SL2-S vs. other cameras - image thread (24 megapixel BSI sensor with Leica lenses)

[Chaemono]

Advertisement (gone after registration)

vor 11 Minuten schrieb la1402:

Have a guess 😀

ISO is 12800. Files are in the download link above.

The Z6 comes with nice PDAF banding, too. 😂

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[la1402]

23 minutes ago, Chaemono said:

The SL2-S doesn’t do that. In order compare noise levels one should, therefore, apply NR and sharpening to the SL2-S files. 

Thanks for posting the article. Interesting. I just wonder - how does it affect daily photography? There is so much done in SW today, also with regards to lens correction, and also by Leica.

BTW, who says the SL2S does not do that?

[la1402]

12 minutes ago, Chaemono said:

The Z6 comes with nice PDAF banding, too. 😂

You see that in the samples I uploaded? Please show.

[BernardC]

36 minutes ago, Chaemono said:

The SL2-S doesn’t do that. In order compare noise levels one should, therefore, apply NR and sharpening to the SL2-S files. 

That's the catch when people claim that they can make one camera look like another in post. It's true, but it means that you have to degrade the better camera's image. No amount of post-processing will restore information that wasn't captured in the first place.

[Chaemono]

vor 1 Minute schrieb la1402:

You see that in the samples I uploaded? Please show.

 "And of course it can be compared, as in the end the output matters." Your quote from #59. So, we'll have a look at that Z6 in-camera RAW processing to see what baggage comes along with its output. Here's what Horshack had to say:

https://www.dpreview.com/forums/post/62354241

The unique aspect of the Z's banding is that it doesn't occur in all images, because unlike banding on other sensors the Z's banding is not a property of the sensor but instead a property of the processing the camera applies to the data coming of the sensor. This causes some to incorrectly conclude they can push an image a certain number of stops without seeing banding (or at certain ISOs), or that the Z's banding is not as bad as reported. What they're really observing is just one of the majority of scenarios that doesn't produce the banding in the first place. It's not a matter of whether you've underexposed+pushed an image a required number of stops to see it (like with traditional banding issues) but rather whether it's there in the image at all. You can actually programmatically detect whether a given raw has banding because the processing Nikon does that causes it leaves behind a unique and unmistakeable anomoly in the raw data.

Based on all the tests I performed there is clearly some sort of algorithm the camera uses to determine when it thinks it needs to apply the processing that ultimately causes the banding. I had a setup where I could control three precise outcomes - one where the banding doesn't occur, one where it occurs a certain percentage of the time, and one where it occurs 100% of the time, all by just changing an exposure parameter by just 1EV in total. You can read about it here, see additional reproduction experiments here ,and read the resulting theory here.

The fact there are exposure ranges where the banding intermittently occurs (which itself is likely is the result of the EV/noise differences straddling the threshold in Nikon's algorithm) means that making conclusions about banding visibility such as ISO thresholds is fraught with error. You make take one shot at ISO 400 and see banding, then not see it at ISO 500, then see it again at 640 - the difference is not the ISO but just the shot-to-shot variability affecting Nikon's detection algorithm that triggers the processing that causes the banding.

One has to take the good with the bad of that in-camere RAW processing in the Nikon Zs. Simple.

[AphotoBerlin]

vor 4 Minuten schrieb Chaemono:

If Sony is using these kinds of tricks to achieve their impressive noise numbers for the a7III, that would, I think, be a shame. This kind of thing is much better done in postproduction.

Isn't everybody doing it? Its not exactly a trick and in general called "dual gain" sensors. Canon does it, Sony does it, Fujifilm does it, Panasonic does and even Leica too. 

My SL2 does it. 3 Steps above base ISO.  If you use the L-Log profile the noise gets worse from ISO400 on (base ISO for L-Log) but at ISO3200 there is a big improvement compared to ISO1600. And I would never be able to smooth the Video like that in post processing that easy. 

I haven't tested it, but the Step without L-Log should be at ISO800 (with base ISO 100). 

 

[Chaemono]

Advertisement (gone after registration)

vor 3 Minuten schrieb AphotoBerlin:

Isn't everybody doing it? Its not exactly a trick and in general called "dual gain" sensors. Canon does it, Sony does it, Fujifilm does it, Panasonic does and even Leica too. 

My SL2 does it. 3 Steps above base ISO.  If you use the L-Log profile the noise gets worse from ISO400 on (base ISO for L-Log) but at ISO3200 there is a big improvement compared to ISO1600. And I would never be able to smooth the Video like that in post processing that easy. 

