Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Canon Forum | Join Upload & Sell

1      
2
       3       end
  

Archive 2012 · ISO sweet spots
  
 
lexvo
Offline
• • • •
Upload & Sell: Off
p.2 #1 · ISO sweet spots


This is a well known discussion. See these links and draw your conclusions
http://photocascadia.wordpress.com/2011/05/13/canon-eos-5d-mark-ii-iso-noise-test/
http://photography-on-the.net/forum/showthread.php?t=1229464



Dec 05, 2012 at 06:29 PM
cgarcia
Offline

Upload & Sell: Off
p.2 #2 · ISO sweet spots


Evangelos wrote:
It's funny, I just spent the last hour looking at different iso stops in pictures I've taken with my 5D3 (now sold) and now I find this thread. Its quite evident to me that 160, 320, and 640 are the best iso's through 800. Beyond that it seems to make no difference. ISO 100 has more evident banding than and blotchiness in the shadows than 160, while 125 is the biggest offender. The same holds true 1 stop up from each of these 1/3 stops. Is it a huge difference? No. But if you feel you may be manipulating the
...Show more

These levels (160/320...) may have less noise, but also less dynamic range. If I remember correctly (John Sheehy, from dpreview forums, was very expert about these topics) Canon implements these intermediate ISOs in most cameras by correcting the RAW values. That is, it not stores a mark in the RAW file meaning "multiply by 1.25" or "multiply by 0.8" during conversion, but instead, it physically performs the multiplication of the pixel RAW values before saving the RAW files, thus losing information (not only noise) either in the hightlights or in the shadows (stored numbers have a limit e.g. of 14 bits). In high end models (only 1D) it apparently uses analog gain instead of tweaking the RAW values, but results are not much better.

By the way, all ISOs beyong certain level used to be internally emulated by pushing the RAW value and were not different at all from a underexposed shot (in fact, the underexposed shot retains more headroom): cameras allowing up to ISO 3200 typically maxed out at 1600 and created 3200 by internally pushing the RAW values a full stop cropping the highlights (with current mega-ISO camera models I don't know).

Latest Sony sensors are so good in readout noise that a 100 ISO photo can be pushed to 1600 looking not very different from a true ISO 1600 shot. Canon still not, but trying to "improve" things by avoiding any push up (even as little as 1/3 f-stop) to 125/250 ... or systematically pusing down to hide noise (160/320...) maybe would not be a good idea in all cases.



Dec 05, 2012 at 06:44 PM
Evangelos
Offline

Upload & Sell: On
p.2 #3 · ISO sweet spots


While they may hold less DR, on not, the one thing I do know they have less of is banding. If I push the shadows of a 100 or 125 iso shot, I see a lot more banding than if I push the shadows of a iso 160 shot. The iso 160 shot will only have banding, if any , in the blackest part of the pushed shadow. I took some shots of my daughter with a 50L. I was shooting at f1.6 and she was sitting on a dark gray bed sheet. The sheets were blurred out and I could see faint traces of banding at iso 125 without even touching the shadows.

Banding, per se, may not affect many shots, and for me it really didn't. However, I've always disliked the blotchy blacks in the shadows when processing raw files with both the 5D2 and the 5D3. These blotchy blacks can easily be seen in large prints. They can be mitigated in post, but, it's nice not to have to. When shooting at iso 160 etc, the blotchiness is less severe. The real revelation for me was when I picked up a refurbished D800E for a low price. I always thought the blotchy blacks were a byproduct of the various raw converters, however, the blacks in the shadows have a very nice smooth transition to them...no blotchiness. This is how I see the iso issue...YMMV

Evan



Dec 05, 2012 at 08:28 PM
BrianO
Offline
• • • • •
Upload & Sell: On
p.2 #4 · ISO sweet spots


cgarcia wrote:
These levels (160/320...) may have less noise, but also less dynamic range.


According to these charts, on the 7D for example the DR drops noticably from ISO 100 to ISO 125, but then goes back up at ISO 160: 8.65 at 100, and 8.64 at 160...an insignificant difference.

