When will (if ever) Leica M get something like Canon DGO sensor?

[jaapv]

Advertisement (gone after registration)

13 hours ago, FlashGordonPhotography said:

*Drastically* is in the eye of the beholder but the M10R does have noticeably improved DR over the M10, plus less noise and more megapickles. I've tested them sided by side and the DR improvements are real.

But there's always going to be scenes with more DR than any camera can handle.

Gordon

We haven’t seen Pickles for a while, let alone the Mega variety

[pgk]

22 hours ago, farnz said:

There is more DR in the picture than even your eyes, with roughly 24 stops of DR, could handle so expecting a camera to be capable of recording the scene without blowing the highlights or the shadows or both is unrealistic even if the camera is specced at 16-stops.  

The eye/brain systems is an 'integrating' system. It handles dynamic range by assessing different areas through movements of the eye and experience. Trying to mimic this photographically will inevitably produce an 'HDR' type appearance. I suspect that this is where things are moving to though because I see an increase in the posting of over saturated and highlight/shadow detailed photographs online, which I find somewhat unnatural. But tastes change as technology and normalities shift. I personally prefer the imagery produced by current cameras. Trying to compress dynamic range into a print or onto a monitor has its limitations for me.

[Chaemono]

Am 7.10.2020 um 06:32 schrieb adan:

Every camera company has their own particular needs.

Leica, especially for the M, has a core need for sensor compatability in sharpness, vigetting and color-stains elimination, with legacy M film lenses. In a niche sales volume.

That means they cannot necessarily chase after the newest and greatest tech (often patented) from other sensor-makers, unless it also fulfills that niche requirement.

Leica may well be able to incorporate such advances as they mature, and Leica (and CMOSIS) figure out how to integrate them.

That’s why we don’t have an M with a 24 MPx BSI sensor like the one found in the S5, yet.

On better DR range of FF cameras, M10-R may have more than the M10-P, but the latter has better ‘push-ability’ of files with less pushed shadows noise. So, to even the ‘DR’ field with the M10-R, one needs to expose more for highlights with the M10-P and recover for the shadows.

There is no way a 40+ MPx FF color camera can match the low shadow noise of pushed 24 MPx FF files. I’ve tried the α7R III/IV and the M10-R and decided to keep the M10-P. The most exciting FF camera in terms this exposure latitude is the S5, IMO, where pushed shadows noise is virtually non-existent up to ISO 6400. Smearing in the corners with M lenses up to 50 mm, I’m afraid, but there are always the Summicron-SL lenses one can use with it. 

[thrid]

The vast majority of movies you see (saw?) in theaters are shot on Arri Alexa cameras, made by Arri in Munich.

The Alexa uses a sensor with a DGO architecture. Basically the camera reads the gain twice on each receptor well. Once for a normal exposure and simultaneously for highlights. These two bracketed exposures are combined into a single frame.

The output of the Alexa is rated at a useable 14-15 stops. It is suspected that the Alexa actually captures much more than that with its dual gain architecture, but that Arri uses the extra data to smoothly roll off the highlights into the 14-15 stops contained in the file. The camera is legendary for its smooth film like  rolloff in the highlights.

I own an Alexa and from my own personal experience it is close to impossible to produce nasty white clipped highlights like you see with single gain cameras. When the Alexa eventually does overexpose it fails gracefully like film did in a very pleasing and organic way.

Yes, there is no projection system short of maybe some HDR systems that can actually show 14-15 stops of captured data. Your standard rec709 television or SRGB monitor can only display about 8 stops of image and print maybe less. But that's not the point. The entire music industry works with sampling rates far higher than anyone can hear. The point is to have the extra data to manipulate in case you want to recover highlights etc and provide overall higher image quality.

I've been advocating for a dual gain architecture sensor in the M-series, along with 16 bit files for a long time. Maybe now that Leica is taking a two pronged approach, with high-res and medium res models they will consider such an option. Sony has been quite successful with a similar strategy in the FS lineup.

 

 

 

 

[Matlock]

1 hour ago, thrid said:

The vast majority of movies you see (saw?) in theaters are shot on Arri Alexa cameras, made by Arri in Munich.

 

Interesting that a growing number of movies, including the latest Bond film, are shot on film.

[lct]

Is Arri still making film cameras? Just curious.

