Help with ISO

[Geoff Port]

Swifty,
Here are a couple of shots that are typical of my problem. I expect you will be able to extract the exe data from them.


This good looking bloke is me sampling a range of ciders. The only tweaking I have done is to sharpen by two clicks in Lightroom.


This shot is of one of the other tour participants. Once again I adjusted the sharpness by two clicks, leaving all other setting alone.

Geoff P

[arthurking83]

Hey Geoff.
Moved your post in the AF Help thread, so as to not confuse and confound that thread.

For the benefit of others reading this, the issue raised by Geoff Port started HERE.

[Geoff Port]

OK Arthur thanks for the heads up.

[swifty]

We're going a bit OT but if the OP doesn't mind, we'll press ahead.
I'm at work so I'm not able to download the photos and view them on my regular imaging software or extract the exif info.
But to facilitate discussion, can you tell us some info if you have it handy:
ISO and exposure info.
WB
Jpeg or RAW.
Whether these are the full images downsized by photobucket or cropped. If cropped, how much are they cropped from the original.

A few things I can comment on immediately though.

Firstly is the light source doesn't appear full spectrum (probably tungsten?). Most cameras will try to compensate with auto WB but that has the effect of lifting data from a colour channel with very little data. Eg. lifting blues for tungsten lighting. So in effect you're lifting the noise in those channels and making it more visible.

Secondly, I don't know how photobucket resize their images. Some hosting sites have terrible resizing algorithms and results in quite poorly presented images, robbing it of some of the sharpness of the original files. Does your originals look sharp in LR at least?
What lens were you using? Do you roughly know how well it performs optimally? Eg. In good light you may be using it a bit stopped down a bit but in poorer light you might be using it wide open which just might not be that sharp for that lens.

Shapening globally also sharpens the noise making it more visible. Try the 'masking' slider in your LR sharpening panel so you're not sharpening the noise in the smooth area.

PS: how were the ciders? I could do with one right about now.

- - - Updated - - -

Oops.. what happened to your post and images, Geoff?

[swifty]

There's your post. Arthur, do you mind moving my response too.

[arthurking83]

I'll start commenting on the topic at hand now!

1. whitebalance is wrong. Too yellow/brown/warm. Strangely, WB setting can affect the quality of the noise level in an image too!
May sound weird, but it should be remembered that WB is just another type of exposure control, affecting red and blue channels quite substantially, and most 'visible' noise is in the red and blue channel.

** delving deeply into a parallel tangential mode now, and risking life and limb and all manner of boredom inducing possibilities ... but what we normally refer too as noise in an image is usually lots of red and or blue noise.
Green channel noise is also there, but it's less annoying to most humans(well at least to me).

Image doesn't look too noisy at the presented scaling!
It may look noisy to you, but I suspect that you tend to view the image hugely magnified compared to the size we have to view it at!!

2. So it's an important aspect to assess the level of aberrations according to the final image output.
That is, complaining that ISO6400 on the 200 Mp camera ABCmkMMXVIII makes no sense if you only make 2Mp images available for the world to see.
Just the subsampled(or downsampled/dowsized) file creation takes care of that aspect automagically for you.
If you crop heavily tho and then make available 2Mp files of a 2Mp cropped image from this hypothetical 200 Mp camera monster .. then it's probably a good idea to get a 2Mp model version of this hypothetical camera.

I don't see cropping or any reason to crop the images above tho .. just more useless info above!

if the images were shot in raw format, first thing is to adjust WB. If you can't find a good WB setting based on the light conditions at the time, click to grey is a handy way to get good WB setting. \
if you don't have a click to grey WB dropper tool, get software that does!

Only issue with using the dropper tool for WB on high ISO images, is that you may click to a noisy set of pixels! Easy to get around that tho .. just keep clicking. Zoom to 100% pixel view(magnified in) to get a more accurate dropper selection.

If you need help with any of that, just let me know. I'm sure I can rattle off more useless info for 'ya!

Can see some exif info in the images, but not the really important stuff.
Can you note down the metering mode used for those images and any metering compensation set into camera.