The difference in read noise, though, is more significant: 6.449 ADU at 100, and 5.661 at 160.

So while the difference between ISO 100 and ISO 160 may, as Dan says, not be visible in the real world, if there's even a theoretical benefit to shooting at an "odd" ISO setting, why wouldn't I?



Dec 05, 2012 at 08:52 PM
Evangelos
Offline

Upload & Sell: On
p.2 #5 · ISO sweet spots


cgarcia wrote:
These levels (160/320...) may have less noise, but also less dynamic range.

BrianO wrote:
According to these charts, on the 7D for example the DR drops noticably from ISO 100 to ISO 125, but then goes back up at ISO 160: 8.65 at 100, and 8.64 at 160...an insignificant difference.

The difference in read noise, though, is more significant: 6.449 ADU at 100, and 5.661 at 160.

So while the difference between ISO 100 and ISO 160 may, as Dan says, not be visible in the real world, if there's even a theoretical benefit to shooting at an "odd" ISO setting, why wouldn't I?


+1



Dec 05, 2012 at 09:03 PM
rolette
Offline
• •
Upload & Sell: Off
p.2 #6 · ISO sweet spots


gdanmitchell wrote:
I've shot a 5D2 for a bit more than four years, making tens of thousands of exposures, and producing many prints in sizes up to 24" x 36". This whole "fractional ISO" business impresses me as measurebation of the most bizarre type. Here's the deal. I could show you two 24" x 36" prints of extremely high quality produced from images shot on the 5D2, with one exposed at ISO 100 and one at ISO 200, and you would be completely unable to tell which is which. I could likely include one shot at ISO 400 and absolutely no one
...Show more

On my old 40D, it made a significant difference in the quality of my indoor sports photos shooting at ISO 1250 instead of 1600. It didn't really make much sense until I saw these noise charts a couple of years later, but it was VERY obvious in the IQ of my photos.

If you are shooting at low ISO values, then sure. It really doesn't matter in practice. Just don't assume that translates to those of us that are shooting in caves/gyms or under stadium lights.



Dec 06, 2012 at 04:00 AM
taemo
Offline
• •
Upload & Sell: On
p.2 #7 · ISO sweet spots


^to be fair though I would never dare shooting higher than ISO 1250 with the 40D as ISO1600 is awful, funny though considering the 5Dc I could push 1600 and 3200 and will still look better than the 40D image


Dec 06, 2012 at 04:46 AM
gdanmitchell
Offline
• • • • •
Upload & Sell: On
p.2 #8 · ISO sweet spots


Eyvind Ness wrote:
OK, Dan. Let's have a real world example. You and I both have a perfect shot of a truly unique moment, but we both missed the exposure by quite a bit, somehow, in the heat of the moment, say by 1 full stop. However, I managed to use ISO 160, while you were at ISO 400. Now, which shot would you rather try to save during post processing?


ISO 400 works great for many situations. I often choose to shoot at 400 when photographing things like birds in flight. For the past few months I've been working on a project photographing orchestral musicians, using natural light on stage and back stage and often shooting at ISO 800 or 1600 and occasionally at 3200 with very good results.

So, "'saving" a 400 ISO shot seems like an unreal concept to me. I've even chosen to shoot landscape from the tripod at 400 on some occasions.

In any case, the real question in the context of claims being made in this thread would call for you to rephrase your question to ask about one of us shooting at 160 and the other at 200. No significant difference...

