Is DR higher, the higher the ISO at a given aperture/shutter?

[_GUI_]

I would like to ask all, specially the op since a forum member in other forum is claiming "captured DR is higher, the higher the ISO is" as a general rule with reference to one of the Cambridge in Colour tutorials (http://www.cambridgeincolour.com/tut...amic-range.htm).

It is well know that we can capture more DR at low ISO values when there is no aperture/shutter limitation.

It is also known that when in low light conditions we cannot increase aperture or reduce shutter speed beyond a certain limit, and we get underexposure, it is best to increase ISO in order to reduce noise.

Do you agree then with the claim: "at a given aperture/shutter more DR can be captured the higher the ISO is set?"

Shots at the same aperture/shutter in both cases:

BR

[McQ]

I fully agree that the captured dynamic range is higher for lower ISO speeds (excluding extended modes, such as ISO 50). On the other hand, I think the question posed here is phrased quite differently than it is usually asked, since the shutter speed is not being adjusted to maintain the same exposure. This is of course not how different ISO speeds would be used in practice, but the dynamic range difference between low and high ISO speeds remains unchanged.

Overall: while it is true that the lower end of the dynamic range is more effectively captured at higher ISO, this does not fully compensate for how much the highlights are compromised as a result. In other words, a shot at 1 second, f/8 and ISO1600 will capture far far more of the scene's shadow detail than an identically lit/composed shot at 1 second, f/8 and ISO100. However, the shot at ISO1600 will clip much sooner than the ISO100 shot...resulting in an overall dynamic range which is still less than the higher ISO shot.

[_GUI_]

Originally Posted by McQ

Overall: while it is true that the lower end of the dynamic range is more effectively captured at higher ISO, this does not fully compensate for how much the highlights are compromised as a result. In other words, a shot at 1 second, f/8 and ISO1600 will capture far far more of the scene's shadow detail than an identically lit/composed shot at 1 second, f/8 and ISO100. However, the shot at ISO1600 will clip much sooner than the ISO100 shot...resulting in an overall dynamic range which is still less than the higher ISO shot.

assuming no highlight clipping, which ISO captures more DR if aperture/shutter remain constant?

[McQ]

Originally Posted by _GUI_

assuming no highlight clipping, which ISO captures more DR if aperture/shutter remain constant?

The lower ISO is capable of capturing more DR, regardless of whether the aperture or shutter remain constant. Whether it actually does in reality capture more DR depends on how much DR the scene contains, and if it was properly exposed.

[_GUI_]

Originally Posted by McQ

The lower ISO is capable of capturing more DR, regardless of whether the aperture or shutter remain constant. Whether it actually does in reality capture more DR depends on how much DR the scene contains, and if it was properly exposed.

Please look at the samples on top: the low light conditions provoqued that even at the slowest affordable shutter speed, and the highest affordable aperture, none of the images (not even the ISO1600) got the highlights clipped.
On the other side, the ISO1600 shot managed to capture far more detail in the shadows as can be seen in the crops.

If captured DR is defined as the range of luminosities in the scene that were captured both unclipped and fairly free of noise, don't you agree that we managed to capture more DR with the ISO1600 shot?

The texture of the wireless phone was totally lost because of noise in the ISO100 shot, while is fairly well recognisable in the ISO1600. That's more captured DR in the shadows end and that means in this case more end to end captured DR.

[McQ]

Originally Posted by _GUI_

Please look at the samples on top: the low light conditions provoqued that even at the slowest affordable shutter speed, and the highest affordable aperture, none of the images (not even the ISO1600) got the highlights clipped.
On the other side, the ISO1600 shot managed to capture far more detail in the shadows as can be seen in the crops.

What were the exposure settings other than ISO? I must be misunderstanding something with this post, because using the same aperture/shutter speed at ISO100 and ISO1600 ought to greatly overexposure or underexpose one of the two shots. ie, if the ISO100 shot was properly exposed, the ISO1600 shot should be 4 f/stops or 16X overexposed.

