ISO performance on SL2-s vs SL2

[kobra]

Advertisement (gone after registration)

Sean Reid from reidreviews.com has tested the SL, SL2 and SL2-S. His testing has demonstrated that on the original SL, ISO50 is indeed the native base ISO, whereas on the SL2 and SL2-S, ISO100 is the native base and ISO50 is a "pull". He sees no advantage to using ISO50 on the SL2 and SL2-S. 

I'm not an engineer, so while I've followed this thread I still don't understand the logic behind why higher ISOs would offer better either IQ, or greater DR, or protect highlights better. Reid's ISO testing is not just theoretical to establish "native base ISO" but to establish the widest DR and the best "highlight headroom". Again, he has tested this to be ISO100 for the newer SL cameras. So why would one intentionally raise ISO above that which produces the best DR and highlight headroom (ISO100)? 

Note, Reid does comment that an RGB histogram would be helpful to ensure that none of the color channels are clipped. 

Brad

[Advertisement]

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

[SrMi]

1 hour ago, BernardC said:

That page says the same thing that others have been arguing: "(ISO 100) does not offer a material advantage over ISO 200".

Yes

1 hour ago, BernardC said:

I'm not sure what he means by "engineering dynamic range." Is it a real thing, or his own name for something else?

Engineering dynamic range is a standard for measuring dynamic range with an SNR=1 criterion.

1 hour ago, BernardC said:

The fact that he found only a third of a stop of difference points toward experimental error. As we know, 1/3 stop is the minimum noticeable difference.

The DR difference has been measured in several cases (including P2P) and has been observed by LUF members, including me.

[SrMi]

3 hours ago, BernardC said:

Did I miss the part where you showed that? Do you mean that 100 and 3200 use different gain circuits, or that 100 is different from 200?

It showed that ISO 800 to ISO 12500 is invariant but not when going from ISO 100 to ISO 1600. The switch occurs at ISO 800.

 

3 hours ago, BernardC said:

Have you tried the test that I suggested: pick a manual exposure, take a series of images at that exposure, switching the ISO between shots.

I do that a lot but did not reply as I did not see any relevance to our disagreed points. If the ISO does not move over the DCG point in SL2-S, there is no relevant difference in noise. The test shows that exposure determines the noise, not the ISO. What are you seeing in the test?

3 hours ago, BernardC said:

PtP isn't measuring what they think they are. Their method isn't well suited to measuring dynamic range. 

I was not referring to PDR measurements in P2P because they say nothing about ISO invariance. To understand the ISO invariance ranges, you have to look at the Shadow Improvement vs. ISO setting:

https://www.photonstophotos.net/Charts/PDR_Shadow.htm#Leica SL2-S

If the graph is flat, the range is invariant.

3 hours ago, BernardC said:

I also find it strange that their DR numbers go down by almost a full stop for every extra stop of ISO.

 As your test shows, the DR/noise is determined by exposure (shutter speed and aperture). As you increase ISO, the maximum possible exposure before clipping is reduced. Hence the drop in DR.

 

[BernardC]

1 hour ago, SrMi said:

Engineering dynamic range is a standard for measuring dynamic range with an SNR=1 criterion.

So that's what Arri calls Photographic Dynamic Range (page 12):

https://www.arri.com/resource/blob/295460/e10ff8a5b3abf26c33f8754379b57442/2022-09-28-arri-dynamic-range-whitepaper-data.pdf

You'll find on the same page that the Alexa 35 uses a dual gain sensor, with the higher ranger starting at EI 2560. That camera wasn't available when the DPR article was written.

[hansvons]

8 hours ago, kobra said:

I'm not an engineer, so while I've followed this thread I still don't understand the logic behind why higher ISOs would offer better either IQ, or greater DR, or protect highlights better. Reid's ISO testing is not just theoretical to establish "native base ISO" but to establish the widest DR and the best "highlight headroom". Again, he has tested this to be ISO100 for the newer SL cameras. So why would one intentionally raise ISO above that which produces the best DR and highlight headroom (ISO100)? 