In terms of focus or sharpness, way too hard to determine any errors/flaws at the presented image size too.
if you're concern is about focus/sharpness/blurriness, once again you need to take into consideration the size of the images presented.
At those sizes, even on my massive monitor, they look fine.

[swifty]

If the light source lacks a certain colour, correctly white balancing the RAW file will still make the image noisy because data is missing in that colour channel. And sometimes other side effects like colour shifts and poor colour saturation etc.
If the light source lacks a certain colour and you shot in jpeg with the wrong WB written into the jpeg file, it is even worse when you try to correct that in post.

[Geoff Port]

Shot #1 ( the handsome drinker)
Lens - Tamron SP f2.8. AF- ON. VC - ON.
Focus -58mm
ISO - set to Auto. Shot @ 6400
Exposure - manual mode
1/500th @ f6.3
Taken in RAW (10.1mp) and converted jpeg when exported to photobucket.

Shot #2
Lens - Tamron SP f2.8 24 - 70mm. AF- ON. VC - ON.
Focus - 40mm.
ISO set to Auto. Shot @ 6400
Exposure - Manual mode
1/500th @ f6.3
Taken in RAW (approx 10mp) converted to jpeg when exported to photobucket.

Metering mode for both - averaged.
EV - 0

[arthurking83]

Ah!
with that additional exif data:

1/500s, VC on, ISO6400

Camera settings to watch for:
you had quite a lot of shutter leeway to reduce ISO.
Technically speaking at the focal lengths used, you could have aperture leeway to spare there too. I reckon F/4 would have given more than enough DOF at that focus distance range too... and the Tammy is plenty sharp at f/4!

So start with the basics and set a shutter speed that always tries to keep ISO to a mininum maximum(if that makes sense).
No problem with high ISO when required, but minimising that high ISO value is the priority.

You had VC on, so even taking into account possible subject movement, you could have set camera to 1/100s easily, and for people 1/60s is plenty fast enough(unless they're moving rapidly).

So using 1/100s as a minimum, ISO would have maxed out at something like ISO1600 or so.
Taken a step or two further, and used f/4 aperture, another stop and a bit saved in ISO .. so more like ISO640-ish would have been set.

One of the reasons I'm not a big fan of manual mode .. you tend to set up some parameters, which kind'a make sense for one scene, but then play havoc with getting the most efficient settings elsewhere in the camera.

Having set Aperture Priority(eg. to f/4 or even f/5.6) and in shutter speed to 1/60s if you're confident, or 1/100s 'just in case' .. auto ISO would have got you in the ISO400-640 region instead.

I think this is where a lot of people don't like AutoISO .. it's not the feature itself that is the problem, but that many people don't fully understand what it's trying to do for 'ya.

BUT! forgetting what could have been, and working on what is .. some suggestions to process the images(using my not so capable memory of LR, or Adobe products in general tho!)

on the raw image, first port of call for processing would be to use the white balance dropper tool to try to white balance the images:
#1, I'd say try your shirt as a white point. so point the dropper tool on that area(collar).
Note that white point is a bit of a strange term too, and the dropper tool doesn't necessarily need to be used on a white subject matter.
Grey and black work just as well, but they have to be pure versions of those colours.
Can't tell the actual blackness of your jumper, but I'd say that could work well too.
Just remember tho, you can click onto small areas and get different results in the white balance rendering even tho you think you're clicking on white.
This will be due to random noise(chroma noise) where the white patch will contain varying degrees of red and green and blue pixels as noise.

On high ISO noisy images what I sometimes do, with various degrees of success, is to increase exposure compensation first to decrease signal to noise ratio on the white area.
That is, it whitens the white, removes some of the RGB chroma noise(relative to the now blown out white pixel data), try to get a good white balance from that area, then drop exposure compensation back to a more normal value again.

Once you have a good WB rendering, I found that the noise reduction tool in LR4 and 5 worked quite well. Actually about the only feature I really liked about Lr back then.
It had good control over chroma noise, and used that as the only sharpening tool on the raw file. Removed any other sharpening that Lr used to automatically do of it's own accord by default.
Sharpening a noisy image only serves to sharpen those high contrasting noisy pixels, so it gets grainier.

