Real iso's?

[bdunbar79]

Can I get some credit here?? In the post that the OP is speaking of, I DID tell dilbert he was wrong!

We give you credit

Now go and fix that 400 mount

:'(

 

[dppaskewitz]

Has anyone done any testing of the 6D? According to Claff's charts, the 5DII, 5DIII (and 30D, for that matter) seem to follow one pattern (less noise at 160, 320, etc.) while the 5D classic (my other body) follows a 100, 200, 400 pattern).

 

[Click]

Very interesting thread. I'm following this closely. Thanks for the great info. 8)

 

[RGF]

IThe golden rule is that the limit is MP/MB+1(file size) so 19MB is in the safe/clean zone and 20MB is pushing your camera. It's no math, just hard numbers.

MxM

Can you explain the golden rule? Why is it true? What does it measure?

 

[RGF]

Very interesting thread. I'm following this closely. Thanks for the great info. 8)

+100

 

[Pi]

As you can see is ISO 160, 320, 640 the camera's sweet-spot...

When you shoot with the cap on...

 

[Sporgon]

I did an ISO test with my 7D.

Put the camera in a dark room (basement) leave the lenscap on, close/block the viewfinder, use RAW or disable ISO long exposure/noise reduction. Choose the settings that you want (Manual mode). Take every picture for at least 10 seconds and you will come up with al list like me.

The 7D is an 18MP camera which is equal to approx. 18MB file size. (the smaller the file size, the cleaner the image is)

As you can see is ISO 160, 320, 640 the camera's sweet-spot... The golden rule is that the limit is MP/MB+1(file size) so 19MB is in the safe/clean zone and 20MB is pushing your camera. It's no math, just hard numbers.

MxM

But what practical data is established by exposing the sensor in the dark for ten seconds ? Surely you are chronically under exposing never mind how long you leave the shutter open. The sensor will start to generate heat.

I see your data matches the Bill Claff charts for the 7D. Looking at the 5D mk2 data it stated that the read noise is the same at ISO 640 as it is at 100. This will give people the impression that your data will show the same amount of noise at these exposures, but this just isn't the case. Shoot a sky at ISO 640 and it is very noisy. Shoot at ISO 100 and it's pretty clean: there is no comparison in data quality - so what's all this read noise meant to mean in practice ?

I think it is very misleading - unless of course you like to shoot inside of lens caps in the dark.

 

[privatebydesign]

Pi and Sporgon, you are missing the point.

The point is that the read noise displayed is added to a regular shot in lesser amounts at shorter exposures. You can argue the relevance, but when you are trying to pull that last bit of detail from a dark shadow that tiny extra amount of noise might make the difference.

It is all about setting yourself up to get the last 1 or 2 % of your sensors capabilities, this small difference is best illustrated with long dark exposures, but is just as valid for shorter brighter ones.

 

[Pi]

Pi and Sporgon, you are missing the point.

No, you are just misreading my posts.

 

[Sporgon]

Pi and Sporgon, you are missing the point.

The point is that the read noise displayed is added to a regular shot in lesser amounts at shorter exposures. You can argue the relevance, but when you are trying to pull that last bit of detail from a dark shadow that tiny extra amount of noise might make the difference.

It is all about setting yourself up to get the last 1 or 2 % of your sensors capabilities, this small difference is best illustrated with long dark exposures, but is just as valid for shorter brighter ones.

Well I'm going to have a look at this tomorrow. By the nature of what I shoot I'm nearly always on ISO 100, so don't have many files I can use as a comparison, but at the moment I believe that ISO 100 is much better than 640 in every way, quality and post process -wise.

 

[Marsu42]

So your recommendation is full iso stops and expose to the right, isn't that basic knowledge and in agreement with the article?

This isn't so, the Magic Lantern devs have figured it out, see this good read about what iso is best (see q&a section): http://magiclantern.wikia.com/wiki/ISO

Then, what is the best ISO?

To the best of my knowledge, the best ISOs are the ones available in recent Magic Lantern versions (April 2012 or later), obtained from ISO 100 multiples adjusted with a small amount of negative digital gain:

* ISO 85, 175, 350, 700, 1400, 2800 - best for Neutral -4 and other low-contrast styles.
* ISO 80, 160, 320, 640, 1250, 2500 - good for a wide range of situations.
* ISO 70, 140, 280, 560, 1100, 2200 - best for high-contrast styles.