[adan]

Advertisement (gone after registration)

Speaking of numbers, a point to keep in mind when giving Leica Camera "helpful" suggestions.

Canon - worldwide employees - 197673

Sony - worldwide employees - 114400

Fujifilm - worldwide employees - 31844

Nikon - worldwide employees - 25729

Leica Camera - worldwide employees - 1800

Edited October 10, 2020 by adan

[Matlock]

22 minutes ago, lct said:

Is Arri still making film cameras? Just curious.

I believe they stopped making film cameras in the very early part of this century, 2002?.

 

[mirekti]

28 minutes ago, adan said:

Speaking of numbers, a point to keep in mind when giving Leica Camera "helpful" suggestions.

Canon - worldwide employees - 197673

Sony - worldwide employees - 114400

Fujifilm - worldwide employees - 31844

Nikon - worldwide employees - 25729

Leica Camera - worldwide employees - 1800

What do these numbers mean in relationships to “helpful” suggestions. Leica wouldn’t make the sensor themselves after all.

[thrid]

On 10/10/2020 at 12:10 PM, lct said:

Is Arri still making film cameras? Just curious.

 As far as I know they stopped production of film cameras a few years ago.

I'm guessing you can still get them serviced. and Arri Rental still carries them

[thrid]

Arri is a pretty small company. The Alexa sensor wasn't designed in-house, but to their specifications by ON Semiconductor. 

Leica does the same thing and may actually have an advantage over Arri due to their partnership with Panasonic / Sigma. Wasn't the sensor in the SL2 a joint venture that is used in multiple cameras? The Alexa sensor is only used by Arri and probably in smaller numbers than Leica sells M cameras...

[SrMi]

On 10/8/2020 at 5:57 AM, Chaemono said:

That’s why we don’t have an M with a 24 MPx BSI sensor like the one found in the S5, yet.

On better DR range of FF cameras, M10-R may have more than the M10-P, but the latter has better ‘push-ability’ of files with less pushed shadows noise. So, to even the ‘DR’ field with the M10-R, one needs to expose more for highlights with the M10-P and recover for the shadows.

There is no way a 40+ MPx FF color camera can match the low shadow noise of pushed 24 MPx FF files. I’ve tried the α7R III/IV and the M10-R and decided to keep the M10-P. The most exciting FF camera in terms this exposure latitude is the S5, IMO, where pushed shadows noise is virtually non-existent up to ISO 6400. Smearing in the corners with M lenses up to 50 mm, I’m afraid, but there are always the Summicron-SL lenses one can use with it. 

Have you compared noise of 40+ vs 24Mpx at pixel level, or when resized to the same dimension?

[Chaemono]

vor 3 Stunden schrieb SrMi:

Have you compared noise of 40+ vs 24Mpx at pixel level, or when resized to the same dimension?

Not the way you imply, i.e. well lighted scenes. I usually test scenes with extreme highlights and shadows, expose for the highlights and lift the shadows to the max. All modern FF sensors do well with respect to pushed shadows noise/banding at base ISO, including 40+ MPx ones, the SL2 better than the α7R III/IV. The higher the ISO, however, the more the 24 MPx sensors pull ahead for pushed shadows noise/banding, and the more the BSI sensors pull ahead, i.e. α7R III/IV do better than SL2.

[SrMi]

5 hours ago, Chaemono said:

Not the way you imply, i.e. well lighted scenes. I usually test scenes with extreme highlights and shadows, expose for the highlights and lift the shadows to the max. All modern FF sensors do well with respect to pushed shadows noise/banding at base ISO, including 40+ MPx ones, the SL2 better than the α7R III/IV. The higher the ISO, however, the more the 24 MPx sensors pull ahead for pushed shadows noise/banding, and the more the BSI sensors pull ahead, i.e. α7R III/IV do better than SL2.

I did not imply any lighting. 
Assume you take the same exposure, of the same scene, with a +40MPx camera and a 24MPx with similar sensor technologies. I have not seen less noise in 24MPx cameras when compared at the same dimension.

Your observation differs from mine. That is why I wondered whether you compared at the pixel level (100% view) or after resizing the +40Mpx image to the same resolution of the 24Mpx camera.