Image #2. Not tyring to be funny but I'd try the blokes grey hair as a WB dropper point!
Sounds funny, but find a thick mass of his hair and try it.
If it doesn't work well, the other points I'd try are the napkin looking things in the background(just near his forehead).
Chances are that those napkins were probably the bleached white types, and they make a good WB reference source.

Rest of the steps I'd do as per my comments in image #1.

Swifty made the comment that the light source may not have been full spectrum, but I don't think that's the problem with the WB rendering.
To me it looks more like lots of ambient light coming through, either an outdoorsy area, or a very large window allowing plenty of natural light coming through .. plus some additional artificial lighting from the cidery.
Unlikely that the cidery used tungsten lighting, and those god awful, CFL types are the more commonly used types nowadays .. and they're the culrpits in the WB rendering.

The overall assessment of the images and the quality of the image at that ISO6400 level tho is that it's very good, and with a bit of mucking about I think will come up looking OK.
There is very little to non existent graininess due to high ISO levels .. so high ISO noise shouldn't be an issue in these images.

[gcflora]

I'll start commenting on the topic at hand now!

1. whitebalance is wrong. Too yellow/brown/warm. Strangely, WB setting can affect the quality of the noise level in an image too!
May sound weird, but it should be remembered that WB is just another type of exposure control, affecting red and blue channels quite substantially, and most 'visible' noise is in the red and blue channel.
<snip>

I have to disagree First, I see nothing really wrong with the white balance on my calibrated monitor. Second, there is no such thing (IMO) as a "correct" white balance -- only what the photographer wants to achieve. For 99.9% of my landscape photos I leave the white balance set to daylight because auto takes all the natural colours and tries to make white white (e.g. taking away the reds/oranges/purples in a sunset that you can see, if you look, being reflected off white surfaces. To me, in this circumstance, auto white balance is incorrect even though from a technical perspective it might be correct). Another example is warm lighting. I happen to like the warm lighting of a fireplace or low lights lighting a room *shrug*. Maybe I'm weird because people say we "see" white even if it's tinged by the ambient or main light source, but I don't -- under tungsten lights I know my paper is white but I don't "see" it as white -- I see it as orange/warm

[Tannin]

The white balance is a little odd. You are not using Adobe RGB in your raw converter are you? It has a bit of the dull, flat feel you get with pictures processed using Adobe RGB instead of the standard SRGB. To agree (more-or-less) with Gcflora, in this case it rather suits the feel of the subject, however, though it does make the cider look more like beer.

As Arthur says, 6400 ISO is way too much here. You want to be somewhere around 125th of a second (which would give you 1600 ISO) or 1/80th (which would give you 1000 ISO).

Also as per the Gospel of Arthur, f/4 would be perfectly OK for these two shots (and get you down to a very clean 400 ISO at 1/80th. I'd have used something like f/2.8 and ISO 400 for the second shot (where you don't really want the depth of field) and let the shutter speed drift up to whatever setting the exposure system wanted - 1/160th in this case. For the first shot, with a greater focal length and a good case for a nice sharp background, I'd be looking to use f/5.6 (or the 6.3 you did use) and consider going to f/8 if the shutter speed stayed sensible and the ISO under about 800.

All of this stuff is much, much easier to do in aperture priority mode with manual ISO. (Why? Because that's the way I've always done it. Also, it's set down in the Gospel of Tannin. Look it up: chapter 7, verse XXIV. What better reason could there be?)

[swifty]

I’ll have to agree with gcflora regarding the point about WB. It’s not wrong/right thing but rather what your intentions are. I don’t always want white to appear white in my photos and often trying to do so has drawbacks to the image quality, particularly if the source light is not full spectrum like most artificial light.

I’m only guessing about the light source in these. As Tannin points out the colours look a bit flat and dull which led me to speculate about the light source and the chosen WB.