Of course, Sean Reid is right in his assessment that the SL2-S' base ISO is ISO 100. I like to see it this way: low ISO protects shadows, and higher ISO highlights in a raw workflow. I hate clipped whites if they are avoidable. Others hate the faintest noise but don't mind burned highlights. Ultimately, something has to give when one chooses the extreme end of ISO selection, e.g. ISO 100 vs ISO 3200. The middle road offers some buffer on both ends. For me, that's ISO 400-800. I can retain highlights and bring up shadows without exhibiting too much noise. At ISO 100, keeping highlights is only possible if the exposure is perfectly nailed (which I rarely accomplish in real life). To mitigate that issue at low ISO, Leica introduced highlight-weighted metering. 

[kobra]

4 hours ago, hansvons said:

Of course, Sean Reid is right in his assessment that the SL2-S' base ISO is ISO 100. I like to see it this way: low ISO protects shadows, and higher ISO highlights in a raw workflow. I hate clipped whites if they are avoidable. Others hate the faintest noise but don't mind burned highlights. Ultimately, something has to give when one chooses the extreme end of ISO selection, e.g. ISO 100 vs ISO 3200. The middle road offers some buffer on both ends. For me, that's ISO 400-800. I can retain highlights and bring up shadows without exhibiting too much noise. At ISO 100, keeping highlights is only possible if the exposure is perfectly nailed (which I rarely accomplish in real life). To mitigate that issue at low ISO, Leica introduced highlight-weighted metering. 

Thanks @hansvons !

I guess where I'm struggling to understand is how "native base (ISO100)" will have the widest DR yet somehow is more in danger of highlight clipping... isn't the definition of DR basically the measurement from the usable noise floor at the dark end (clean shadows) up to the max highlight headroom at the bright end (ie, just before clipping)? Everything I thought I knew was that raising ISO above base will almost always decrease DR... and if so then I don't see how raising ISO can help with highlight clipping?? 

So, I know you've stated the above before but I'm still not getting the logic behind it. Can you send links to other info to help me understand? Thanks! 

Brad

[SrMi]

Advertisement (gone after registration)

1 hour ago, kobra said:

Thanks @hansvons !

I guess where I'm struggling to understand is how "native base (ISO100)" will have the widest DR yet somehow is more in danger of highlight clipping... isn't the definition of DR basically the measurement from the usable noise floor at the dark end (clean shadows) up to the max highlight headroom at the bright end (ie, just before clipping)? Everything I thought I knew was that raising ISO above base will almost always decrease DR... and if so then I don't see how raising ISO can help with highlight clipping?? 

So, I know you've stated the above before but I'm still not getting the logic behind it. Can you send links to other info to help me understand? Thanks! 

Brad

You are correct (still photography). ISO 100 has no more danger of highlight clipping than any other ISO (except the weird ISO 50).

Anyone can run an experiments to verify it.

[LocalHero1953]

@BernardC has posted this graphic before, and I find it helpful to understand his and @hansvons comments. Yes, it's from a video source, but I have not heard why it shouldn't be relevant to stills.

 

[Stuart Richardson]

10 hours ago, hansvons said:

Of course, Sean Reid is right in his assessment that the SL2-S' base ISO is ISO 100. I like to see it this way: low ISO protects shadows, and higher ISO highlights in a raw workflow. I hate clipped whites if they are avoidable. Others hate the faintest noise but don't mind burned highlights. Ultimately, something has to give when one chooses the extreme end of ISO selection, e.g. ISO 100 vs ISO 3200. The middle road offers some buffer on both ends. For me, that's ISO 400-800. I can retain highlights and bring up shadows without exhibiting too much noise. At ISO 100, keeping highlights is only possible if the exposure is perfectly nailed (which I rarely accomplish in real life). To mitigate that issue at low ISO, Leica introduced highlight-weighted metering. 

