The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Backlight landscape photography realized - say no to silhouette

Hi,

Yes, the question is how?

Best regards
Erik
The same way I could get tons of DR and 16-bits on my 645Z with exposure bracketing. HDRx is basically a technique where the RED camera uses it's incredibly high capture speed overhead to alternate high and low exposures every other frame (via shutter speed), so if you're shooting 24fps for film, the camera is actually capturing 48fps with half the frames having a different exposure level. Obviously this doubles file sizes, but the benefit is that once you load the clip into REDCINE-X you can choose a blending method for the highlight/shadow transitions and have at it. The downside is that the shutter is altered for the different exposure levels, so you risk having blurry highlights if you don't pay attention to the speed for the brighter exposure before shooting.

The Dragon sensor has a "mechanical" 20 stops DR and a usable ~16, so it's not like it really needs exposure bracketing to make use of 16-bit.

It will be interesting to see how the 007 cmos performs and how it compares to the Sony.
James Bond is taking way too long to come to market though.

37.5MP is fine if we're talking about enormous DR and low noise, even by 645Z standards, but it's hard to compete with Sony purely on sensor manufacturing prowess. The spec sheet gives an ISO range of 200-6400, and ISO100 is curiously a pull setting. If the finer sensor process lets CMOSIS cram the electrical bits and produce bigger pixels than the 645Z, in addition to the lower pixel count in the first place, the S should have been capable of 25K ISO at the least. And why list DR as 13 stops? even if they're being extremely conservative, an MF sensor with 6 micron pixels and on-sensor readout should be capable of at least 15 stops or so.

Incidentally, I find it strange that DxO hasn't benchmarked the 645Z yet, despite it's popularity.
 

fotografz

Well-known member
It will be interesting to see how the 007 cmos performs and how it compares to the Sony.
I'll be interested in seeing that also. Being different from the Sony's CMOS homogenization of all other MFD sensors could be a good thing. Choice is always good.

I'm currently working with a S-006, and for what and how I shoot, I still prefer the "antiquated and inferior" aesthetics of CCD with its rendering that many feel has a pleasing "organic" look, rather than the "clinically correct" one favored by the data driven photo-nerds. Yet, I'll be wide open to the prospect of shooting with a S-007 also ... if it doesn't stray to far from the look I like.

Admittedly, the S-006 is mostly using lighting (studio and location) which is why I got the camera in the first place, and why I used a Hasselblad H before that. However, even for shooting in natural available light, I prefer exposing at the right time, in the right light as seen, rather than as forced looking HDR looking torture tests with visibly weak color depth. Also, I freely admit that shooting "sun trails" isn't my cup of tea. That is not the "different" sort of image that others can't shoot I'm striving for.

That, and the S/CS lenses are still the main reason for using the S system by a large margin.

The possible addition of a CMOS sensor in an S camera would have a few operational/functional advantages of interest, and if it follows the CMOS image advantages of more DR (for the exceedingly few times I may need it), and higher ISO, then I'll use both CCD and CMOS S cameras in concert, perhaps enabling me to finally jettison all my 35mm Sony stuff ... since I have not come to love the look and feel after years of using their cameras. Call it belief, religion, or what-ever ... artistic perceptions are my guide, and when something fulfills one's vision then, in my experience, it's a good idea to stick with it.

- Marc
 

thrice

Active member
The S007 is the same underlying architecture as the M typ 240, expect similar DR, high iso and push/pull.

