indirect lighting indoors? Forget it.

There's always room to improve autofocus. I don't need more points- I have the Canon set to only use the centre one to focus and recompose anyway. But faster performance, especially in low light, would be welcome, and every little bit of accuracy helps - Hassy's TrueFocus sounds worthwhile. As various web reviewers are discovering, there's more to getting an accurately-focussed sharp 30+ megapixel image than slapping a finer grained sensor in a Nikon body- you need to pay attention to all these little physical niggles like focus shift, shift on recompose, etc.

A base ISO 400-800 sensor at 645 size with 40-ish megapixels and all the precision of MF? Sold.

  Cheers, Hywel.

[BJL]

I am curious on the CMOS vs CCD issue.  Is it just not possible to get a good live view from a CCD?  Does only CMOS allow this?  The  are CCD's in the 60mp and 80mp sensors not single chips, but instead they are 8 separate chips ...

It is not possible to get the video feed needed for live view from the Full Frame(*) type of CCDs that is used in MF. That is why interline type of CCD was developed, as used in video cameras and many smaller format still cameras until CMOS started taking over. For video/live view, it makes no sense moving to interline CCD, even though MF suppliers like Kodak/Truesense offer interline CCDs, because they have worse noise and dynamic range than either full frame CCD or modern active pixel CMOS sensors.

Another video option is frame transfer CCDs, also used in some video cameras, but that technology only works well for quite small sensors: it requires a separate storage frame on the silicon next to the sensor and as big as the sensor itself.

The CCDs you mention are indeed single chips, though in fabrication on the silicon wafer, each sensor is etched part at a time. From what I have read, the segmentation you see on those big 60MP and 80MP sensors is not on the sensor's silicon chip itself, but in the coatings atop it, like color filter arrays and microlenses, which have to put on in several pieces on such large sensors.


(*) "Full Frame" has referred to a type of CCD design since long before it came to mean "36x24mm or 645 format, but nothing smaller or in-between."

[BJL]

Thanks Erik,

I even tried to move the title of that other thread to your title here, to get away from the futility of asking for a yes or no answer and move towards discussing a range of possibilities and hopes.

Two recap some ideas that got buried in that thread, it now looks to me that for formats larger than 36x24mm, a move to more modern sensor technologies like active pixel CMOS with on-chip, column-parallel ADC is both necessary in the long term, and becoming feasable due to the rise of sensor designers like CMOSIS, which seems willing and able tom design a large, high quality, relatively low-volume CMOS sensor, as indicated by its design work with Leica for the new Leica M.

I would bet on a small sensor design house like CMOSIS or Aptina (the latter worked with Nikon on designing the Nikon One sensors) rather than Sony, because Sony seems less interested in small volume custom designs.

Design is the main issue: actual fabrication can then be outsourced to one of several competent foundries. None of Nikon, CMOSIS, or Aptina have ever manufactured the sensors that they have designed over the years.

[FredBGG]

I currently use a Hassleblad H3Dii-31 and a Canon 7D, shooting for my own website which I'd characterise as "fetish fashion" (http://www.restrainedelegance.com if anyone is interested - not safe for work...)

I use the Hasselblad wherever possible because of its clarity, colour rendition and sharpness. I like the big bright optical viewfinder, the delay on the mirror flip (which makes a perceptible difference in critical sharpness even with a very short delay), the fact that the camera compensates for focus shift at different apertures, and the leaf shutters which further reduce vibration and allow me to work with short shutter speeds with studio flash.The lack of optical low pass filter is a mixed blessing- increases perceived sharpness and micro-contrast, but is the devil's own job to shoot a model wearing stockings and suspenders.

Nothing stops the 35mm dSLR crowd from matching the salient points here to achieve similar levels of sharpness and perceived image quality.

However, I don't believe the dSLRs are there yet. My technique has probably got a bit sloppy but I'm spoilt by not having to worry too much about camera shake. When I have to fall back to the Canon and shoot at 1/160th of a second, with OLPF, and lenses which don't match the quality of the Hasselblad ones, I certainly notice the difference big-time at 100% zoom (let alone the lack of megapixels).

I don't shoot tethered and I dislike live view as a method of shooting. If I was a fan, I'd probably be trialling using my RED Scarlet for stills as well as video.

Until electronic viewfinders have resolution similar to the Mk I eyeball plus a Hassy lens and viewfinder, I want to stick to fast, unencumbered shooting with direct zero-lag vision of what's going on to pick the moment to press the shutter release.