 

[Zen]

There are basically three 'types' of ISOs:

• Base ISO - this is the 'real' ISO for the sensor before any amplificaition. For most sensors, this is actually in the ISO 60 to ISO 80 range, not ISO 100 as many people assume.
• Native ISOs - analog amplification applied to the base signal, prior to analog to digital conversion (ADC). These values have numbers for selection (e.g. ISO 3200)
• Expanded ISOs - digital amplification applied to the signal after the maximum analog amplification, occurring after the ADC. These valuse have letters for selection (L, H1, etc.).

You're talking about 'tweener' ISOs - those are seen in some cameras but not others. Some analog amplifiers (most of them, for Canon sensors, actually) are only capable of full stop incremental amplification, so digital amplification (pushing/pulling occurring after the ADC) is applied to the signal after the 'native' full-stop analog amplification. If you look at Bill Claff's data for Canon sensors, you can see the zig-zag curves that result from these 'tweener' ISO settings, with all their cameras except the 1D X.

So, dilbert is wrong about the 5DII's ISO settings - ISO 6400 is the highest 'real' (aka native) ISO. He may be confused because ISO 3200 is the highest available ISO when ISO is set to Auto. I have no idea what he means by stating, " Once the ISO is past the point where IQ drops more than 1 stop per ISO stop" - what the heck is '1 stop of IQ'?!? Last time I checked, IQ was not measured in stops anywhere outside of dilbert-land (a fanciful place where lenses are sometimes cameras). But the reality is, noise and DR scale linearly in the 5DII from ISO 1600 all the way through H2, so the 'IQ difference' between ISO 1600 and ISO 3200 is the same as the difference between ISO 3200 and ISO 6400.

Having said that, the highest ISO one will use is a personal judgement call. With the 5DII, ISO 3200 was my highest setting for non-emergency use (but that doesn't make ISO 6400 'not real' - it's still a native ISO setting).

How in the world do you guys remember all this stuff, or even better, how do you take advantage of it when shooting your images? I'm completely lost when it comes to this highly technical data. Does it actually make a difference in your images?

Guess I'm a real boob! Just can't seem to get my arms around it all . . .

Thanks for trying to pound it into my thick skull just the same.

Zen ???

 

[Valvebounce]

WOW
Thanks guys for all the answers, I am going to confess that I am still unclear as to where my native (real to use my incorrect description) iso's run out and when they become expanded, is it in theory the one before H, or is it in multiples of the base iso, 60 to 80 which I also don't know for the 7D! Or is it still best to work in multiples of the lowest selectable iso of 100 (no L setting on 7D)?
Thanks to Neuro for his in depth answer, perhaps I should have asked for the simplified version, might make more sense when less tired and stressed!
Really enjoy reading this forum, not sure how much is going to stick.
So much info so little ram left I wonder what just got over written!

Cheers, Graham.

 

[privatebydesign]

WOW
Thanks guys for all the answers, I am going to confess that I am still unclear as to where my native (real to use my incorrect description) iso's run out and when they become expanded, is it in theory the one before H, or is it in multiples of the base iso, 60 to 80 which I also don't know for the 7D! Or is it still best to work in multiples of the lowest selectable iso of 100 (no L setting on 7D)?
Thanks to Neuro for his in depth answer, perhaps I should have asked for the simplified version, might make more sense when less tired and stressed!
Really enjoy reading this forum, not sure how much is going to stick.
So much info so little ram left I wonder what just got over written!

Cheers, Graham.

The 7D's best iso level, the one with least noise, is 160iso; 100 and 200 are a touch noisier, then 318 and 126, then 251 and then they all get progressively more noisy.

So for shooting photos the "best" iso would be 160, as you needed more go up 1/3rd to 200, skip 251 (2/3rds up), after that just keep it as low as is practical. Although these are odd numbers they are the actual iso values registered when the camera is adjusted in 1/3rd stops of iso.

All this will only make the absolute tiniest of differences to dark areas of shadow detail and even then only if you shoot RAW.

 

[LetTheRightLensIn]

I did an ISO test with my 7D.

Put the camera in a dark room (basement) leave the lenscap on, close/block the viewfinder, use RAW or disable ISO long exposure/noise reduction. Choose the settings that you want (Manual mode). Take every picture for at least 10 seconds and you will come up with al list like me.

The 7D is an 18MP camera which is equal to approx. 18MB file size. (the smaller the file size, the cleaner the image is)

As you can see is ISO 160, 320, 640 the camera's sweet-spot... The golden rule is that the limit is MP/MB+1(file size) so 19MB is in the safe/clean zone and 20MB is pushing your camera. It's no math, just hard numbers.

