I was recently shooting my new G80 in glaring midday sun conditions (shooting an event, couldn't choose the time of day) and was surprised to see what seemed like a bit too much noise at the base ISO, with so much available light. It was easily noticeable on people's faces, but only at 100% magnification, nothing really terrible. To calibrate "a bit too much" - this is in comparison to shots under more normal/diffused lighting (including shots from my em10ii). I suspect it had to do with some relative underexposure as the camera might have been fooled by the overall light available, and yet the faces were partially in the shade. But there was also quite a bit of noise in the blue sky. I see in LR that there was easily at least a stop of headroom available without blowing any highlights (in fact the whole scene as sun drenched). I was shooting in shutter priority and the camera chose the f-stop and therefore exposure, no exposure compensation dialled by me. Unfortunately I can't share the pictures, sorry. Anybody has similar experiences and thoughts on what could be causing this unintuitive outcome? Shall I always try to expose to the right? What is the best way to consistently achieve it without blowing the highlights? Anything else that I may be missing? Thank you.
The light that your camera gets is dictated by the exposure set, not how much light is available. 'Underexposure', contrary to what some say, doesn't cause noise because it's 'under' some 'correct' exposure, just because it's a smaller exposure. The rule is simple, the smaller the exposure the more the noise. To get less noise, use a bigger exposure. If you work to your exposure meter, then the way to get a bigger exposure is to use a lower ISO. I generally use my GX80 (which I guess is much the same as the G80) at 100 ISO, unless there are bright highlights, which can get clipped at the 'Lo' setting.
Also, sky noise os a particular thing, not to be confused with general noisiness. Firs, viewed in fine detail, the sky really is quite noisy,
due to uneven scattering (it is the scattering that gives the sky its blue colour) and specular reflections off water droplets and ice crystals in the atmosphere. If you have a camera without an AA filter, such as the G80, these tiny bright spots can alias and become more apparent rather than being smoothed out and invisible, as they should be. Then, a camera's sensor is at its least efficient in the blue part of the spectrum, so blue things will always look more noisy than other colours.
You posted a similar conjecture about "uneven scattering" due to ice crystals in the atmosphere five years ago, and it was pretty definitively addressed at the time by kenw
here. Has something changed since then to prompt you to once again submit it as a contributing factor?
Do you keep a database of my transgressions? That was five years ago, and I had forgotten all about that conversation.
No, I just have sufficient mental faculties to recall a thread in which I participated.
Glad to hear that you're not as obsessive as you appear.
Good. Now that we've established that I'm not as obsessive as I appear, shall we discuss whether you're as paranoid as you appear?
I wouldn't think it's paranoid to be amused by someone keeping tabs of you sufficiently to remember the details of a five year old discussion and further, to have a ready link to that discussion.
The timing of that particular discussion was propitious because it occurred when I was doing a lot of processing of images from a trip that featured numerous shots that contained deep blue skies and waters and in which blue sky/water noise was a common issue. Your conjecture struck me then as pretty preposterous. It still does.
So, explain why you think it's preposterous, in your own words, not Ken's.
No, you don't get to shift that burden to me.
I'm not shifting the burden. In any sensible discussion instead of making insulting observations (it's preposterous) you give a reasoned critique. That much shows respect for the person you're discussing with. if you wish to discuss in an entirely negative and insulting way, rather than giving reasoned arguments, I have no obligation to discuss with you at all, and unless you make the decision for yourself to discuss this sensibly, I'll simply terminate it. Up to you. You know how it goes, you've taken this line many times before.
Left=one of kenw's test crops; Right=portion of kenw's test crop upsized 200% using bicubic smoother in PS
OK, that wasn't what Ken claimed..
I don't know how you can read Ken's post any other way than I am, but why don't we see if he's willing to clarify (I've PM'd him about this thread).
That's how disagreements happen. Two people interpret the text different ways. The way to learn is to explore why they are seeing it a different way. I didn't see Ken making the claims for his images that you did.
Moreover, the presentation of them in these little crops, at a very low lightness is not at all typical of sky, and doesn't make it easy to judge anything.
Odd that you didn't raise any of these objections back when Ken presented the results of his testing.
What I did then is not relevant to the present discussion. You can impugn my motives if you want, but talking that attitude will simply lead to the end of the discussion. I don't have to explain to you why after a gap of five years, when I review the discussion it looks a bit different to me now than how it did then.
So two questions.
i) Why do you think this is inconsistent with what I said?