The key improvement that would sell me on an MF upgrade is light sensitivity. The main reason I fall back to the Canon is available light shooting. Even though the H3Dii-31 has microlenses I find it hugely light-hungry, shooting ETTR I rate it somewhere around ISO 64. Indirect lighting indoors? Forget it.

There's always room to improve autofocus. I don't need more points- I have the Canon set to only use the centre one to focus and recompose anyway. But faster performance, especially in low light, would be welcome, and every little bit of accuracy helps - Hassy's TrueFocus sounds worthwhile. As various web reviewers are discovering, there's more to getting an accurately-focussed sharp 30+ megapixel image than slapping a finer grained sensor in a Nikon body- you need to pay attention to all these little physical niggles like focus shift, shift on recompose, etc.

A base ISO 400-800 sensor at 645 size with 40-ish megapixels and all the precision of MF? Sold.

  Cheers, Hywel.

Why do you have crop sensor cameras? You go on about the magnificent Hasselblad lenses and then up a small sensor behind them throwing away
much of the clarity/resolution they project.

You also say that 35mm DSLRs are not there yet, but choose to use a crop sensor Canon with a 1.6 crop factor.
Take a look at how the D800 compared to the H3D 40:
http://www.luminous-landscape.com/forum/index.php?action=dlattach;topic=69391.0;attach=64261;image



You say that even with the canon you use a center focus point and recompose... well that's relatively OK if you don't shoot wide open. No very shallow depth of field on your website.

One of the pluses of MF is the larger sensor and how it produced nice shallow depth of field. To make this more usable better focusing support, both manual and auto,
would be a strong improvement.

Full frame sensors are the way to go for MFD in the future IMO. MF crop sensors just overlap with current top of the line 35mm dslr.

Regarding mirror up functionality. Both the Hasselblads and the D800 have mirror up with or without delay.

[FredBGG]

A really nice compact folding MFD camera like the Fuji GF680 would be a very nice camera for the luxury market.
A luxury MFD camera that someone can actually carry around. However it would require a higher level if compact electronics
and a more compact battery.

I can see a better future to the prestige of MF brands through that then the crazy Lunar BS.

[ErikKaffehr]

Hi,

CCD is essentially based on popping charge between adjacent cells. So you sort of pop each cell to the next one, taking several thousands of pops to get the charge to the readout circuitry. That takes time and generates heat. CMOS is more straightforward, you are reading charge in place.

The reason that large sensors are stitched is that the devices exposing the masks on silicon substrate are limited to small apertures. So most sensors larger than APS-H are stitched. This stitching is not done physically but when exposing the sensor chips.

Regarding live view, there are dual reasons for it being important. One is that getting rid of the mirror is good for flexibility. For instance you can use fairly symmetrical wide angle designs with mirrorless but inverted telephoto is needed for wide angles on SLR.

The other factor is that LV removes variations in ground glass screen and sensor registration as the actual signal from the sensor is used for focusing. For instance, the Alpa cameras can be shimmed within 0.01 mm, but thermal expansion from say -20 to +20 degrees may expand the body more than that!

Live view is simply "what you see is what you get", and it comes no better than that.

Best regards
Erik

I am curious on the CMOS vs CCD issue.  Is it just not possible to get a good live view from a CCD?  Does only CMOS allow this?  The  are CCD's in the 60mp and 80mp sensors not single chips, but instead they are 8 separate chips, matted into one chip.  You can see this when you view the sensor in bright light.  You can also sometimes actually see the hard physical lines in your image  when the back has not been tuned well. The size of the separate CCD's on the IQ160 Chip appear to be about the size of a APC-C sensor.  As I recall on the older Kodak chips in the P45+ you did not see the 8 separate segments, instead it was just one large chip.  I have always assumed you have to create the chip first, then build a camera around the chip.  I am surprised that it has taken this long to get a composite style chip similar to the Dalsa CCD chips out of CMOS.  I assume it must be lack of market demand?

The Tech camera solution is more a requirement due to optical shortfalling by the MF wide angles.  I have only used the Mamiya/Phase line up, so cannot speak to Hassi or Contax.  The 35mm and 28mm both were just too soft in the corners for my work.  The ability to add tilt to a medium format wide angle lens is also a key component.  You can create a very impressive  hyperfocal distance.   All you have to do is shoot with a wide on a tech camera and see the differences in clarity/color/contrast and overall focus, you will not want to go back.   On a side note, this same issue has now appeared with the D800 and wide angle lenses, and I am sure will also appear with Canon when they bring a higher resolution solution to the market. 

