Noise is unwanted erroneous data. Ideally, there would be no noise, but the camera adds some (various sources, bucketed under the convenient term "read noise"), and light has a random quantum nature we call "shot noise".
Shot noise can't avoided, read noise can be controlled to a degree. Modern cameras have good control over read noise.
There is always some noise present
Noise isn't objectional if it is kept low compared to the desired data
The best way of controlling noise is to get more light on the sensor (controlled by level of ambient light, shutter and aperture settings).
ETTR i.e. allowing sufficient light on to the sensor to swamp the noise without letting so much in it saturates the sensor is one common technique used by photographers to minimise noise
Boosting ISO does not get more light on the sensor, it merely makes the midtones less dark (careful avoidance of disputed terminology!)
Both mischaracterisations. You're engaging in a very common (and ego based) fallacy which might be called "I'm a pro so I know." There are several problems with it. First, there's quite a few pros that don't know, so saying just that you're a pro doesn't prove anything. Second, it kind of assumes that you speak for all pros, and thus leads you to wrongly decide that anyone who disagrees with you can't be one. So far as Iliah goes, I've come to realise that he doesn't say anything without a great deal of thought. It's also a lot to do with learning style, as I said to Don B upthread somewhere. Different folks approach learning and its application to problem solving in different ways. Some take a trial and error approach, where they try different things, find out what works for them and adopt it. No problem with that. Others take a more model based approach - which is to try to understand how the system they are dealing with works then apply that knowledge to the problem they want to solve. In my experience those that adopt the second approach generally end up with better solutions more quickly, whilst those that take the first will often find good solutions, generally in a more time, depending whether you factor in the tie taken to understand how the system works or not.
ETTR can still be considered when using auto ISO in ultra low light, because having set your shutter and aperture you can still go to halve your ISO by applying 1 stop +EV to the right, subject to not clipping the highlights.
This approach will give you a bit more DR to play with in post for such challenging situations.
DR is measured between saturation and a noise floor (arbitrarily decided, there are various measures such as engineering DR, photographic DR) which put boundaries for acceptable noise at different levels
The ideal exposure for maximum DR is ETTR - it gets the maximum amount of light on the sensor
If you give less exposure that ETTR, noise becomes a greater proportion of the overall exposure and more visible
Raising ISO sometimes (depending on camera) reduces the read noise component visible in the deepest shadows but overall noise is dominated by shot noise so it doesn't do a lot.
As raising the ISO doesn't change the sensor sensitivity, it normally results in less light hitting the sensor - and thus more visible noise. Visible noise is one component in measuring DR, so raised ISO is usually associated with reduced DR. Raising ISO also raises values of highlights and drives them towards clipping so this could potentially reduce DR as well.
NB
I haven't addressed your other bullet points as I'm not sure what you mean by them.
Your all good Jim ,i wish others would join the conversation with there experiences. this is not a battle but a good platform for others to enter
with there own subjective or common views.
All I've said and backed up with examples of how to shoot in the real world is designed to help real world photographers.
When challenged it basically comes down to Ilah eventually saying "you do know who I am don't you?" which is a change of strategy.
That's because his first strategy of cherry picking bits of my posts to reply to (and out of context) wasn't working.
And then Bob you come in in and try and be clever and turn it round like I'm trying to be some superior pro.
Nice try but it doesn't wash.
What's happening here is scientists trying to prove how clever they are, then trying to be even more smug with cherry picking selectively the bits they want to reply to.
Then you pop up trying to be the ultimate cherry on top. But all you're doing is defending your scientist mates. Having probably not read the whole thread properly (in your defence).
OK. I could quibble with some of that, but let's take that as a working definition, at least for now.
Both can be controlled. Shot noise by controlling the number of photons incident on the sensor (which equates to exposure if we ignore sensor size). In the majority of cameras the ISO control offers some control over read noise, but not in a documented or standardised way - you need to investigate you camera to know how it works in this respect.
Yes
I find that to be so simplistic as to be unhelpful. In the best of all possible worlds, the best SNR is the highest one. How much or which type of noise is 'objectionable' is dependent on so many factors - usage - viewing size - characteristic of the noise - artistic intent - whether the consequences of reducing the noise are more objectionable - and on and on.