But your chosen exposure settings will have the same negative effects here. There’s something like 3 or more stops of exposure leeway either through shutter speed or aperture or both.
2^3 = 8 so you could be getting at least 8x more light on that sensor. Provided you don’t blow the highlights in any channels you could use a greater exposure setting with a potential lower iso setting.
Here it gets a bit fuzzy cos the chosen iso will depend on your sensor characteristics and how you choose to post process.
I know for conventional wisdom, best practice is to use the lowest iso possible for something that looks ‘right’ on the camera lcd but from my understanding it really depends on how your camera implements iso.
For some cameras, as long as you don’t blow the highlights you get better results going with a higher iso (up to a certain point) but shooting ‘overexposed’ then pulling back in post.
For other cameras it’s the opposite. Shoot ‘underexposed’ and push up in post. A true ‘iso-less’ camera with no read noise added would be an ideal candidate. The modern Sony sensors are getting quite close to being iso-less.
But now we’ve introduced a need to post process so the editing software ability comes into play. If you like to shoot what looks ‘right’ straight out of camera then just ignore my whole section about pushing or pulling in post.

But it’s important to remember exposure is affected only by aperture and shutter speeds. Not iso. I’m aware of the concept of exposure triangles but that is only a convenient learning tool and in reality iso is not part of exposure. Only the aperture and shutter speed will alter the amount of light hitting a given sensor or film which is what exposure is.

So for ideal image quality you always want to maximize exposure where possible. And how you want to choose iso will depend on sensor characteristics, how you like to shoot, scene DR etc. Its actually the same concept as ETTR - maximizing signal. I’m just talking about optimizing your data for image quality here. In real life of course you’d need to consider the creative and technical effects of your chosen shutter speed and aperture for your scene.

Going back to your original images, possibly the light quality may be poor and/or exposure as shot is suboptimal has resulted in dull colours, unnecessary increased noise etc. I’d be tempted to do a BW conversion.

[Geoff Port]

Thank you all for your in-depth input here. I have a lot of information I need to assimilate now.
A few comments after reading your posts.
I have always shot in Aperture priority mode prior to this trip to Tassie.
I nearly always attempted to shoot with ISO set on 100 thinking that I would be reducing the risk of grainy shots.
Nearly all my shots in this mode seemed to me to be over exposed and no matter what I did I could not bring it back satisfactorily.
This trip I decided to run with manual mode letting the ISO levels take care of themselves. I have not used manual mode on this camera before so I was running a bit blind as it were but on the camera my captures seemed to be working well.
I always attempt, and I emphasize attempt to capture scenes as close to natural as possible. ie; If light levels are low, natural and cast a warmth over the subject and the background is dark and shaded then that is what I attempt to capture. I don't particularly enjoy tweaking with LRoom because I am producing a fake image but I do it out of necessity. (I attempt to fix by stuff ups.)
The information you have all given me will be invaluable to me going forward and I thank you once again.
When I first purchased the 60D I thought, "now I can take good photos. THis machine will set everything up for me and leave me to only worry about composition." Hahaha. A rookie error it would seem.

[Tannin]

Cheers Geoff.

It is quite common for cameras to consistently expose other than as you expect. I've never explored the issue properly because it is so easy to fix. (More on this in a moment.) My 5D II does it, as an example. My guess - and it is just a guess - is that there is some dirt or other muck partly obscuring the metering sensor, and that a trip to a camera repairer for a clean would sort it out. But I've never got around to it because it is so easy to live with. I simply set the exposure compensation 2/3rd of a stop down and get the exposure I expect. (I.e., about the same as the exposure I get with other cameras.) All I have to do is regard the "-2/3rds" mark as the "centre" marker and work normally from there. If, for instance, I have a scene where I'd want +1/3rd on the 7D or the 1D III, I dial the 5D II up to -1/3rd; for a scene where I'd want -2/3rds on the 7D, I select -1 & 1/3rd on the 5D II. Sound difficult? Honestly, it isn't hard at all. You very quickly get used to it and it becomes second nature. Even easier for ]i]you[/i] because you don't have to remember which camera you are using and switch mental gears. (Or, of course, you could always take it somewhere to have it adjusted.)

ISO 100 isn't any cleaner than ISO 200, at least not enough for any normal human to see, and ISO 400 is still just fine for most purposes. So why do people use ISO 100 at all if 200 is just as clean? Because ISO 100 gives you more dynamic range. For a scene where there is not a huge difference between the darkest part of the picture and the brightest, ISO 200 is perfectly OK, and ISO 400 generally fine. You want ISO 100 for things like scenes with bright sky and darker ground. (Use it any other time too if you like, provided you are OK for shutter speed, but never be afraid to go to 200, and bear 400 in mind in case you need it. ISO 800 and up are best reserved for situations where ISO 400 simply won't do.