MxM

Don't forget that the clipping point also shifts though. ISO160 is nothing more than over-exposing ISO200 by 1/2rd stop (in RAW). So for RAW it really doesn't matter. For JPG and movies it can matter though because of how they apply the tone curves.

 

[Pi]

The 7D's best iso level, the one with least noise, is 160iso; 100 and 200 are a touch noisier, then 318 and 126, then 251 and then they all get progressively more noisy.

So for shooting photos the "best" iso would be 160, as you needed more go up 1/3rd to 200[...]

This is wrong. ISO 160 is pushed ISO 200 by 1/3 stop (ETTR). You can push ISO 100 to ISO 80, if you want. ISO 100 and ISO 200 have the same shadow noise but away from the shadows, ISO gets better and better, up to 2.6 dB better at the top. Nowhere in the range is ISO 200 better.

The illusion of the "better" ISO 160 comes from the hidden overexposure by 1/3 stop. That does not change the DR; in fact, the DR at ISO 200 (or 160) is slightly worse than at ISO 100.

 

[privatebydesign]

The 7D's best iso level, the one with least noise, is 160iso; 100 and 200 are a touch noisier, then 318 and 126, then 251 and then they all get progressively more noisy.

So for shooting photos the "best" iso would be 160, as you needed more go up 1/3rd to 200[...]

This is wrong. ISO 160 is pushed ISO 200 by 1/3 stop (ETTR). You can push ISO 100 to ISO 80, if you want. ISO 100 and ISO 200 have the same shadow noise but away from the shadows, ISO gets better and better, up to 2.6 dB better at the top. Nowhere in the range is ISO 200 better.

The illusion of the "better" ISO 160 comes from the hidden overexposure by 1/3 stop. That does not change the DR; in fact, the DR at ISO 200 (or 160) is slightly worse than at ISO 100.

Well if you take empirical measurements, or read the page that was linked to http://home.comcast.net/~NikonD70/Charts/RN_ADU.htm you would see that the 7D has the lowest read noise at iso 159 (160).

If you look further on the site you will find this page http://home.comcast.net/~NikonD70/Charts/PDR.htm this measures DR to be the same at 100 and 200 iso, well within 0.01 of a stop.

 

[Pi]

The 7D's best iso level, the one with least noise, is 160iso; 100 and 200 are a touch noisier, then 318 and 126, then 251 and then they all get progressively more noisy.

So for shooting photos the "best" iso would be 160, as you needed more go up 1/3rd to 200[...]

This is wrong. ISO 160 is pushed ISO 200 by 1/3 stop (ETTR). You can push ISO 100 to ISO 80, if you want. ISO 100 and ISO 200 have the same shadow noise but away from the shadows, ISO gets better and better, up to 2.6 dB better at the top. Nowhere in the range is ISO 200 better.

The illusion of the "better" ISO 160 comes from the hidden overexposure by 1/3 stop. That does not change the DR; in fact, the DR at ISO 200 (or 160) is slightly worse than at ISO 100.

Well if you take empirical measurements, or read the page that was linked to http://home.comcast.net/~NikonD70/Charts/RN_ADU.htm you would see that the 7D has the lowest read noise at iso 159 (160).

But higher shot noise above the shadows, per DXOmark. This makes the statement that ISO 160 is better incorrect. It has to be better everywhere, to be declared better.

And of course, you can just dial in EV=+1/3 at ISO 100 and beat ISO 160 everywhere (DXOmark). With FF bodies, you can do even better: ISO 50.

If you look further on the site you will find this page http://home.comcast.net/~NikonD70/Charts/PDR.htm this measures DR to be the same at 100 and 200 iso, well within 0.01 of a stop.

I said "slightly worse" (DXOmark). But getting the same DR does not make it better. The shot noise is still worse.

 

[wickidwombat]

Anyone got any idea how many shots you miss while agonising over this? ???

seriously I dont think it makes much of a difference

 

[privatebydesign]

This makes the statement that ISO 160 is better incorrect. It has to be better everywhere, to be declared better.

No it doesn't and, no it doesn't.

Look we are talking about tiny differences in the ability to process a RAW file, an actual picture. Specifically with a 7D shadow detail is the problem, it is so easy to adjust highlights and midtones 1/3rd stop either way with no loss of IQ, but where we struggle is getting that 1/3 stop extra in the shadows. If using 160 instead of 200 gets you less shot noise and no reduction in DR then surely that is a no brainer?

Better can mean many things, in my experience anything you can do with Canon RAW files to lessen problems in the shadows makes for better images.