That's the only way of increasing the shot noise SNR. It has no effect on read noise.
That's not a definition of ETTR that I would subscribe to. As I said earlier, ETTR is a means to an end, not an end in itself. The end goal is to maximise exposure subject to your working and pictorial constraints. ETTR - that is increasing exposure until a displayed histogram shows that all the recorded exposures in the image are within the bounds of the histogram, but there is no free space 'to the right' - is one technique that can be used to achieve that goal, but there are others. One of the issues with ETTR is what the histogram actually represents. So far as I know, no present camera offers a raw histogram. There are hacks to try to get around that of various kinds, some more successful than others.
Nor does it get less light on the sensor - so if the exposure is limited by working and pictorial constraints, boosting ISO is not an issue. And since on most cameras, at least over some of the ISO range, boosting ISO will reduce read noise, it's worthwhile in those cases.
Theres a question worth asking pure ETTR advocates. If you've set the slowest shutter speed that you can whilst avoiding shake, and you've set the widest aperture that you can, which will give sufficient DOF, and the meter shows 'underexposed' at base ISO, what will you do?
1. Take the photo at base ISO with that exposure?
2. Increase exposure and risk shake or insufficient DOF?
3. Increase the ISO?
4. Decide not to take the photo?
and why is your choice the right one?
Number 4, of course, real photographers and all that jazz ;-).
Seriously, #3, increase the ISO but stay clear of histogram's right size or zebras/blinkies. I would increase the ISO to make the EVF sufficiently bright for framing.
I feel you are just picking in holes in the clarity of my writing and nothing you have said is in disagreement with what I've said (if you allow yourself to see past the loose words and concentrate on the meaning). This is a disagreement over writing style. I feel you are demanding an unrealistic standard. This is not a thesis, it's a chat. I'm happy with the sloppiness and imprecision as long as we both know what the other means.
Nothing you've said confirms any material conceptual error on my part, just picky points not really relevant for field use. I'm happy my concepts of exposure, noise, DR and ETTR are good enough for practical purposes.
Talking of picky, I'm not sure why you are picking on my use of the ETTR term, either. It's not a technical term, it's just shorthand for "expose as hot as you dare". One has to decide on some sort of exposure strategy. Mine is to point the camera at my subject, and use the preview histogram to maximise exposure, ignoring what that does to the look of the image on the screen. Collect as much light as I can without blowing essential highlights. Once I have the image, I check the post capture histogram just to be sure. That's all I mean by ETTR. I know the histogram is not a raw histogram, I know the same goes for zebras and the like, but you've got to use something and that's what I use.
To answer your last question, previously I would have taken the exposure at base ISO and reduced shutter speed and damn the consequences to dof and camera shake. Then I realised that you can fix noise in post but you can't fix camera shake, so I switched to an auto ISO strategy. Then I learned from the mega threads here that ISO is not what I thought of that so I have changed to leaving everything at base ISOm choosing whatever shutter and aperture I need and just accepting less light on the sensor (raising the midtones in post). Jim assures me that the read noise benefits of raising ISO in dim conditions are minor and I won't be leaving a lot on the table with this approach. It simplifies things a lot as I am currently using 6 cameras, too many things to remember.
Agree 100% i was shooting test cloud images yesterday as they were moving fast and the lighting was changing every 10 secs ,so shooting an event i use A and float iso/shutter speed with a min shutter set to 1/250 sec.
Thank you for clarifying the term, but be aware that others may use it differently. I did not understand why you wrote that raising the ISO is effectively underexposure? What you probably meant is that you increase the ISO when you have "underexposure" or that you can raise ISO when you have "underexposure."
I understand its use. Some people just want useful advice for field work, others are having an academic discussion. What matters for each scenario can be different.
Technical discussions can get very heated and quibbling over the finest distinctions is the norm. A lot of the time, those quibbles are irrelevant for field use. If you are out there approximating everything to the nearest stop, who cares about things that affect the sixth decimal place.
The problem, of course, is context. Exactly when are the quibbles material for field use? That can get tricky. And if the discussion is not relevant to field use and is in the academic domain, of course the sixth decimal place is important!