Take care,

Dan



Dec 06, 2012 at 08:12 AM
kaycephoto
Offline
• • •
Upload & Sell: Off
p.2 #9 · ISO sweet spots


on both the 1D4 & 5D2, i definitely found shooting at ISO320/640/1250 yielded much cleaner results than at 400/800/1600.. technical findings from other sources matched that conclusion

on the 1Ds3, i typically stayed at ISO100/200 when i could, going up to 400/800/1600 in full stops only.. data again backed this up.

the 1DX is basically the first camera I've ever shot where i feel free from being inclined to choose one set of ISO values over the other.. from the technical measurements I've seen, both the DR & read noise levels are relatively linear from ISO100 to ISO5000..

ymmv



Dec 06, 2012 at 11:08 AM
rolette
Offline
• •
Upload & Sell: Off
p.2 #10 · ISO sweet spots


taemo wrote:
^to be fair though I would never dare shooting higher than ISO 1250 with the 40D as ISO1600 is awful, funny though considering the 5Dc I could push 1600 and 3200 and will still look better than the 40D image


Tolerance for noise varies pretty dramatically. I've seen folks claim that they don't like shooting their 5D2 above ISO 800. On my 1D4, I live in the ISO 8000 world. When shooting sports, there are certain realities you have to deal with to get the shot. Noise can be dealt with. Subject motion, not so much...

No doubt the 5D had better IQ than the 40D, but for shooting sports, no way I would have made that trade. 40D has twice the frame rate and better AF.

Jay



Dec 06, 2012 at 12:46 PM
 

Search in Used Dept. 



Monito
Offline
• • • • • •
Upload & Sell: Off
p.2 #11 · ISO sweet spots


ratsnest74 wrote:
makes it look like the 5DIII is worse than the 5Dc, the 1DsII, the 1DIII, the 3Ds and the D4, going backward Canon!


Please read more carefully.

As stated at the site, for many technical reasons, you can't compare between cameras. One in particular, as the site page says, is that the measures are not normalized for area.



Dec 06, 2012 at 02:21 PM
Monito
Offline
• • • • • •
Upload & Sell: Off
p.2 #12 · ISO sweet spots


With the 5D Classic, I shot the ISO 100, 200, 400 sequence and occasionally 800. I rarely went above that, but I did if necessary.

With the 5D2, I shoot ISO 160, 320, 640, and sometimes 1250 (more often than I shot 800 before). I have gone above that too, more frequently than with the 5D. Above 1250, the step effect is not really apparent, so set it to the lowest that meets your needs (photography is the art of compromise).

The step effect at multiples of 160 is real (better dynamic range and less noise, except for the 5D c), but matters only marginally. However, I will take every little bit of quality I can eke out of whatever camera/film/sensor/lens combination I have in my hands at the time.

Tapeman wrote:
Small things add up. If I get a few of them right it may improve my images. [...] I don't see a downside to minor adjustments that may improve my pics.


Yup.

One never knows which little bit of quality of photograph will be the tipping point that flips a viewer's choice from one photograph to another.



Dec 06, 2012 at 02:26 PM
cgarcia
Offline

Upload & Sell: Off
p.2 #13 · ISO sweet spots


BrianO wrote:
According to these charts, on the 7D for example the DR drops noticably from ISO 100 to ISO 125, but then goes back up at ISO 160: 8.65 at 100, and 8.64 at 160...an insignificant difference.

The difference in read noise, though, is more significant: 6.449 ADU at 100, and 5.661 at 160.

So while the difference between ISO 100 and ISO 160 may, as Dan says, not be visible in the real world, if there's even a theoretical benefit to shooting at an "odd" ISO setting, why wouldn't I?


You are right. When ISO 160 is internally implemented by the camera (except in 1D series) as ISO 200 overexposed by 1/3 and correcting the exposure down 1/3 to emulate ISO 160, this is just a kind of ETTR (expose to the right) feature done by the camera, and ETTR is well known for dramatically improving the shadows. When there is enough DR or if we aren't able to notice any ill effect in DR, in cameras with not very good shadows, these intermediate ISOs (160/320...) may be a handy and effortless way to improve them. Unfortunately, the other intermediate ISOs (125/250...) are just the opposite.