The problem with all this pushing shadows in a backlit situation is that you're pushing poor quality light, despite the fact that you might get it up to appreciable levels and be able to add the contrast back it still lacks dynamism and realism. Just my 2c
 
Before getting too excited about Sony sensor shadow push capabilities I think one should consider tonality. Even if noise is low I think it looks like colors are dull, kind of brownish or otherwise going towards monochromatic, those pushed areas look like how old CMOS cameras rendered color, ie not that good. My guess is that there's too little signal (photons) captured to get good color.
Funny you should mention the Sony chip and tonality. I had a Sony A7r and sold it. It was a good camera and the 2 Sony lenses I bought for it (35 and 55) were very good but even though one could push the shadows it always looked "thin" to me which is perhaps what you have correctly named as a tonality problem. It was very good at a low ISO but as the ISO increased the file got "thin", in my estimation. Also the ability to push the shadows seems to encourage that which means a lot of photos are turning up with that extended HDR look which isn't to my taste either. The really big deal in the new CMOS backs is LIVE VIEW and long exposure. Now all we need are bigger chips that behave better with shifting.
The problem with all this pushing shadows in a backlit situation is that you're pushing poor quality light, despite the fact that you might get it up to appreciable levels and be able to add the contrast back it still lacks dynamism and realism.
One can always call the CCD-look as "film grain" (low SNR), and the CMOS-look as "plastic, cheapo, artificial" (high SNR).

It would be good to know whether your claimed lack of color / dynamism / realism is true from pushed shadows. I don't see such difference imo. Below shows the identical color / look / saturation etc from two bracketed exposures 2ev apart.

 
Last edited:

RVB

Member
I'll be interested in seeing that also. Being different from the Sony's CMOS homogenization of all other MFD sensors could be a good thing. Choice is always good.

I'm currently working with a S-006, and for what and how I shoot, I still prefer the "antiquated and inferior" aesthetics of CCD with its rendering that many feel has a pleasing "organic" look, rather than the "clinically correct" one favored by the data driven photo-nerds. Yet, I'll be wide open to the prospect of shooting with a S-007 also ... if it doesn't stray to far from the look I like.

Admittedly, the S-006 is mostly using lighting (studio and location) which is why I got the camera in the first place, and why I used a Hasselblad H before that. However, even for shooting in natural available light, I prefer exposing at the right time, in the right light as seen, rather than as forced looking HDR looking torture tests with visibly weak color depth. Also, I freely admit that shooting "sun trails" isn't my cup of tea. That is not the "different" sort of image that others can't shoot I'm striving for.

That, and the S/CS lenses are still the main reason for using the S system by a large margin.

Marc,its the S glass that keeps me in the system.. I love the IQ and sold my Hasselblad system after using both for a while,while I enjoyed the H sytem the S glass impressed more.

The possible addition of a CMOS sensor in an S camera would have a few operational/functional advantages of interest, and if it follows the CMOS image advantages of more DR (for the exceedingly few times I may need it), and higher ISO, then I'll use both CCD and CMOS S cameras in concert, perhaps enabling me to finally jettison all my 35mm Sony stuff ... since I have not come to love the look and feel after years of using their cameras. Call it belief, religion, or what-ever ... artistic perceptions are my guide, and when something fulfills one's vision then, in my experience, it's a good idea to stick with it.

Using both CCD & CMOS in your toolbox is a good plan,Leica claims 1stop of extra Dr for cmos and I like the base ISO of 200,a friend is using an IQ280 for portrait work and the base of 35 is very light hungry,the 007 base of 200 is also pretty good for handheld shots natural light.

Another dynamic in the IQ will be the CFA,will it produce a different color response to the M240?

- Marc
P.S the 3.5fps could be useful

Rob
 

fotografz

Well-known member
The S007 is the same underlying architecture as the M typ 240, expect similar DR, high iso and push/pull.

The problem with all this pushing shadows in a backlit situation is that you're pushing poor quality light, despite the fact that you might get it up to appreciable levels and be able to add the contrast back it still lacks dynamism and realism. Just my 2c
I think this is the crux of the situation. It is shooting in "unbalanced" light that is overly poor to start with.

While it is amazing what can be pulled out of these files, as a whole they ring false or unrealistic to the eye, lacking the organic nature of light that the human eye is used to seeing.

Maybe it is just taken too far to demonstrate the DR abilities of the new CMOS technologies?

