Jump to content

Understanding SL2 and SL2-S ISO settings


kobra

Recommended Posts

Advertisement (gone after registration)

2 hours ago, kobra said:

Hmm, I'm not sure on that... I don't think ISO on digital sensors is changing how much light hits the sensor; rather it is changing how sensitive the pixels are to the light that hits them - unlike shutter or aperture that actually will increase or decrease the light hitting the sensor pixels. 

So, in your example above, if using ISO100 and -1 EC, wouldn't that force ISO50? Or maybe I'm not understanding what you meant. 

Thanks,

Brad

ISO does not change sensor’s sensitivity.

If you use automatic metering with fixed ISO, and you apply -1 EC, ISO will stay fixed but either aperture will close one stop or shutter speed will go up one stop. You will have the same exposure (shutter speed and aperture) as if you had selected ISO 200 with 0EC in automatic metering. The advantage of ISO 100 with -1 EC vs. ISO200 with 0EC is that former (ISO100) preserves highlights better.

Link to post
Share on other sites

8 minutes ago, SrMi said:

ISO does not change sensor’s sensitivity.

If you use automatic metering with fixed ISO, and you apply -1 EC, ISO will stay fixed but either aperture will close one stop or shutter speed will go up one stop. You will have the same exposure (shutter speed and aperture) as if you had selected ISO 200 with 0EC in automatic metering. The advantage of ISO 100 with -1 EC vs. ISO200 with 0EC is that former (ISO100) preserves highlights better.

Ok, now I'm confused - https://digital-photography-school.com/understand-iso-digital-camera/ 

If ISO is not the sensor's sensitivity to light, what is it?

Thanks!

Brad

Link to post
Share on other sites

33 minutes ago, dem331 said:

This is the typical thread in which you start by thinking you are acquiring new knowledge and you end up understanding neither what you have learned nor what you knew in the first place. 

Yes, I feel your pain :D

Brad

Link to post
Share on other sites

32 minutes ago, kobra said:

Ok, now I'm confused - https://digital-photography-school.com/understand-iso-digital-camera/ 

If ISO is not the sensor's sensitivity to light, what is it?

Thanks!

I've posted this image before. It's from Arri, but you'll find similar charts for other camera brands:

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!

This only applies for the main ISO range. Many cameras have "dual ISO," where the sensor has a second read path that kicks-in at higher ISOs.

What you need to know is that lower ISOs have less margin in the highlights, and more in the shadows. In this case, you'll get an even split around ISO 640. That's what you want for video and for events, because it leaves a lot of room for correction.

The reason to shoot at lower ISOs is because noise lurks in the shadows. You might want to expose your highlights close to the upper limit of the sensor, because that means your shadows have less noise. The problem is that you are at the limit of clipping, so you need to check your histograms to make sure you haven't gone too far. Gratuitous car analogy: you are more likely to crash when you are driving at the limit!

The other caveat is that noise is less objectionable in moving images. It adds texture, and you only see it for 1/24 of a second (or less) because it's distributed randomly. You notice noise more in a print because it doesn't move around, and it's easier to peer into one small section of the image. In the end, it's an aesthetic choice. For instance, shadow noise/grain is common in "street" images, but it stands-out more in Ansel Adams-style landscapes.

Link to post
Share on other sites

2 minutes ago, BernardC said:

I've posted this image before. It's from Arri, but you'll find similar charts for other camera brands:

This only applies for the main ISO range. Many cameras have "dual ISO," where the sensor has a second read path that kicks-in at higher ISOs.

What you need to know is that lower ISOs have less margin in the highlights, and more in the shadows. In this case, you'll get an even split around ISO 640. That's what you want for video and for events, because it leaves a lot of room for correction.

The reason to shoot at lower ISOs is because noise lurks in the shadows. You might want to expose your highlights close to the upper limit of the sensor, because that means your shadows have less noise. The problem is that you are at the limit of clipping, so you need to check your histograms to make sure you haven't gone too far. Gratuitous car analogy: you are more likely to crash when you are driving at the limit!

The other caveat is that noise is less objectionable in moving images. It adds texture, and you only see it for 1/24 of a second (or less) because it's distributed randomly. You notice noise more in a print because it doesn't move around, and it's easier to peer into one small section of the image. In the end, it's an aesthetic choice. For instance, shadow noise/grain is common in "street" images, but it stands-out more in Ansel Adams-style landscapes.

That is the part that I meant is not relevant for still photography :).

Link to post
Share on other sites

Interesting topic. Have tried to reproduce this in real life. There is no difference in highlight clipping between ISO 100 and ISO 800. Have exposed for skin like described you would as well in the video world, and adjusted exposure accordingly. 