I haven't tested it, but the Step without L-Log should be at ISO800 (with base ISO 100). 

 

Could be. Depending on the algorithm and the amount of in-camera smoothing applied, the ISO noise will still differ among cameras that basically use a variant of the same 24 MPx sensor. For full disclosure, we should then mention the good and the bad that goes along with it.

[la1402]

3 hours ago, Chaemono said:

 "And of course it can be compared, as in the end the output matters." Your quote from #59. So, we'll have a look at that Z6 in-camera RAW processing to see what baggage comes along with its output. Here's what Horshack had to say:

https://www.dpreview.com/forums/post/62354241

The unique aspect of the Z's banding is that it doesn't occur in....

You did not answer my question, which was wether you saw banding in the files I uploaded. I think we do not need to repeat a general discussion about banding in Nikon Z. I have not seen it in many thousands of pictures, and I do not see it in the ones I took here which served the sole purpose of comparison with the SL2S. So I assume you also did not see it, as it probbaly belongs to "the majority of scenarios that doesn't produce the banding".

I think all manufactures try to optimize signal processing to maximize the results and generate the look they want, as they have little influence on the performance of the sensor itself. And signal processing is in SW since decades. Assuming that the majority of 24MP BSI FF sensors come from the same manufactorer, the main  difference will be in filters and signal processing, and that is ok. The idea that Panasonic/Leica produce some kind of purer signal is not realistic at all.

I am not trying to defend the Z series here - I try to understand the difference in results between it and the SL2S. And I can do that better by looking at  pictures shot at the same scene and comparable settings with both cameras than by following generic discussions. So far I like what I see.

 

 

Edited December 24, 2020 by la1402

[Daedalus2000]

2 minutes ago, la1402 said:

You did not answer my question, which was wether you saw banding in the files I uploaded. I think we do not need to repeat a general discussion about banding in Nikon Z. I have not seen it in many thousands of pictures, and I do not see it in the ones I took here which served the sole purpose of comparison with the SL2S. So I assume you also did not see it, as it probbaly belongs to "the majority of scenarios that doesn't produce the banding".

I think all manufactures try to optimize signal processing to maximize the results and generate the look they want, as they have little influence on the performance of the sensor itself. And signal processing is in SW since decades. Assuming that the majority of 24MP BSI FF sensors come from the same manufactorer, the main  difference will be in filters and signal processing, and that is ok. The idea that Panasonic/Leica produce some kind of purer signal is not realistic at all.

I am not trying to defend the Z series here - I try to understand the difference in results between it and the SL2S. And I can do that better by looking at the same pictures shot with both cameras than by following generic discussions. So far I like what I see.

 

 

I agree with you, there are things we can control and things we cannot control on the image processing pipeline, so comparing output at the level before we apply post processing makes sense to me.

BTW, I downloaded your raw files, sampled parts of the image where the color seems to be uniform and I see that the SL2-S has lower variability so yes it looks a bit cleaner. I cannot quantify how much cleaner, but definitely slightly cleaner.

[Chaemono]

vor 1 Stunde schrieb la1402:

You did not answer my question, which was wether you saw banding in the files I uploaded. I think we do not need to repeat a general discussion about banding in Nikon Z. I have not seen it in many thousands of pictures, and I do not see it in the ones I took here which served the sole purpose of comparison with the SL2S. So I assume you also did not see it, as it probbaly belongs to "the majority of scenarios that doesn't produce the banding".

...

You still don't get. The smooth high ISO performance that you show here has a flipside, arbitrary banding at the whim of the camera. If you don't see, I'll show you.

Edit - The Z7 will do for purpose of illustration.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

Edited December 24, 2020 by Chaemono

[SrMi]

7 hours ago, Chaemono said:

The α7 III doesn’t have an AA filter. I’ll compare ISO noise later.

<snip>

Sony a7 III has a low pass (AA) filter: https://blog.kasson.com/a7iii/sony-a7iii-raw-filtering/

[SrMi]

7 hours ago, Chaemono said:

There’s in camera RAW smoothing going on in the Z6. De facto it means one can apply the same smoothing and sharpening to the SL2-S in LR first and then compare. 

See here: https://blog.kasson.com/nikon-z6-7/nikon-z6-z7-fwc-and-read-noise/

That is the result of the conversion gain (the ratio between the voltage on the pixel and the number of electrons stored there) changing.