There are a lot of fans of auto-ISO on this form. Fair enough, whatever works for them is fine, but honestly, I reckon it's a weird and not-so-useful way to work. Av or manual every time for me. Aperture priority because your aperture is the first thing you want to be aware of: it essentially controls everything else.

In normal use, aperture is the only factor in the exposure triangle which is creatively significant. The other two only matter if you screw them up. Fairly rare exceptions (mentioned below) aside, there is no "perfect" shutter speed or ISO for any given shot; they just need to be as high (shutter) or as low (ISO) as possible in the circumstances. (The exception: shutter speed can be important if you are deliberately trying to blur moving objects; typically this is only for certain genres such as sport, propeller-driven aeroplanes, or those awful frozen water landscape things. ISO is never important unless you muck things up and get too much noise.)

In short: of the three factors in the triangle:

• Aperture is by far the most important. This is why aperture priority makes so much sense: you are directly controlling the thing that matters most.
• Shutter speed matters less. Mostly, you don't care too much about it so long as it is X or higher, where "X" is the lowest speed you judge will give you sharp results. For portraits, as an example, most 'togs regard 1/60th or perhaps 1/100th as the minimum and don't care if it drifts higher. For my bird work with long lenses, I generally aim for 1/1000th or better - more with certain subjects because they move so fast. (Again, there are the genre exceptions I mentioned above, but they are just that: exceptions.)
• ISO matters not at all so long as it is low enough to give you the clean results you need. Most people set it to something reasonable and then forget about it in order to concentrate on the creative factors, only adjusting it up when the desired aperture and shutter speed cannot be obtained, and adjusting it down again when, at the desired aperture, the shutter speed is lots higher than needed for the shot.

[swifty]

Hey Geoff, go with what you’re comfortable with.
Most of what I described are shooting styles that optimize data.
And that’s not always the aim and I don’t always follow my own advice either although i’m moving more and more towards that style.
Most of the time I shoot aperture priority and playing with exposure compensation but in non-changing light, you could just set your manual settings and forget about it until you leave that environment, in which case just a single change back to aperture priority. Auto-iso doesn’t even exist on my dinosaur DSLR but if it did, I would probably use it if it gave me enough parameters to set it up properly.
A final note, if your aim is good accurate colours I have found that trying to achieve SOOC perfection is the best practice. Maximizing data is all well and good for tonality and PP flexibility etc. but colours are sometimes a bit off, especially skin tones. I think I know why but haven’t really researched it so I’m not entirely sure if it’s my camera model anomaly or across the board. When shooting portraits I try to nail it in-camera usually.

[Tannin]

A final note, if your aim is good accurate colours I have found that trying to achieve SOOC perfection is the best practice. Maximizing data is all well and good for tonality and PP flexibility etc. but colours are sometimes a bit off, especially skin tones.

Good point, Swifty. I think you have to have both a special kind of eye, and a highly developed touch with developing software to achieve accurate colours from an original that isn't right to begin with.

The first ability - the eye for accuracy - is akin to having perfect pitch (the ability to hear a note and say "that's a c#" without having any other known notes to compare it to). People sometimes say that you are either born with perfect pitch or you are not, but that is untrue. Some people (probably most people) can learn it, and you lose the ability if you don't practice it now and again. Presumably, the "eye for natural colour" is similar.

I can't comment on the second ability, except to say that I don't have it. (To be fair, I haven't tried terribly hard to learn it; possibly it is within the ability of many of us if we are prepared to work on it hard enough).

The last point to make in this context is that a lot of photographers seem to have a well-developed ability to turn any given scene into a "perfect scene". By this I mean that they have a clear idea not of what natural or accurate light is, but of what a scene might look like on a "perfect" day. Their work looks very impressive to begin with, but after a while you start to see that every picture they produce is (so to speak) in the key of C Major with no sharps, flats, or accidentals.