[jaapv]

19 hours ago, SrMi said:

I did not imply any lighting. 
Assume you take the same exposure, of the same scene, with a +40MPx camera and a 24MPx with similar sensor technologies. I have not seen less noise in 24MPx cameras when compared at the same dimension.

Your observation differs from mine. That is why I wondered whether you compared at the pixel level (100% view) or after resizing the +40Mpx image to the same resolution of the 24Mpx camera.

The sensors cannot be the same as the sensels will have a different size.The larger ones will have a better quantum efficiency thus less noise per pixel. Your observation is contradicted by physics. Do consider that the sensor output is processed in the camera before the raw file is written. The in-camera noise reduction may well give the illusion of similar noise, but the noise reduction will show up in other subtle ways, like colour shift or loss of micro contrast.
Another way to equalize the noise is to refine the sensor, for instance using more pure Silicon, or improved metal compounds for the wiring on the sensor, even when using the same architecture, but that is quite expensive.

[SrMi]

4 hours ago, jaapv said:

The sensors cannot be the same as the sensels will have a different size.The larger ones will have a better quantum efficiency thus less noise per pixel. Your observation is contradicted by physics. Do consider that the sensor output is processed in the camera before the raw file is written. The in-camera noise reduction may well give the illusion of similar noise, but the noise reduction will show up in other subtle ways, like colour shift or loss of micro contrast.
Another way to equalize the noise is to refine the sensor, for instance using more pure Silicon, or improved metal compounds for the wiring on the sensor, even when using the same architecture, but that is quite expensive.

To clarify, I was talking about similar sensor technologies, like Nikon Z 6 vs. Z 7.
All sensels have the same noise as I understand it, but the signals are weaker with smaller sensels (thus lower SNR, what you probably meant by noise). However, that is compensated by having more sensels. In the end, it is the sensor size that matters more than the sensel size (when viewed at the same output size).

https://www.dpreview.com/articles/5365920428/the-effect-of-pixel-and-sensor-sizes-on-noise/2

 

[jaapv]

Not so fast -

 

https://dtmateojr.wordpress.com/2014/03/09/noise-performance-comparisons/

[farnz]

1 hour ago, jaapv said:

Not so fast -

 

https://dtmateojr.wordpress.com/2014/03/09/noise-performance-comparisons/

Apologies, Jaap, and it's in no way detrimental to the point you were putting, but I just had to pull out this little gem from Demo Mateo's:

" ... In the end, the crazy consumers do the justification in behalf of camera manufacturers in the form of heated arguments in forums. ..."

I don't think I've seen such an overstated belief in forum opinion's/blog's worth and reach.  I suspect that he had to grind his teeth to stop himself from including the term "influencer".  

Faced with detailed business cases and cost justifications I think it highly unlikely in the extreme that Sony, Canon, Nikon and the rest would pay a microsecond's attention to 'heated forum arguments' etc when selecting their next technology direction or product strategy.  (Oh, I'm still giggling. ) 

Right ... on with the discussion.

Pete.

[SrMi]

1 hour ago, jaapv said:

Not so fast -

 

https://dtmateojr.wordpress.com/2014/03/09/noise-performance-comparisons/

That article confirms what I wrote: higher resolution has higher image noise at pixel size, but (almost) the same noise at the same output dimension. What Mateo wrote is even more "damning" for lower resolution sensors:  If you downsample the D800 it will outperform the D700.

What I disagree with Mateo is whether pixel level or output size matters.

 

[Advertisement]

Hallo,

Look here: Leica M

[John Miranda]

I think that individual pixel sensor size and its S/N varies on various variables over time due to technology advancements beyond the theoretical notion that a larger pixel level sensor well captures more photons. I recall a 2005 era camera forum arguing that Nikons 12MP D2X must have more noise than the 6MP D100 model.., then the same for the 24MP D3x, then the same for the 36MP D800, and so on.... and today, the D10R, in every aspect, blows away the Nikon D3X in dynamic range, noise, and MP, both having the same sensor size.

I believe better fabrication technology, IC packaging, processing and overall system technology allows for improvement over time, which has given us a rising MP while improving S/N and dynamic range in the past 20 years of digital photography.

  The current rate of improvement suggests that curve of improvement is not flattened quite yet.  It is an AND, not an OR.   So let us enjoy it!

 

Edited October 14, 2020 by John Miranda