Sony is using the same trick with α7 III at ISO 640. See here: 

https://blog.kasson.com/a7iii/sony-a7iii-raw-filtering/

The a7III appears, like many alpha cameras, to use the Aptina DR-Pix conversion gain algorithm, and to switch to high conversion gain at ISO 640. And indeed the ISO 640 plot does not look all that different from the ISO 100 one. The pattern starts to repeat as the ISO goes up.

If Sony is using these kinds of tricks to achieve their impressive noise numbers for the a7III, that would, I think, be a shame. This kind of thing is much better done in postproduction.

I do not see any reference to raw smoothing in the links that you mentioned. 
Conversion gain is a recent technology to increase high ISO performance. I do no see it as a trick but as a useful enhancement of sensors. It seems that Leica SL2 uses conversion gain (at ISO 800), and SL2-S likely has it as well. If an SL2-S owner would work with Bill Claff, we could see the details of the SL2-S sensor.
You cannot replace dual-conversion technology with postproduction.

About dual-conversion gain:

https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

 

[bilbrown]

I have the Lumix S1, Sigma FP and had the original SL until yesterday.. when I added the SL2-S!

 

Let me know if anyone is interested in me posting the 24.6MP flagship cameras shooting portraits with the same lenses, and possibly video.

[Daedalus2000]

1 hour ago, bilbrown said:

I have the Lumix S1, Sigma FP and had the original SL until yesterday.. when I added the SL2-S!

 

Let me know if anyone is interested in me posting the 24.6MP flagship cameras shooting portraits with the same lenses, and possibly video.

Yes please that will be amazing. If you can add any comments on the Sigma FP in terms of performance and overall usability vs the SL2-S it will be great. I value lower weight but I dislike the lack of viewfinder...

[Chaemono]

vor 4 Stunden schrieb SrMi:

Sony a7 III has a low pass (AA) filter: https://blog.kasson.com/a7iii/sony-a7iii-raw-filtering/

He concludes it has a weak one. If one compares the α7III with the α7II, the former produces a lot sharper images. In addition, the α7III shows the same moire as cameras without an AA filter: https://www.smugmug.com/gallery/n-xjDrXB/. To me it looks like it doesn’t have it. 

[Chaemono]

vor 4 Stunden schrieb SrMi:

I do not see any reference to raw smoothing in the links that you mentioned. 
Conversion gain is a recent technology to increase high ISO performance. I do no see it as a trick but as a useful enhancement of sensors. It seems that Leica SL2 uses conversion gain (at ISO 800), and SL2-S likely has it as well. If an SL2-S owner would work with Bill Claff, we could see the details of the SL2-S sensor.
You cannot replace dual-conversion technology with postproduction.

About dual-conversion gain:

https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

 

There is no reference to raw smoothing but to arbitrary in-camera raw processing even at base ISO. No FF camera shows deep shadows banding at base ISO when exposure is pushed by 2.33 stops or shadows are lifted by 100. The Z7 RAW files come out looking really contrasty and sharp. It looks to me like there is a lot more RAW file cooking with the Z cameras than what other manufacturers do. It’s not a far fetched conclusion, therefore, that there is smoothing and sharpening applied to high ISO files. Dual-conversion gain at ISO 800 is just one part of it.

[SrMi]

30 minutes ago, Chaemono said:

There is no reference to raw smoothing but to arbitrary in-camera raw processing even at base ISO. No FF camera shows deep shadows banding at base ISO when exposure is pushed by 2.33 stops or shadows are lifted by 100. The Z7 RAW files come out looking really contrasty and sharp. It looks to me like there is a lot more RAW file cooking with the Z cameras than what other manufacturers do. It’s not a far fetched conclusion, therefore, that there is smoothing and sharpening applied to high ISO files. Dual-conversion gain at ISO 800 is just one part of it.

Hmm...
Every camera does raw processing before saving; I would not call it arbitrary, though. Jim disagrees with how the blue and red channels are pre-scaled; that is all, AFAICT.
AFAIK, it is generally agreed that Z 7 raw files are not cooked (much) or smoothed (noise reduction?). There is a difference in default settings of raw files for Nikon Z cameras, passed on through built-in profiles, but that is not cooked in raw files. There is a reason why Jim Kasson's favorite FF camera is Z 7.
I do not understand why you are bothered by conversion gain? It affects the hardware side, not the camera's internal raw processing. If SL2-S sensor is based on a7rIII, than it is very likely that it also contains dual conversion gain.

[Chaemono]

vor 5 Stunden schrieb SrMi:

...I do not understand why you are bothered by conversion gain? It affects the hardware side, not the camera's internal raw processing. If SL2-S sensor is based on a7rIII, than it is very likely that it also contains dual conversion gain.