Originally Posted by _GUI_

If captured DR is defined as the range of luminosities in the scene that were captured both unclipped and fairly free of noise, don't you agree that we managed to capture more DR with the ISO1600 shot? the texture of the wireless phone was totally lost in the ISO100 shot, while was still quite recognisable in the ISO1600. That's more captured DR on the shadows end.

Sure, then in this case the captured DR was greater at ISO1600 than ISO100, if I am understanding the test case correctly. However, this was not because ISO1600 necessarily is capable of capturing more DR than ISO100-- it is just because ISO1600 was used by the camera operator more effectively. If using the same exposure for both ISO speeds, the ISO100 shot will capture more dynamic range (if it is there to be captured).

[_GUI_]

Originally Posted by McQ

What were the exposure settings other than ISO? I must be misunderstanding something with this post

All time I am referring to shots taken at the same aperture/shutter; in the experiment: T=1/4s, f/13, 50mm, and obviously exposure-corrected in postprocessing to find out the degree of detail captured in the shadows.

I think the key to understand what I was talking about was in my first post:

"It is also known that when in low light conditions we cannot increase aperture or reduce shutter speed beyond a certain limit, and we get underexposure, it is best to increase ISO in order to reduce noise".

I just simulated a situation with restricted aperture/shutter to get a proper exposure in the RAW file, like could be shooting dark indoor scenes, concerts, weddings without flash,...

[McQ]

Originally Posted by _GUI_

All time I am referring to shots taken at the same aperture/shutter; in the experiment: T=1/4s, f/13, 50mm, and obviously exposure-corrected in postprocessing to find out the degree of detail captured in the shadows.

Ok, good, I was understanding you correctly. All above statements I made hold true.

Originally Posted by _GUI_

I think the key to understand what I was talking about was in my first post:

"It is also known that when in low light conditions we cannot increase aperture or reduce shutter speed beyond a certain limit, and we get underexposure, it is best to increase ISO in order to reduce noise".

I just simulated a situation with restricted aperture/shutter to get a proper exposure in the RAW file, like could be shooting dark indoor scenes, concerts, weddings without flash,...

The key here is "capturable" dynamic range. This is always greater for lower ISO speeds, excluding extended modes, regardless of what aperture or shutter speed is used. In your case we are talking about *captured* dynamic range (ie, in the RAW file itself), which was intentionally crippled for your ISO100 example. So yes, the ISO100 photo did capture less dynamic range in its RAW file. In situations where one is limited to using only a very narrowly defined aperture and shutter speed, then yes, the ISO1600 shot *may* give more dynamic range if the conditions are right...which I understand is what you were trying to simulate in the ISO100/ISO1600 example images you gave.

Overall though, when someone says that "ISO100 has more dynamic range than ISO1600", this statement is true because it commonly refers to the capturable dynamic range. Whether this DR advantage holds true in practice depends on whether the camera's capturable dynamic range has been maximally utilized by the chosen exposure setting. For example, I could easily record a greater dynamic range in a compact camera than a high-end SLR (even though the latter is capable of much more) if I blew the exposure setting with the digital SLR.

[McQ]

To add to my earlier replies: on second look it is actually quite remarkable that the ISO100 shot fared so well at the same shutter/aperture setting as the ISO1600 shot. Really amazing that there's that much shadow detail hidden away.

I'd also like to add to the following statement:

"at a given aperture/shutter more DR can be captured the higher the ISO is set?"

At the same aperture/shutter, if the lower ISO image is properly exposed (and as a result the high ISO image is waay overexposed) then the lower ISO image will still have more dynamic range. The DR advantage can go either way.

[_GUI_]

Originally Posted by McQ

I'd also like to add to the following statement:

At the same aperture/shutter, if the lower ISO image is properly exposed (and as a result the high ISO image is waay overexposed) then the lower ISO image will still have more dynamic range. The DR advantage can go either way.