Dec 06, 2012 at 03:18 PM
gdanmitchell
Offline
• • • • •
Upload & Sell: On
p.2 #14 · ISO sweet spots


Monito wrote:
The step effect at multiples of 160 is real (better dynamic range and less noise, except for the 5D c), but matters only marginally. However, I will take every little bit of quality I can eke out of whatever camera/film/sensor/lens combination I have in my hands at the time.


I'd actually go so far as to rephrase your point in a way that is more accurate in terms of real world photography, as differentiated from test-bench theoretical discussions: "The step effect at multiples of 160 can be measured but is invisible in photographs."

Added later:

A. It simply isn't worth worrying about.
B. Even the relatively small differences between whole ISO values (e.g. - 100 and 200) are barely visible in most cases. I challenge anyone to look at a 16" x 24" print and tell me whether it was shot at 100 or 200. In fact, it would be interesting to expand that to include a 400 ISO print.
C. The amounts of noise that folks are fretting over are not only entirely insignificant, but a bit of noise in an image often creates a more pleasing photograph than one in which noise has been overly reduced. In some cases we actually choose to add a bit of noise for this purpose.

Take care,

Dan


Edited on Dec 06, 2012 at 08:43 PM · View previous versions



Dec 06, 2012 at 05:36 PM
chez
Offline
• • • • •
Upload & Sell: On
p.2 #15 · ISO sweet spots


gdanmitchell wrote:
I'd actually go so far as to rephrase your point in a way that is more accurate in terms of real world photography, as differentiated from test-bench theoretical discussions: "The step effect at multiples of 160 can be measured but is invisible in photographs."

Take care,

Dan


So...why not use multiples of 160 if there is a measureable difference. If we look at all our equipment this way, not caring about the small added improvements, then in the end...these small improvements or the lack of them can result in an inferior print. We use tripods, try to not use apertures where diffraction starts to occur etc... Why not also use the ISO settings that produce the best measured result? This is just good practice IMHO.



Dec 06, 2012 at 07:32 PM
Rajan Parrikar
Online
• • •
Upload & Sell: On
p.2 #16 · ISO sweet spots


These are things I would like to know about regardless of their visibility in the output. Such knowledge never subtracts, it can only add (as a famous American physicist used to say). If this does not come with any penalty, I will implement it into my work mode.


Dec 06, 2012 at 09:24 PM
Access
Offline
• • •
Upload & Sell: Off
p.2 #17 · ISO sweet spots


chez wrote:
So...why not use multiples of 160 if there is a measureable difference. If we look at all our equipment this way, not caring about the small added improvements, then in the end...these small improvements or the lack of them can result in an inferior print. We use tripods, try to not use apertures where diffraction starts to occur etc... Why not also use the ISO settings that produce the best measured result? This is just good practice IMHO.

Many of these things are (in reality) tradeoffs, if you think this way, ie. "I am gaining something while losing nothing", then you are only seeing one side of it and not the whole story.




Dec 06, 2012 at 11:20 PM
Monito
Offline
• • • • • •
Upload & Sell: Off
p.2 #18 · ISO sweet spots


Ordinarily I agree with you, Dan, but here I must differ slightly.

gdanmitchell wrote:
B. Even the relatively small differences between whole ISO values (e.g. - 100 and 200) are barely visible in most cases. I challenge anyone to look at a 16" x 24" print and tell me whether it was shot at 100 or 200. In fact, it would be interesting to expand that to include a 400 ISO print.


The test you describe is bogus. It is obvious you do not know how to do psychophysiological experiments; or if you do, you are being purposefully bogus.

Your suggested "test" is as ridiculous as showing a print of a landscape and asking someone if it was made with a SoNikon or a Canon or a Pentax.

The proper way to do this kind of test is to show a number of people pairs of prints, one at 100 and one at 200 and ask them to decide which is which. Then you see if the performance of the group rises above chance. My experience comes from having been a research assistant, but the real scientists are even more sophisticated.

The visual difference between ISO 100 and ISO 200 will usually be slim to none. It is more about the 320 / 640 / 1250 set where it counts more, on modern Canons (after 5D classic), compared to the 400 / 800 / 1600 set. Also, much depends on how much shadow area is in the images and how crucial they are to the quality of the image.