It is not dissimilar to using lighting and over-doing a back-lit scene by adding too much light to the foreground subject and ending up with that fake "flashed look" even if exposed correctly ... the balance has to be natural looking to the eye, not just numerically or technically balanced.

What would be of real interest is how the DR/Noise is with these CMOS backs when shot at higher ISOs. Perhaps that may be the more useful attribute since shadows really start plugging up, and the color goes all wonky as the ISO goes up?

- Marc

BTW, thanks for the S-007 info.
 

RVB

Member
The same way I could get tons of DR and 16-bits on my 645Z with exposure bracketing. HDRx is basically a technique where the RED camera uses it's incredibly high capture speed overhead to alternate high and low exposures every other frame (via shutter speed), so if you're shooting 24fps for film, the camera is actually capturing 48fps with half the frames having a different exposure level. Obviously this doubles file sizes, but the benefit is that once you load the clip into REDCINE-X you can choose a blending method for the highlight/shadow transitions and have at it. The downside is that the shutter is altered for the different exposure levels, so you risk having blurry highlights if you don't pay attention to the speed for the brighter exposure before shooting.

The Dragon sensor has a "mechanical" 20 stops DR and a usable ~16, so it's not like it really needs exposure bracketing to make use of 16-bit.



James Bond is taking way too long to come to market though.

37.5MP is fine if we're talking about enormous DR and low noise, even by 645Z standards, but it's hard to compete with Sony purely on sensor manufacturing prowess. The spec sheet gives an ISO range of 200-6400, and ISO100 is curiously a pull setting. If the finer sensor process lets CMOSIS cram the electrical bits and produce bigger pixels than the 645Z, in addition to the lower pixel count in the first place, the S should have been capable of 25K ISO at the least. And why list DR as 13 stops? even if they're being extremely conservative, an MF sensor with 6 micron pixels and on-sensor readout should be capable of at least 15 stops or so.

Incidentally, I find it strange that DxO hasn't benchmarked the 645Z yet, despite it's popularity.
The Leica Bond edition.. ;-)

Leica have said they will be applying changes to this sensor compared to the M240 based on what they learned from that sensor so I'm optimistic and may keep my 006 for a while just in case..

Rob
 
The Leica Bond edition.. ;-)

Leica have said they will be applying changes to this sensor compared to the M240 based on what they learned from that sensor so I'm optimistic and may keep my 006 for a while just in case..

Rob
I'm optimistic too, if only for curiosity's sake, as I've already bought the Z. The lower price, higher resolution, proven sensor characteristics, tilt screen and many other bits and bobs it has have cemented it as the way to go for me, but if Leica can knock things out of the park on the image quality front, it'll be all the better for those who invested in the system.
 

Ken_R

New member
Its gonna be a sad day when all high end cameras/backs use similar Sony CMOS Sensors even if they are different sizes. Imagine if back in the day there was only one film available. :(

Anywho. Nice comparisons. For those landscape situations without using ND Grad filters you are definitely at the very cutting edge of what today's sensors can provide in regards to dynamic range. I would say around 14 stops is close to the minimum required to pull those shots off but 15-16 might be ideal. Where you set exposure is also critical obviously to maximize that dynamic range.
 

Uaiomex

Member
It's my understanding that in the last years just about any new DMF back had a Dalsa sensor.
Eduardo

Its gonna be a sad day when all high end cameras/backs use similar Sony CMOS Sensors even if they are different sizes. Imagine if back in the day there was only one film available. :(
 
Its gonna be a sad day when all high end cameras/backs use similar Sony CMOS Sensors even if they are different sizes. Imagine if back in the day there was only one film available. :(
Well, on the other hand, back in the day every film type was available in every size, so if you really liked a kind of film, you could use it in your Leica, Hasselblad and 8x10. Some photographers I personally know used to stick to a couple of film types for pretty much everything they shot regardless of camera - just tons of boxes of the same-kind film.