For all of these reasons, I habitually shoot with the intention of producing an image in-camera that is, bar minor cropping, a finished product, complete with daylight white balance to properly reproduce the variation of colour tone that each day provides. (Of course, there is often a gap between intention and result. Much too often!)

[arthurking83]

.....
I have always shot in Aperture priority mode prior to this trip to Tassie.
I nearly always attempted to shoot with ISO set on 100 thinking that I would be reducing the risk of grainy shots.
Nearly all my shots in this mode seemed to me to be over exposed and no matter what I did I could not bring it back satisfactorily.
...............

two ... three reasons you would get over exposure in Aperture priority mode

* exposure compensation set to a +ve value
* not enough shutter speed. eg. you needed 1/16000s but camera maxes out at 1/8000s. Alternatively aperture in lens not operating correctly(sticking)
* metering mode set to spot, and directed towards a dark area with zero exposure compensation(where some -ve exposure comp would have been appropriate).

Otherwise as Tony said, probably an error with the camera.

Post some examples with exif, or check the exif for yourself(and post here).
The items of interest in the exif in those over exposed are: metering mode used, exposure compensation set, and exposure triangle settings used.

EDIT: stuffed up on point 2! brain fade. edited that point to shutter speed!

[Geoff Port]

THank you all again for your advice and the sharing of your knowledge.
I this Riverina environment dust could well be a problem so I might just take it in for a service.
Metering mode may well have been my major mistake I have been using center weighted most of the time in the past and occasionally going to evaluative metering without really having any real knowledge or plan as to why.
I don't know what SOOC is swifty .
I'm one of those members of the human race without a natural anything so I really work hard at "seeing"the photo in a scene. I know its there but I have trouble finding it. Having said that, I took a pic of the early sun striking our bookcase this morning and really liked the result. THe shot just stood out for me.
I have been playing around with the camera over the past 24 hrs. All my shots have been taken indoors with numerous including the outside light through the lounge window forcing me to deal with major variations in light intensity.
I used AV and ISO 100. Using your advice as a base I was able to take shots that resulted in reasonable exposure all over. Very pleasing.
The greatest change in the results for the better came when I changed from center weighted to evaluative metering.
I have an A4 version of the 60D manual. I read and re- read all sections on exposure, metering ISO and several other sections. Amazing what I discovered / rediscovered.
Thank you all again for your excellent advice and patience. I really hope others read this thread as there is so much to be taken from it.

[swifty]

Straight out of camera (SOOC)

[Tannin]

SOOC stands for Swifty Often Only Confuses.

The great thing with metering modes is to pick one and stick to it. Doesn't really matter which one, just so long as you pick one and become familiar with it.

Evaluative is good because it is significantly smarter than the others and is more likely to give you a good result, and bad because it's smart enough to be hard to second-guess, meaning that adding +_ exposure compensation becomes a bit of a lottery as you never quite know what the clever little computer is going to do next.

Centre-weighted average is nowhere near as smart but does a good job most of the time and is much easier to predict. I can usually look at a scene and say "that will need +2/3rds EC" with centre-weighted average metering, and get it right first time without chimping. Good luck trying that with evaluative.

Spot is good because it places you in control and you can meter off exactly the thing you want and nothing else. It is particularly useful for take-your-time manual shooting where you meter off two or three things and decide on a suitable compromise exposure setting. It is bad because you can't just frame and shoot, you have to fiddle about metering first because the point you want to expose off seldom happens to be in the exact centre of the frame. Spot metering is useless for fast-moving action of any kind and at its best with predictable subjects like landscapes, portrait victims subjects, and soup tins. People who spot meter everything wear sandals with socks, keep their small change in a shovel purse, and still miss film.

Partial is good because it offers many of the advantages of spot but is much more forgiving. You can very often shoot normally with partial metering (i.e., not meter and recompose, spot-style) and get a sensible exposure most of the time. Partial is bad because it's not as accurate as spot and thus more easily fooled, but doesn't take the whole picture into account the way evaluative and CW-average do. Neither fish nor fowl.

Oh, and it's possible that some people might think that SOOC actually stands for Straight Out Of Camera.