In my tests of the 5D classic, I can see the difference in noise between ISO 400 and 500. Also between 800 & 1000 and 1000 & 1250. It's in the shadows.

gdanmitchell wrote:
C. The amounts of noise that folks are fretting over are not only entirely insignificant, but a bit of noise in an image often creates a more pleasing photograph than one in which noise has been overly reduced. In some cases we actually choose to add a bit of noise for this purpose.


Being an artist is about control of the materials and techniques.

It is much better for artists to have the choice to add the noise or not, rather than having the noise there without the choice.



Dec 06, 2012 at 11:40 PM
gdanmitchell
Offline
• • • • •
Upload & Sell: On
p.2 #19 · ISO sweet spots


Turns out I actually do know something about assessment and testing methods, though I don't often offer up the full story in photography forum posts. I'm boring enough as is! :-)

The test I propose is most certainly not "bogus," much less "purposefully bogus," though that does not mean that there are not other ways to test such things.

What I'm after is not so irrelevant to actual photography as suggesting that so-called objective equipment tests that use some arbitrary baseline method of conversion and apply it to all cameras equally tell us something useful about the results that photographers will obtain in real use.

For example, if camera A turns out to be "more red" (to make up a difference) than camera B - which could equally be described at "camera B is less red than camera A" - this tells us little of value if a) color balance is almost always adjusted in post and b) the results after typical post adjustments are essentially the same.

In addition, having some background in testing and the analysis of test results, perhaps you have heard of an alternative to your AB test, which I have heard described as an ABX test. In this type of test, three samples are presented to the subject. One is case A, one is case B, and the other ("X") could be either A or B, but the subject is not told which.

Several things are possible in this sort of test, and it addresses the psychological issues you raise, notably the tendency of test subjects to find a way to a) see differences whether or not they are there when they are asked to compare - e.g. look for differences - among two test cases, and b) determine whether or not these purported subjective ratings are based on something they can really see or not.

I first heard of this a few decades ago when an audio magazine used it to test the reported ability to distinguish between the audio outputs of some (at that time) very, very expensive compact disk players and the output of some rather inexpensive players. CD playback was synchronized between the A and B machines (one of which was the "high priced spread" and the other was the relative cheapie) and audio levels and so forth were equalized. The test subjects were, as I recall, given a three position switch with A, B, and X positions, and were invited to switch among the three inputs as they listened to music. The subjects included audiophiles, musicians, regular folks, and so forth.

Almost all, as I recall, expressed a preference between the A and B options. However, there was a second request, namely to identify whether X matched A or B. It turned out that those claiming prefer A or B were unable to determine in any statistically meaningful way whether X was their favored A/B option or not.

So, yes, I'm somewhat familiar with a range of issues related to testing.

Meanwhile, back in the world of photography, I think it is quite important to step away from the supposedly objective world of measurable but trivial differences between things and ask some basic questions about this focus. If we are so sure that A is substantially better than B, why is it that really fine photographers seem to be able to produce outstanding work using A and using B? If we are really concerned about the "quality" of our work, technically or aesthetically (he timidly proposes in the "technical" forum...) why do we focus so much on trivialities that make so little difference to our photography and spend (some of us, at least) so little time and energy focusing on the things that demonstrably do have a very real effect on the photographs we produce?

So many of the "differences" that we debate here - though not quite all of them - are differences that don't matter or differences that cannot be seen in the product of our work, either because they are vanishingly tiny anywhere other than the test bench and/or because they are utterly swamped by other factors.