I understand that, but drew a slightly different conclusion. I learned photography shooting slides, so I am more comfortable exposing for highlights and I tend to like dark shadows. So for me I stick to base ISO mostly and often shoot with -.5 or -1 exposure compensation. Nailing the exposure is not too hard for me, but that is also because a lot of what I do is not fast paced. I have not found that to be the same as shooting ISO 200 without compensation, but it may well be! I wonder if people who were used to negative film would prefer the method you mention. Color negative in particular is best shot with more generous exposure.

Edited April 14, 2023 by Stuart Richardson

[jaapv]

Fortunately we do have EVFs now which allow a histogram in the viewfinder and exposure compensation under our thumb. I find that, depending on the subject, optimal exposure can vary from -3 to +2. 

I might add, that ISO value is not a valid way to control exposure and prevent blown highlights or noise.  It does not alter the physical properties of the sensor, nor its sensitivity. If you hit a pixel with more photons than it can handle it will be blown out. If there are fewer photons than it can record it will be below the noise floor.  The only way you can control the number of photons hitting the sensor is by aperture and shutter speed. 

[hansvons]

14 minutes ago, Stuart Richardson said:

I understand that, but drew a slightly different conclusion. I learned photography shooting slides, so I am more comfortable exposing for highlights and I tend to like dark shadows. So for me I stick to base ISO mostly and often shoot with -.5 or -1 exposure compensation. Nailing the exposure is not too hard for me, but that is also because a lot of what I do is not fast paced. I have not found that to be the same as shooting ISO 200 without compensation, but it may well be! I wonder if people who were used to negative film would prefer the method you mention. Color negative in particular is best shot with more generous exposure.

Great point, Stuart, as it shows that this isn't much of a scientific/engineering discussion (to which I cannot contribute anything insightful anyway) but rather the question of how to handle ISO regarding exposure. And here, as you pointed out very well, things become personal. I exposed only about 500 slides in my life but a couple of kilometres of Kodak negs. Plus, I'm used to working with digital cinema footage, which is log-based and somewhat close to film negatives in behaviour, with the crucial exception that highlights exposure is not as forgiving as with film and needs careful consideration (although camera manufacturers work relentlessly to improve highlights roll-off and highlights exposure). 

Whether you compensate via EV, dialling up ISO, or closing the aperture is the same on a raw-based workflow (as long as the sensor behaves invariant). I'm the ISO guy, as I have never exposed negs at box speed but tend to expose them fatter, which means using a lower ISO in the light meter. With B&W, home souping etc., that's somewhat different, but protecting shadows is the best practice with ENC2 and the usual Kodak/Fuji candidates. Hope that makes sense!

17 minutes ago, jaapv said:

Fortunately we do have EVFs now which allow a histogram in the viewfinder and exposure compensation under our thumb. I find that, depending on the subject, optimal exposure can vary from -3 to +2. 

That's right, of course. But I often find myself in a fast-paced situation and thus need some leeway. We are all different animals with different hunting grounds. 

[jaapv]

6 minutes ago, hansvons said:

Great point, Stuart, as it shows that this isn't much of a scientific/engineering discussion (to which I cannot contribute anything insightful anyway) but rather the question of how to handle ISO regarding exposure. And here, as you pointed out very well, things become personal. I exposed only about 500 slides in my life but a couple of kilometres of Kodak negs. Plus, I'm used to working with digital cinema footage, which is log-based and somewhat close to film negatives in behaviour, with the crucial exception that highlights exposure is not as forgiving as with film and needs careful consideration (although camera manufacturers work relentlessly to improve highlights roll-off and highlights exposure). 

Whether you compensate via EV, dialling up ISO, or closing the aperture is the same on a raw-based workflow (as long as the sensor behaves invariant). I'm the ISO guy, as I have never exposed negs at box speed but tend to expose them fatter, which means using a lower ISO in the light meter. With B&W, home souping etc., that's somewhat different, but protecting shadows is the best practice with ENC2 and the usual Kodak/Fuji candidates. Hope that makes sense!