Doing the same for ISO 3200 and above that seems to work however.

Could it be that Leica changed this with the later firmwares? Running the newest one.

 

Link to post
Share on other sites

On 1/24/2023 at 8:12 PM, SrMi said:

ISO does not change sensor’s sensitivity.

This.

At ISO 3200 (SL2-S), when amplification kicks in and the sensor stops working invariantly, the sensor's sensitivity still doesn't change. However, the image will be brighter. That's similar to a pushed development process with B&W film. The film's sensitivity remains (it's inherent with the emulsion), but the brightness will be pushed with all the downsides like exacerbated grain/noise, especially in the shadows.

38 minutes ago, OleBe said:

Doing the same for ISO 3200 and above that seems to work however.

That's because, at ISO 3200, the sensor isn't invariant anymore.

Link to post
Share on other sites

vor 17 Minuten schrieb hansvons:

This.

At ISO 3200 (SL2-S), when amplification kicks in and the sensor stops working invariantly, the sensor's sensitivity still doesn't change. However, the image will be brighter. That's similar to a pushed development process with B&W film. The film's sensitivity remains (it's inherent with the emulsion), but the brightness will be pushed with all the downsides like exacerbated grain/noise, especially in the shadows.

That's because, at ISO 3200, the sensor isn't invariant anymore.

Think I understand this, but I could not reproduce the mid point shift from ISO 100 to 800. The highlights seem to clip exactly at the same stops over middle grey, no matter the ISO setting. This should be different right? 

 

Link to post
Share on other sites

17 minutes ago, OleBe said:

Think I understand this, but I could not reproduce the mid point shift from ISO 100 to 800.

I guess you refer to these infamous DXO graphs? I can’t see a meaningful relation of that graphs to my real-world experiences. Grey is all theory.

Link to post
Share on other sites

vor 34 Minuten schrieb hansvons:

I guess you refer to these infamous DXO graphs? I can’t see a meaningful relation of that graphs to my real-world experiences. Grey is all theory.

No, I do not care for that graph. If I expose for a middle grey subject, at ISO 100 the clipping to white occurs approx 3 stops over the subject. If you set ISO 800 and adjust exposure so that the subject is still middle grey, the clipping should appear 6 stops over. But it is clipping at 3 stops over still, so no matter the ISO setting, that means the camera is not behaving like for instance an Alexa. At least not in still mode.

Link to post
Share on other sites

4 hours ago, hansvons said:

At ISO 3200 (SL2-S), when amplification kicks in and the sensor stops working invariantly, the sensor's sensitivity still doesn't change

Where did you get that information from? P2P measurement, the standard for evaluating invariance, disagrees with what you wrote.

Link to post
Share on other sites

7 hours ago, OleBe said:

No, I do not care for that graph. If I expose for a middle grey subject, at ISO 100 the clipping to white occurs approx 3 stops over the subject. If you set ISO 800 and adjust exposure so that the subject is still middle grey, the clipping should appear 6 stops over. But it is clipping at 3 stops over still, so no matter the ISO setting, that means the camera is not behaving like for instance an Alexa. At least not in still mode.

isn't the Alexa  chart in video and LOG capture?

I would imagine that in photo mode the Dynamic range changes from ISO100 to 800 and above.
I am using the SL2, so it is very different then SL2-s.

Link to post
Share on other sites

vor einer Stunde schrieb Photoworks:

isn't the Alexa  chart in video and LOG capture?

I would imagine that in photo mode the Dynamic range changes from ISO100 to 800 and above.
I am using the SL2, so it is very different then SL2-s.

Yes, the chart of the Alexa is for video mode. I thought after reading this thread that the SL2-S should behave similar in stills mode, but at least for my testings with the latest firmware, that does not seem to be the case. 

Would be really great if that would be possible though. :) 

 

Link to post
Share on other sites

23 hours ago, SrMi said:

P2P measurement, the standard for evaluating invariance, disagrees with what you wrote.

They are hardly a standard, except by their own estimation. Their methodology is inherently flawed, and the actual measurements are farmed-out to punters who have no hope of achieving repeatable results. The intent is to crowd-source enough results so that errors cancel-out, but that won't work with less popular models.

There's been a lot of talk over the years about how to measure dynamic range in a repeatable manner. First-off, you need a dedicated test setup, with consistent temperature (ideally you want to measure temperature variance). You also need to measure DR within a single exposure (which is very hard to do), using a standard lens.