Who knows what will happen in 5 years time, CMOS could step up to full 645, or maybe a new and different sensor tech will step in altogether, like one that can capture luminance in addition to color.
 

thrice

Active member
The Leica Bond edition.. ;-)

Leica have said they will be applying changes to this sensor compared to the M240 based on what they learned from that sensor so I'm optimistic and may keep my 006 for a while just in case..

Rob
Hi Rob, I must admit when I spoke to Tony at Photokina and he said "exactly the same architecture but different processor" the S007 was still in prototype phase, but sensor changes within 6 months of launch are pretty risky.
 
Hi Rob, I must admit when I spoke to Tony at Photokina and he said "exactly the same architecture but different processor" the S007 was still in prototype phase, but sensor changes within 6 months of launch are pretty risky.
Well, "architecture" is a pretty loose term, using computer chips as an example, Intel's Haswell architecture is utilized in everything from bargain processors powering $500 computers to monstrous $5000+ 18-core chips that can be used in pairs.

Sony's own Exmor architecture has been in use for many years on sensors of every size and complexity from inexpensive camcorders to, supposedly, the 645Z as well.
edit: Since the 645Z is claimed to feature better performance than the other cameras that use the same sensor, it could be said that a lot rests on aspects outside of just the sensor itself. Since the 007 features "a different processor", the difference in performance could be potentially greater than the M240 on a per-pixel level.
 
Last edited:

RVB

Member
Hi Rob, I must admit when I spoke to Tony at Photokina and he said "exactly the same architecture but different processor" the S007 was still in prototype phase, but sensor changes within 6 months of launch are pretty risky.
What do you think of the M240 sensor?? I have a few test files from it and it looks pretty good to my eyes,good DR and decent high ISO,I have a test file shot at 1000iso which is quite good,decent blacks and low noise and takes a push quite well.

I'm also really interested in the live view for shooting cityscapes at night and the new AF system in the 007,it can still improve even though the last firmware update on the 006 was a nice improvement.

I would love to see Leica offer two resolutions in the future,the current res is great for handheld work and the 007's 3.5fps is a very useful but an 80mp sensor camera for landscape would be hard to resist (as long as it wasn't insanely expensive),at some point they'll have to boost the resolution.

Rob
 

thrice

Active member
Well, "architecture" is a pretty loose term, using computer chips as an example, Intel's Haswell architecture is utilized in everything from bargain processors powering $500 computers to monstrous $5000+ 18-core chips that can be used in pairs.

Sony's own Exmor architecture has been in use for many years on sensors of every size and complexity from inexpensive camcorders to, supposedly, the 645Z as well.
edit: Since the 645Z is claimed to feature better performance than the other cameras that use the same sensor, it could be said that a lot rests on aspects outside of just the sensor itself. Since the 007 features "a different processor", the difference in performance could be potentially greater than the M240 on a per-pixel level.
I'm going to have to use the weight of my computing background here to point out that haswell and most other CPU's out there are based on a setup that is very scaleable. They can implement 2 cores or they can implement 18. They can scale the L3 cache to anywhere up to 45MB and the performance jumps are huge.

An imaging sensor is based on a series of quantum wells with a fixed efficiency attached to an array of amplifiers and readout circuitry. That is sensor architecture. Despite still being fabricated from silicon, saying 'architecture' is a much more fixed qualifier in imaging sensors than it is in SoC (system-on-chip) design.

I worked for Sony and I worked for Leica. I will point out that Exmor refers only to the layout of the CMOS DAC and on-chip NR. The sensitivity across the Exmor, Exmor R and Exmor RS series' varies wildly and the sensor size does as well. Dark-current noise and quantum efficiency can vary hugely through tweaking the architecture whilst still being a "Sony Exmor CMOS sensor".