The Tech camera route, leads to a need for a better Live view, so hopefully Phase will follow up in 2013 with a new back with Live view.  I have to commend Phase for having put into place an excellent software solution in Capture One that can accommodate the LCC process so well.

Paul 

[yaya]

Live view is simply "what you see is what you get", and it comes no better than that.

Kind of...at the moment it is more like "what you see is what the camera interpolates out of the data read off the sensor"...hence the difference in LV image quality at 1:1 between cameras. Compare the 5DIII to the D800 to see the differences

Yair

[Hywel]

Why crop sensor Hasselblad?

1) All I could afford at the time (when I added in lenses, etc.)
2) 1 stop improvement in light sensitivity over the non-crop frame versions (micro lenses on sensor)
3) The crop is fairly modest and gives "look around" capability on the viewfinder, which I've grown to like
4) To date, the improvements in newer Hasselblad models hasn't provided a compelling reason to change. As I said, base 400 ISO and 40-ish megapixels and I'd be there

Why crop sensor Canon?

1) I bought it for video, and at the time it had significantly less compromises than the 5Dii (got fixed in firmware later on but it didn't have 25 fps IIRC which is an issue in PAL land)
2) I sold my 5D Mark 1
3) It is only the backup camera, so I bought a RED instead of a full-frame replacement backup stills camera. I definitely will go back to a full-frame Canon when I get around to replacing the 7D

Good to know that the D800 implements a mirror flip-delay system. As I said, for my purposes I don't see any particular reason why 35mm full frame shouldn't deliver the kind of results I'm currently getting from the Hassy, so maybe they're one step closer than I realised, but I believe there are still issues with focus shift etc.. And if I do update the Hasselblad, at least I have lenses capable of supporting the higher spec readout at the back end without investing in a sack full of expensive new glass.

  Cheers, Hywel.

[ErikKaffehr]

Hi!

I don't have neither the 5DIII nor the D800, so I cannot comment on that. It is my understanding that maximal enlargement on the D800 is larger than actual pixels.

Live view is just a tool and we need to learn how to use it. I am a Sony user and still learning how to use live view, but I love it!

Best regards
Erik

Kind of...at the moment it is more like "what you see is what the camera interpolates out of the data read off the sensor"...hence the difference in LV image quality at 1:1 between cameras. Compare the 5DIII to the D800 to see the differences

Yair

[FredBGG]

Why crop sensor Hasselblad?

1) All I could afford at the time (when I added in lenses, etc.)
2) 1 stop improvement in light sensitivity over the non-crop frame versions (micro lenses on sensor)
3) The crop is fairly modest and gives "look around" capability on the viewfinder, which I've grown to like
4) To date, the improvements in newer Hasselblad models hasn't provided a compelling reason to change. As I said, base 400 ISO and 40-ish megapixels and I'd be there

Why crop sensor Canon?

1) I bought it for video, and at the time it had significantly less compromises than the 5Dii (got fixed in firmware later on but it didn't have 25 fps IIRC which is an issue in PAL land)
2) I sold my 5D Mark 1
3) It is only the backup camera, so I bought a RED instead of a full-frame replacement backup stills camera. I definitely will go back to a full-frame Canon when I get around to replacing the 7D

Good to know that the D800 implements a mirror flip-delay system. As I said, for my purposes I don't see any particular reason why 35mm full frame shouldn't deliver the kind of results I'm currently getting from the Hassy, so maybe they're one step closer than I realised, but I believe there are still issues with focus shift etc.. And if I do update the Hasselblad, at least I have lenses capable of supporting the higher spec readout at the back end without investing in a sack full of expensive new glass.

  Cheers, Hywel.

3) The crop is fairly modest and gives "look around" capability on the viewfinder, which I've grown to like

Good point. Look around is a nice thing to have. Something to be kept in mind in future MF development.

D800 offers various crop abilities and gives you two options for indicating the crop in the finder. Either greyed out of a guide frame. Can be very handy.
The crop feature also increases frame rate.

Could be a good option for medium format too. 80MP at a slow frame rate and 15 or 20 much faster.