I agree, but the whole thread didn't refer to the case in which the lower ISO image was properly exposed, but clearly underexposed:

It is also known that when in low light conditions we cannot increase aperture or reduce shutter speed beyond a certain limit, and we get underexposure, it is best to increase ISO in order to reduce noise.

Do you agree then with the claim: "at a given aperture/shutter more DR can be captured the higher the ISO is set?"

BR

[McQ]

Sure, in the situation that you have no choice but to increase ISO to increase exposure, more ISO yields improved dynamic range. This way the tonal mapping is done in analog using the camera's internal electronics as opposed to being done digitally in software. The only caveat is that for many photographers I would contend that this scenario represents a corner case.

[wnelson]

Originally Posted by McQ

Overall though, when someone says that "ISO100 has more dynamic range than ISO1600", this statement is true because it commonly refers to the capturable dynamic range......

Except the tests that I have done with my 30D. In this case there appears to be 1/3rd stop *less* DR with ISO100 than with ISO200 and above. This is with properly exposed images which I realise is different to what is being discussed here.

Wayne

[McQ]

Originally Posted by wnelson

Except the tests that I have done with my 30D. In this case there appears to be 1/3rd stop *less* DR with ISO100 than with ISO200 and above.

Yeah, I think that the "native" ISO on Canon models has shifted from the original ISO100 more towards ISO200-- similar to what the Nikon cameras were from early on. Just a qualitative observation though, with no knowledge of if/how the inner workings of the sensor/read-out have changed...

[xeliex]

Originally Posted by McQ

To add to my earlier replies: on second look it is actually quite remarkable that the ISO100 shot fared so well at the same shutter/aperture setting as the ISO1600 shot. Really amazing that there's that much shadow detail hidden away.

I'd also like to add to the following statement:

At the same aperture/shutter, if the lower ISO image is properly exposed (and as a result the high ISO image is waay overexposed) then the lower ISO image will still have more dynamic range. The DR advantage can go either way.

This makes sense.
Thanks GUI for posting this topic.

This reminds me of the technique "exposing to the right" where we try to maximize DR by maximizing the noise to signal ratio. You can read more at Luminous Landscape

What I got from this thread is;
Practically speaking, if you are shooting a birthday party with your f/4 max 70-200mm. It is better (less noise / more detail / more DR) to crank up the iso to 1600 and shoot your 70mm at f/4 1/60 than it is to shoot at ISO 400 at the same exposure and recover the lost shadows later post processing.

McQ, what do you exactly mean when answering wnelson?
Will my say Rebel XTi capture a higher absolute dynamic range at ISO 200 than ISO 100?
Also, why do some photographers advise you to shoot ISO 200 in harsh sun?

[McQ]

Originally Posted by xeliex

McQ, what do you exactly mean when answering wnelson?
Will my say Rebel XTi capture a higher absolute dynamic range at ISO 200 than ISO 100?
Also, why do some photographers advise you to shoot ISO 200 in harsh sun?

In general, the lowest ISO will always provide the maximum "capturable" dynamic range. Only the extra low ISO settings that are not enable by default are the ones which generally have lower dynamic range (such as ISO 50 with the Canon 5D and 1-series). Whether you actually do capture my dynamic range all depends on whether you've properly exposed the image though...

Not sure about the advice for using ISO 200 in harsh sun though... Perhaps this is to minimize the amount of sun your sensor gets (and therefore reduce heat) during the exposure. This is really not something one should ordinarily be considering though if that is indeed the reason for the advice.

[jamey15]

Originally Posted by McQ

In general, the lowest ISO will always provide the maximum "capturable" dynamic range. Only the extra low ISO settings that are not enable by default are the ones which generally have lower dynamic range (such as ISO 50 with the Canon 5D and 1-series). Whether you actually do capture my dynamic range all depends on whether you've properly exposed the image though...

Not sure about the advice for using ISO 200 in harsh sun though... Perhaps this is to minimize the amount of sun your sensor gets (and therefore reduce heat) during the exposure. This is really not something one should ordinarily be considering though if that is indeed the reason for the advice.