I'm quite certain Tony meant that the underlying architecture was near identical, but in order to hit 37.5MP they will have to change the pixel pitch which can have all kinds of effects. They can alter the CFA, and maybe do some processing tricks. The RAW readout can be tweaked quite significantly before giving you your 'RAW' file, so we'll wait and see.
 
Last edited:

RVB

Member
I'm going to have to use the weight of my computing background here to point out that haswell and most other CPU's out there are based on a setup that is very scaleable. They can implement 2 cores or they can implement 18. They can scale the L3 cache to anywhere up to 45MB and the performance jumps are huge.

With an imaging sensor which is based on a series of quantum wells with a fixed efficiency attached to an array of amplifiers and readout circuitry. That is sensor architecture. Despite still being fabricated from silicon, saying 'architecture' is a much more fixed qualifier in imaging sensors than it is in SoC (system-on-chip) design.

I worked for Sony and I worked for Leica. I will point out that Exmor refers only to the layout of the CMOS DAC and on-chip NR. The sensitivity across the Exmor, Exmor R and Exmor RS series' varies wildly and the sensor size does as well. Dark-current noise and quantum efficiency can vary hugely through tweaking the architecture whilst still being a "Sony Exmor CMOS sensor".

I'm quite certain Tony meant that the underlying architecture was near identical, but in order to hit 37.5MP they will have to reduce the pixel pitch which can have all kinds of effects. They can alter the CFA, and maybe do some processing tricks. The RAW readout can be tweaked quite significantly before giving you your 'RAW' file, so we'll wait and see.
Is the pixel pitch on the S007 not the same as the M240?
 

thrice

Active member
at some point they'll have to boost the resolution.
Rob
Hi Rob,

They're targeting commercial photographers working at f/8 or smaller.
The fear is that they will be compared to a 40x54mm P1 at 80MP and the 30x45mm sensor at 80MP and be found wanting due to diffraction.

FYI the IQ180 hits a diffraction limit at f/8.0, short of digital trickery (Sony have an algorithm they use to try and overcome diffraction) or theoretical negative refractive index materials there is no way around this. No lens will get sharper from f/8.0 to f/11 on a 5.2µm pixel pitch at the pixel level within the plane of sharpness.

The pitch on the current S is 6µm, to get to 60MP which all the sales and marketing guys at Leica begged for would require 4.8µm and 80MP would take 4.1µm. Those are diffraction limits of f/5.6 and ~f/4.5 respectively. Unacceptable in the commercial photography and landscape world. It might amuse you to learn that the Canon 5DS is diffraction limited at f/4.0.
 

thrice

Active member
Is the pixel pitch on the S007 not the same as the M240?
It is 5µm on the M typ240 and the S is 6µm

The well size could be EXACTLY the same to avoid a redesign but they can use gapless microlenses and the larger surface area to effectively gain a larger pixel pitch. The M has very strange and offset microlenses.
 

RVB

Member
Hi Rob,

They're targeting commercial photographers working at f/8 or smaller.
The fear is that they will be compared to a 40x54mm P1 at 80MP and the 30x45mm sensor at 80MP and be found wanting due to diffraction.

FYI the IQ180 hits a diffraction limit at f/8.0, short of digital trickery (Sony have an algorithm they use to try and overcome diffraction) or theoretical negative refractive index materials there is no way around this. No lens will get sharper from f/8.0 to f/11 on a 5.2µm pixel pitch at the pixel level within the plane of sharpness.

The pitch on the current S is 6µm, to get to 60MP which all the sales and marketing guys at Leica begged for would require 4.8µm and 80MP would take 4.1µm. Those are diffraction limits of f/5.6 and ~f/4.5 respectively. Unacceptable in the commercial photography and landscape world. It might amuse you to learn that the Canon 5DS is diffraction limited at f/4.0.
Very interesting..I like to print at around 34"x36" and 37.5MP is more than adequate at this size although for fabrics and other high frequency details a little extra would be nice.

What do you think is the ideal pixel pitch?+-5um might be just right..

Rob
 
Top