That's right, of course. But I often find myself in a fast-paced situation and thus need some leeway. We are all different animals with different hunting grounds. 

We are fortunate to live in a time when our gear affords us such leeway.

[kobra]

12 hours ago, LocalHero1953 said:

@BernardC has posted this graphic before, and I find it helpful to understand his and @hansvons comments. Yes, it's from a video source, but I have not heard why it shouldn't be relevant to stills.

 

Thanks Paul, 

I admit as much as I have been following this thread and that one, I still don't understand. (Also, on the other thread, there was talk of "middle grey" which I think is not used in stills photography?)

Either way there has been a lot of technical discussion that went over my head, but even worse were people disagreeing on even the basics... For example, from my layman's view and relating to stills photography, here's my summary (hope it's accurate) of 2 different theories expressed:

Theory #1) base ISO (ISO100 in the SL2 and SL2-S) always gives the maximum sensor DR available. Comments have suggested that if you raise ISO, such as to ISO400, you are wasting stops of DR, so you have all the extra weight of a full frame system with the DR of something like an M4/3 system. Message -- always use base ISO100 whenever there is enough light to do so.

Theory #2) if you use base ISO you will be more likely to blow highlights. So, instead use a middle ISO, like ISO800, because I guess that will force you to under expose the sensor by 4 stops while raising the ISO amplification to match. Thus, because you have severely reduced the actual light reaching the sensor, you are less likely to have blown highlights or you have more room to recover them (at least that's the only sense I can make of it). Further justification is that with modern sensors, there won't be much noise in the shadows at ISO800 anyway. Message -- don't use base ISO as you will more easily blow highlights. Instead, use ISO800 to have more latitude that will protect highlights and don't worry about the shadow noise as it won't be bothersome anyway. 

I admit to being confused because the above are examples of what seems to be completely opposing views. Do I do what I've been doing, and always use the lowest possible ISO to maximize the light reaching the sensor? Or am I missing something by not raising ISO by a few more stops? 

And, related to THIS thread and why I am asking for clarity on this topic... if I cannot find agreement on the basics of ISO and which setting to use, how can I understand the more technical differences between the SL2 and SL2-S? Differences such as "ISO invariant vs non-invariant", or "dual gain", or "scaling"? So that is why I am hoping those with more expertise in these matters can help us layman to get the basics down before we see how the more technical differences impact the images we are trying to make. 

Specifically why that matters to me? I have an SL2 and am disappointed in the low light/high ISO situations, so am trying to determine if there is enough of a difference in the SL2-S to get that as a second body now; or do I make do with the excellent SL2 and adjust my technique somewhat and not worry about the SL2-S? 

I am always ready to learn, so further comments to help clarify are very welcome. And I do hope there are others that will benefit from further discussion of this as well. 

Thanks again! 

Brad

[LocalHero1953]

18 minutes ago, kobra said:

I am always ready to learn, so further comments to help clarify are very welcome. And I do hope there are others that will benefit from further discussion of this as well.

As I have posted elsewhere, I am averse to testing, so if a camera matches my needs in practice, then I don't worry about it!

As for your particular concerns, I can understand the arguments, and have reached my own conclusions on them, but I am qualified neither academically nor by practical testing and experience to weigh in.

[hansvons]

1 hour ago, kobra said:

So, instead use a middle ISO, like ISO800, because I guess that will force you to under expose the sensor by 4 stops while raising the ISO amplification to match.

This is where, at times, the discussion starts to become complex because some people believe the sensor in question (SL2-S) is invariant until approx ISO 1600/3200. Thus, no signal amplification takes place (that's me). While others think differently and believe that there's amplification going on at a much lower ISO. If the latter is the case, the base ISO delivers the highest DR. If the sensor is invariant over 5 stops, e.g. from ISO 100 to ISO 1600, the DR remains the same regardless of ISO because the sensor's signal isn't affected by amplification. 