Of course!!!, I am confused!!!

I have a Canon Pro 1. I am looking for the type of photo acheived by the Sigma DP1..(?? maybe no way??) I read high iso is better ///low iso is better...Help.. I have seen photos from the Pentax optio A 30 a tiny point and shoot..That appear to have better dynamic range(luminosity) ?? any thoughts. Thank you....Jim
jamey15@socal.rr.com

[sebasj]

Originally Posted by jamey15

Of course!!!, I am confused!!!

I have a Canon Pro 1. I am looking for the type of photo acheived by the Sigma DP1..(?? maybe no way??) I read high iso is better ///low iso is better...Help.. I have seen photos from the Pentax optio A 30 a tiny point and shoot..That appear to have better dynamic range(luminosity) ?? any thoughts. Thank you....Jim
jamey15@socal.rr.com

Don't confuse... McQ and GUI are referring to a very specific kind of situation. As a generalization, it is better to shoot at the slowest ISO speed you can to achieve a proper exposure.

The case that GUI compares is simple indeed: I have a scene where the appropiate exposure is -to say a number- 1/60 f/8 ISO 1600.

As we know, ISO 1600 is noisy. So, to reduce noise, what would happen if we shoot at ISO100 (maintaining f/8 and 1/60) and then recover shadows -because it will be highly underexposed- in a photo editing software? Well, as GUI and McQ said, the dynamic range is affected. Thus, in that kind of scene, it is better to expose at ISO 1600 and reduce noise later on some noise reduction software perhaps. Properly exposed -for example 1/15 f/4 for ISO 100- the ISO 100 image is better of course.

The rule of thumb still applies: it is best to shoot at the slowest ISO sensitivity possible -while maintaining a correct exposure- to reduce noise issues and good dynamic range.

Best,
Sebas.

[TommyO]

GUI,
I tend to think of it this way....

In normal low light environments, while the high ISO is more "sensitive", there is less available light. Hence, it is now sensing random noise just as much as it would the low light. Thus, noise shows up in the image.

In controlled light environments (like your example), there is ample available light. Hence the chip is sensing more real light than noise. Thus, noise is reduced although still present. In the low ISO shot, you simply prevented enough photons from interacting with the chip by limiting the shutter speed. The internal algorithm for the camera processed the available light longer (not knowing you would do that) and thus amplified the noise more.

ISO no longer is a fixed element, like film was. It is an algorithm determined by the camera manufacturer, and we have no idea what they are doing. Once you override the exposure, you may get different results for different cameras.

[Graystar]

I would submit that there is something wrong with the OP’s images. They are not representative of a 4-stop increase in exposure compensation.

There was a discussion on exposure compensation over at DPR back in December. Here is a link to the images that were used in the final experiments.

http://gregorywilson.smugmug.com/gal...38516662_rzo7p

What we found was that applying 4 stops of EC to an underexposed ISO 200 image was nearly as good as the ISO 3200 image, taken at proper exposure. The high ISO image was better (we supposed) because the gain is applied to the analog signals from the sensor. So EC applied during RAW processing suffers from quantization error. However, it wasn’t anywhere near as bad as the OP’s images would suggest.

Here’s a sample I just snapped. The image is of Canon’s 4x6 Photo Paper Glossy box. My camera only goes from ISO 80 to 800, so the test is limited to 3 stops, between 100 and 800. These are screen captures from the RAW processing software.



There's not much difference between the two. As for dynamic range, it is as McQ says. The OP’s images were properly exposed at ISO 1600, so you wouldn’t expect the highlights to be overexposed. That doesn’t mean that high ISO has more dynamic range. Also, just because computer software can’t do as good a job applying gain as a camera can, that doesn’t mean that the original underexposed image doesn’t contain a greater dynamic range than the high ISO image. It does...it’s just too dark to see. The bottom line is that due to the results of many tests, performed by many people, it is well accepted that high ISO images have less dynamic range.