I’m sure you know the article, but for the sake of argument let me post a link and an excerpt:

https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

This scheme requires an added transistor and capacitor in the pixel layout, which can eat up real estate that could otherwise be utilized to increase the photodiode size. It also adds metal-line routing for the DCG signal control, which reduces the size of the aperature above the photodiode. These drawbacks create extra challenges as pixels shrink beyond the sub-2 micron mark. Also, this scheme is most useful when the photodiode (PD) size is large enough to be the defining factor for the FW capacity. When the amount of charge held in the PD is so large that it cannot be measured within the pixel’s allowed voltage swing, then extra capacitance must be added to the FD in order to measure all of the PD’s charge. In this case the DCG feature can be added to the pixel design to add the HCG mode for improved low-light performance.

It is interesting to note that not all 24 MPx Sony sensor cameras achieve the same reduction in read noise with dual conversion gain. The α7III’s DCG kicks in at ISO 640, the Z6’s at ISO 800. Now this could be explained by some discretion in the ISO scale and ISO 640 in the Sony is really ISO 800 in the Nikon. But while the α7RIII has an ISO 640 read noise equal to base ISO 100, the Z6 amazingly has a much lower read noise at ISO 800 than at base ISO. Again, it’s the same sensor. Either the Aptina dual gain conversion is implemented differently in the Z6 or there is some additional software smoothing going on in the Nikon. And all I’m saying is, there must be trade-off to achieve this kind of high ISO noise reduction in RAWs.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[SrMi]

8 hours ago, Chaemono said:

I’m sure you know the article, but for the sake of argument let me post a link and an excerpt:

https://www.photonstophotos.net/Aptina/DR-Pix_WhitePaper.pdf

This scheme requires an added transistor and capacitor in the pixel layout, which can eat up real estate that could otherwise be utilized to increase the photodiode size. It also adds metal-line routing for the DCG signal control, which reduces the size of the aperature above the photodiode. These drawbacks create extra challenges as pixels shrink beyond the sub-2 micron mark. Also, this scheme is most useful when the photodiode (PD) size is large enough to be the defining factor for the FW capacity. When the amount of charge held in the PD is so large that it cannot be measured within the pixel’s allowed voltage swing, then extra capacitance must be added to the FD in order to measure all of the PD’s charge. In this case the DCG feature can be added to the pixel design to add the HCG mode for improved low-light performance.

It is interesting to note that not all 24 MPx Sony sensor cameras achieve the same reduction in read noise with dual conversion gain. The α7III’s DCG kicks in at ISO 640, the Z6’s at ISO 800. Now this could be explained by some discretion in the ISO scale and ISO 640 in the Sony is really ISO 800 in the Nikon. But while the α7RIII has an ISO 640 read noise equal to base ISO 100, the Z6 amazingly has a much lower read noise at ISO 800 than at base ISO. Again, it’s the same sensor. Either the Aptina dual gain conversion is implemented differently in the Z6 or there is some additional software smoothing going on in the Nikon. And all I’m saying is, there must be trade-off to achieve this kind of high ISO noise reduction in RAWs.

I assume that you are looking at "Read Noise in DNs vs. ISO setting." On that page:
"These raw values are not appropriate for comparing camera models because they are not adjusted for gain or area."

High ISO noise is about the same in Z 6 as is in a7rIII, see Z 6 vs. A7rIII. There is also some noise reduction applied to very high ISO of a7rIII (triangle down). No noise reduction was measured for Z 6. An example of using noise reduction to lift the DR are the new Canons: R5 and R6.

A manufacturer has a choice at which point to apply the DCG jump. Fuji GFX even changed the DCG point when moving from pre-production (at ISO 800) to the final version (at ISO 500).

BTW, I posted the link to the white paper in #75 .

Happy holidays!

[Chaemono]

Am 25.12.2020 um 03:11 schrieb SrMi:

...There is a reason why Jim Kasson's favorite FF camera is Z 7.

[...]

Actually, now that I got rid of the SL2, I’m keeping the Z7 for situations like these https://www.smugmug.com/gallery/n-7qQZzW/ in case there is some moire with the SL2-S. But just like with the α7III, visible moire with the SL2-S only shows up in heavily underexposed dark fabrics when shadows are lifted later. It must have a very weak AA filter because I was stunned how little moire showed in the second picture here: https://www.smugmug.com/gallery/n-qs73fM/. It was well exposed. 

The Z7 also works really well with the 50 and the 75 M lenses I have.