If a sensor is invariant, one can dial up ISO by four stops and close the aperture by four stops, leading to an identical image. If the sensor is not invariant and one sets different ISO values that inevitably lead to amplification, the exercise, as mentioned earlier, will lead to different results in colour, DR and noise. And then, of course, base ISO wins in terms of IQ.

But I find all of that somehow moot as I shoot my stuff as I want it, and others shoot as they please to do.

1 hour ago, kobra said:

Specifically why that matters to me? I have an SL2 and am disappointed in the low light/high ISO situations, so am trying to determine if there is enough of a difference in the SL2-S to get that as a second body now; or do I make do with the excellent SL2 and adjust my technique somewhat and not worry about the SL2-S? 

I can't add much wisdom to that because I'm me, and you are you. I bought the SL2-S because I don't need many pixels. Others have different use cases and need many pixels. However, the more pixels, the smaller they become at a given sensor size. Larger pixels collect light better than smaller pixels, which is somewhat similar to film grain which gets more coarse the more sensitive the film material is.

That can be irrelevant in everyday light situations. But having some leeway is a good thing when things get a bit tighter. In cinematography, moody light is, by definition, not an everyday situation and is often crucial to the story. The better your sensor or your film stock can deal with low-light situations, the easier/better setting up moody lighting becomes (and budget-friendlier). That's why Arri developed a sensor with larger pixels for their cine camera Alexa than the competition because they knew that their clients (directors and DOPs) prefer juicy shadows and light sensitivity over resolution - by a large margin. 

All of that can be quite different for stills photography - or not. As usual, it depends. For me, having some extra leeway in the shadows is important. That can be very different for you. I'd give the SL2-S a test drive and then decide. 

[kobra]

1 hour ago, hansvons said:

This is where, at times, the discussion starts to become complex because some people believe the sensor in question (SL2-S) is invariant until approx ISO 1600/3200. Thus, no signal amplification takes place (that's me). While others think differently and believe that there's amplification going on at a much lower ISO. If the latter is the case, the base ISO delivers the highest DR. If the sensor is invariant over 5 stops, e.g. from ISO 100 to ISO 1600, the DR remains the same regardless of ISO because the sensor's signal isn't affected by amplification. 

If a sensor is invariant, one can dial up ISO by four stops and close the aperture by four stops, leading to an identical image. If the sensor is not invariant and one sets different ISO values that inevitably lead to amplification, the exercise, as mentioned earlier, will lead to different results in colour, DR and noise. And then, of course, base ISO wins in terms of IQ.

But I find all of that somehow moot as I shoot my stuff as I want it, and others shoot as they please to do.

I can't add much wisdom to that because I'm me, and you are you. I bought the SL2-S because I don't need many pixels. Others have different use cases and need many pixels. However, the more pixels, the smaller they become at a given sensor size. Larger pixels collect light better than smaller pixels, which is somewhat similar to film grain which gets more coarse the more sensitive the film material is.

That can be irrelevant in everyday light situations. But having some leeway is a good thing when things get a bit tighter. In cinematography, moody light is, by definition, not an everyday situation and is often crucial to the story. The better your sensor or your film stock can deal with low-light situations, the easier/better setting up moody lighting becomes (and budget-friendlier). That's why Arri developed a sensor with larger pixels for their cine camera Alexa than the competition because they knew that their clients (directors and DOPs) prefer juicy shadows and light sensitivity over resolution - by a large margin. 

All of that can be quite different for stills photography - or not. As usual, it depends. For me, having some extra leeway in the shadows is important. That can be very different for you. I'd give the SL2-S a test drive and then decide. 

Thanks @hansvons - the above is helpful, and challenges some of my prior assumptions. 

I think your advice is valid - I am going to try an SL2-S and use it for some of my photography. On paper, I'm worried that I will lose megapixels but on the other hand I don't know if that will make enough of a practical difference to worry about. What I do expect is to gain a lot on low light situations. Thanks again, this is enough for me to order the SL2-S and try it out! 