I can "see" a difference between ISO 100 and 200 if I stare long enough and careful enough at things like pure sky or similar and jack up contrast or create a very steep curve and look at 100% magnification and compare two samples side by side. But I cannot see a difference between ISO 100 and ISO 200 in a large photographic print from my 5D2 - and, as a consequence, while I tend to shoot at 100 out of pure habit, I will switch to 200 when I want an extra stop without worrying about it in the least. I can more readily "see" a difference between 100 and 400 if I compare the same way (100%, side-by-side, smooth gradient or solid area, staring very closely) but I can produce a print from a ISO 400 image that shows none of this - I simply use a slightly different balance of NR and masking and so forth. On my screen as I write this is a natural light photograph of a violinist shot at ISO 3200, displayed at 100% and containing tones from pure black to pure white. I apply a slight amount of NR (20-25 in Lightroom) along with a bit of masking (about 40-50) of Amount:25 sharpening and it looks great - will make a fine 18" x 24" print.

About: "Being an artist is about control of the materials and techniques."

No. And the nature of "artist" and what constitutes "art" happens to be something that I know a lot about by professional training and decades of professional work. Much of it is mystery, but if by "control of materials and techniques" you intend to draw a parallel to something as mundane as reducing noise by some tiny increment in a photograph, you are very, very wrong. Control of the methods and materials of one's art is not the same as things like minimizing noise. To that point, the "control" of noise as an aspect of the creation of art often means decisions about the role of noise in the work and it can and does involve decisions to increase or control the quality of noise as much as to diminish it. If not, how to explain artists whose style is at least partially defined by using a gritty, grainy style.

Mastery of the materials of an art is often a characteristic of an "artist," but that simply means understanding those materials so well, to the level of intuition, that he/she can manipulate them it whatever way is best for the work at hand.

In any case, we do have choices regarding noise that makes a difference in our work. We can apply all sort of effective NR strategies in those few situations where this is necessary or we can add it in post if we think that is useful. Or we could even choose to shoot at ISO values that create a lot of noise and then leave it - sort of like how folks used to (and a few still do) shoot ISO 400 films and even shoot/develop to accentuate the grain.

I'll stop the philosophical and theoretical business about art here - though it seems to me to be far more interesting and important than the difference between noise at ISO 160 and ISO 200. To anyone who thinks that fretting about whether to use full or fractional ISO values will make a sensible difference in their work... I urge you to look elsewhere for things that will make your work better in ways that people will actually notice.

Take care,

Dan

Monito wrote:
Ordinarily I agree with you, Dan, but here I must differ slightly.

The test you describe is bogus. It is obvious you do not know how to do psychophysiological experiments; or if you do, you are being purposefully bogus.

Your suggested "test" is as ridiculous as showing a print of a landscape and asking someone if it was made with a SoNikon or a Canon or a Pentax.

The proper way to do this kind of test is to show a number of people pairs of prints, one at 100 and one at 200 and ask them to decide which is which. Then
...Show more



Dec 07, 2012 at 12:35 AM
Monito
Offline
• • • • • •
Upload & Sell: Off
p.2 #20 · ISO sweet spots


Very interesting discussion of ABX testing. Thank you.

gdanmitchell wrote:
Mastery of the materials of an art is often a characteristic of an "artist," but that simply means understanding those materials so well, to the level of intuition, that he/she can manipulate them it whatever way is best for the work at hand.


Exactly.

In terms of the current discussion it becomes very simple and intuitive. When shooting 5D Mark II, I shoot the ISO 160, 320, 640 and 1250 set. I shoot at 160 if I can, switch to 320 or 640 without a second thought when I need it, and pause briefly if I need to go to 1250.

I really don't need the values between (like 400 and 500). It is easier to adjust the shutter or aperture by a third of a stop.

It's that simple and intuitive.

When I shoot the 5D classic, I shoot ISO 100 when I can and go to 200, and 400 without a second thought, and pause a little longer than briefly if I need to go to 800. Similarly to the other camera, I don't really need the values in between (like 125 and 160).

Simple.

160 is no more or less "fractional" than 100 or 200.



Dec 07, 2012 at 01:18 AM
1      
2
       3       end




FM Forums | Canon Forum | Join Upload & Sell

1      
2
       3       end
    
 

You are not logged in. Login or Register

Username   Password    Retrive password