You also need to include subjective analysis of the results. Are the shadows usable? How is the highlight roll-off? How close can you get to each extreme safely? What's the quality of the exposure range in-between the two extremes?

The only people who do this kind of thing are camera manufacturers, and motion picture rental houses. Rental houses test to see if a camera is "out of spec," so they don't care much about having an accurate number, just a consistent one.

Link to post
Share on other sites

1 hour ago, BernardC said:

They are hardly a standard, except by their own estimation. Their methodology is inherently flawed, and the actual measurements are farmed-out to punters who have no hope of achieving repeatable results. The intent is to crowd-source enough results so that errors cancel-out, but that won't work with less popular models.

There's been a lot of talk over the years about how to measure dynamic range in a repeatable manner. First-off, you need a dedicated test setup, with consistent temperature (ideally you want to measure temperature variance). You also need to measure DR within a single exposure (which is very hard to do), using a standard lens.

You also need to include subjective analysis of the results. Are the shadows usable? How is the highlight roll-off? How close can you get to each extreme safely? What's the quality of the exposure range in-between the two extremes?

The only people who do this kind of thing are camera manufacturers, and motion picture rental houses. Rental houses test to see if a camera is "out of spec," so they don't care much about having an accurate number, just a consistent one.

Bill has occasionally collected measurements for the same camera from multiples sources and has not found any variations to speak of.

Bill has occasionally collected measurements for the same camera from multiple sources and has yet to find any significant variations. Yes, there are justified discussions about whether the way he measures the DR makes sense (PDR).

However, that is off-topic. Instead, I commented on  @hansvonsinvariance claim (my claim was: "P2P measurement, the standard for evaluating invariance") :

On 1/28/2023 at 2:45 AM, hansvons said:

At ISO 3200 (SL2-S), when amplification kicks in and the sensor stops working invariantly, the sensor's sensitivity still doesn't change.

While I do not know what kind of amplification hansvons refers to, we see that SL2 starts scaling at 3200, while SL2-S has no scaling at all (link). We also see here that SL2-S is practically invariant from ISO1600 and not from ISO 100. I would like to learn how hansvons came to his claims of invariance and amplification.

The scaling, noise reduction, and shadow improvement measurements are not controversial, AFAIK.

Link to post
Share on other sites

20 hours ago, SrMi said:
  21 hours ago, BernardC said:

They are hardly a standard, except by their own estimation.

 

20 hours ago, SrMi said:

Bill has occasionally collected measurements for the same camera from multiple sources and has yet to find any significant variations.

That's funny, and it confirms what I wrote.

Link to post
Share on other sites

On 1/29/2023 at 5:03 PM, SrMi said:

I would like to learn how hansvons came to his claims of invariance and amplification.

That's how I learn how a camera's sensor works in terms of ISO, noise and the rest of it.

1. Select ISO 100. Underexpose the image by, say, four stops. Open up the image in your RAW editor of choice. Pull the exposure slider to the right until the picture looks right. 

2. Select ISO 1600 and shoot the same motive. Import the image in your editor. It should look roughly identical to the image above. Do some pixel peeping and compare the noise of the two images.

If the noise looks the same, the sensor is invariant. If not, some amplification occurs, and the sensor is not invariant at the chosen ISO settings. If I recall correctly, at ISO 3200 (maybe it's ISO 6400), the SL2-S noise started to look different.

 

 

 

 

Edited by hansvons
Link to post
Share on other sites

3 hours ago, hansvons said:

That's how I learn how a camera's sensor works in terms of ISO, noise and the rest of it.

1. Select ISO 100. Underexpose the image by, say, four stops. Open up the image in your RAW editor of choice. Pull the exposure slider to the right until the picture looks right. 

2. Select ISO 1600 and shoot the same motive. Import the image in your editor. It should look roughly identical to the image above. Do some pixel peeping and compare the noise of the two images.

If the noise looks the same, the sensor is invariant. If not, some amplification occurs, and the sensor is not invariant at the chosen ISO settings. If I recall correctly, at ISO 3200 (maybe it's ISO 6400), the SL2-S noise started to look different.

I applied your approach, and I see a considerable difference in noise between ISO 1600 (upper) and ISO 100 (lower) (100% view in LrC, screen grab):

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!

 

SL2-S handles noise really good, so that I need to push both images quite a bit to make the difference obvious.

On the other hand, applying the same acrobatics to ISO 800 and 12500 shows similar noise (ISO invariance).

My test confirms what P2P measurements show: SL2-S has an invariant range from 800 to at least 12500, but is not invariant from 100 to 1600.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...