Brad

[kobra]

2 hours ago, LocalHero1953 said:

As I have posted elsewhere, I am averse to testing, so if a camera matches my needs in practice, then I don't worry about it!

As for your particular concerns, I can understand the arguments, and have reached my own conclusions on them, but I am qualified neither academically nor by practical testing and experience to weigh in.

Thanks @LocalHero1953 - I appreciate you weighing in. I hope my posts don't come across as overly concerned or frustrated; in fact not at all. I seek to understand, and this thread has been helpful.

I feel like you in that I don't feel qualified to fully understand, but I also am averse to testing so rely on my own experiences in the end. It's why I ended up making a major change to the Leica SL system from my prior system - no amount of comparing and reading of others experiences was equal to me getting the system and starting to use it. In the end, I love so much about the system that I know it was the right choice. However, also from my experience, I suspect if I went back a few months in time I would likely have chosen the SL2-S over the SL2. 

The comments from you and others have been helpful enough to push me to order an SL2-S. I did purchase my SL2 used, so if I do like the SL2-S a lot more, I may sell the SL2 and hopefully not lose too much. Or I may keep them both as long as I can get it past my live-in bookkeeper  

Brad

[helged]

3 hours ago, kobra said:

Thanks @LocalHero1953 - I appreciate you weighing in. I hope my posts don't come across as overly concerned or frustrated; in fact not at all. I seek to understand, and this thread has been helpful.

I feel like you in that I don't feel qualified to fully understand, but I also am averse to testing so rely on my own experiences in the end. It's why I ended up making a major change to the Leica SL system from my prior system - no amount of comparing and reading of others experiences was equal to me getting the system and starting to use it. In the end, I love so much about the system that I know it was the right choice. However, also from my experience, I suspect if I went back a few months in time I would likely have chosen the SL2-S over the SL2. 

The comments from you and others have been helpful enough to push me to order an SL2-S. I did purchase my SL2 used, so if I do like the SL2-S a lot more, I may sell the SL2 and hopefully not lose too much. Or I may keep them both as long as I can get it past my live-in bookkeeper  

Brad

Having used SL2 since it was released, and then SL2-S, I ended up selling the SL2 and have since used the S-version as the main FF body. SL2 was (and is) excellent. But I prefer the files from SL2-S: Somewhat more natural colours, improved high ISO performance, less tendency for highlight clipping, possibly faster/more reliable af (but still lagging the af from eg Canon and Nikon), etc. - everything based on my eyes & experience. High resolution files I get from S3 (and S006, the latter I love due to its colour rendering). For me, SL2-S is the most versatile of the SL bodies up to now. I guess SL3 may change this, but time will tell... 

[Advertisement]

Hallo,

Look here: Leica L-Mount

[LocalHero1953]

To @kobra

I bought the SL2 around August 2020, to replace my SL. I used it for four months before the SL2-S was launched. I traded in the SL2 (Leica Mayfair offered me a very good deal) based on the reviews of people I trust, and I've been very happy that my switch to the SL2-S was correct. The underlined text is the key part. I don't read every review (I never watch video reviews). There are a number of reviewers, and especially members who post here, whom I have come to trust over the years; I find that their sort of photography chimes with mine, they look for the same features and behaviours, and they identify big advances and failings that concern me. I ignore the other reviewers, because they are irrelevant - to me.

If you struggle with the technical aspects of camera choice and comparative arguments, I suggest you try the same approach*. Your list of trusted reviewers will not match mine, because you will have a different style of photography, needs and wants, and a different way of taking in information.

 

* I'm comfortable with understanding technical details, but I find analysing them to death is tedious when I'd rather take photos. 

Edited April 15, 2023 by LocalHero1953

[hansvons]

55 minutes ago, LocalHero1953 said:

* I'm comfortable with understanding technical details, but I find analysing them to death is tedious when I'd rather take photos. 

Exactly my sentiment.