You need to be more specific here. I don't understand what the "this" is you're referring to and which of your statements you're referring to.
If you don't want to discuss, you don't what to discuss. If you do want to discuss, you'll make a sensible interpretation of what I said and produce a response tailored to increasing understanding, rather than point scoring.
ii) You say these show 'typic;' sky noise. You ay what I said is 'preposterous'. What is your non-preposterous explanation?
Image noise. Plain and simple. In the case of blue sky, it's often the red channel that's polluted with noise. As I explained in my earlier post in the thread, the problem is triggered by a less than optimal exposure setting and then frequently made worse by adjustments that accentuate the problem (either in-camera or in post/processing). Subtle tonal ramps in blue sky can also lead to JPEG compression artifacts but that generally has a different more mottled look to it.
OK.
i) What do you mean by 'a less than optimal exposure setting'? My understanding is that as far as exposure goes bigger is better. I'm wondering whether you understand that there is some optimum other than bigness that we should be looking for.
ii) Please explain how JPEG compresses skies differently so as to leave artefacts in what should be a plain colour (JPEG is generally pretty good at that - it removes high frequencies, rather than adding them).
iii) Why do skies apparently behave differently from other plain colour areas? Or maybe you say that they don't.
So, I'd disagree that it was 'pretty definitively addressed'. In fact, coming back after five years, I think his case looks weaker than it did then.
You might like to look up the formation of light pillars, and you'll see that reflection off ice crystals certainly can cause visible effects in the sky. From the
Wikipedia article :
The crystals responsible for light pillars usually consist of flat, hexagonal plates, which tend to orient themselves more or less horizontally as they fall through the air. Each flake acts as a tiny mirror which reflects light sources which are directly above or below it, and the presence of flakes at a spread of altitudes causes the reflection to be elongated vertically into a column.
Easy to see how some random orientation of such crystals can cause random reflection in the sky, through to create a visible pillar, it need a particular orientation to the light source.
So what?
The so what is that there is a physically possible cause of the phenomenon I described.
We can also see rainbows. But rainbows and light pillars are relatively rare conditions.
Not the point. The point is that Ken claimed that my explanation was physically impossible because there aren't any particles in the air large enough to cause specular reflections. Yes there are. And we aren;'t talking about light pillars, we're talking about the ice crystals that cause ice pillars potentially causing other effects.
Again, you've provide zero explanation for how ice crystals many miles away are able to generate the specular reflections sufficiently large enough to look like well distributed and uniform noise but also very distinctly different from any well-known and well-explained naturally occurring atmospheric phenomenon.
This photo is taken from the Wikipedia article on halos
I don't think it's a stretch of the imagination to thing that the same phenomenon that causes clearly visible effects like this could, in a less ordered and more random alignment with the light, cause effects which are observable by a high resolution camera as being akin to noise.
Blue "sky" noise is common and easily produced under all sorts of atmospheric and other conditions.
Sure. All sorts of atmospheric conditions have ice crystals in the higher levels of the atmosphere. It's cold up there.
Moreover, the claimed "randomness" of the reflection can't be correct because the light source of the specular reflections is the sun, which is always at some fixed position in the scene.
But the orientation of the ice crystals will be somewhat randomised. That's exactly waht is required. Have you ever seen 'glitter'?
The glitter, if distributed like ice crystals will have to be spread in a rather uniform/normal way across the sky, which at the scale we're talking about and with the randomized positioning of the surface of each piece of glitter, I would expect the effect to behave similarly to any of the number of well-described atmospheric optical effects (halos, sun dogs, glories, etc.) depending on the relative position of the sun. There will be specular outliers scattered around that stand out relative to their immediate neighborhood, but those will be isolated and not extremely uniform across the entire plane as is the case with blue sky noise.
That is your expectation, not mine. I would think that ice crystals in many situations could quite evenly be distributed around the atmosphere (see the 360 degree halo), and if the sun isn't in the right alignment for a visible halo, all you'll get is a random brightness variation at a very small scale from those crystals that are randomly in the correct alignment.
Again, I invite you to prove me wrong with even one reputable explanation of a known atmospheric optical condition that depends on specular reflection and that looks extraordinarily like regular image noise.
We're discussing possible cause., and as I said, if these ice crystals can provide structured, large scale visible phenomena with the sun in the right alignment, there is no reason to believe they won't provide more randomised, less structured effects with the light in other alignments.
There's a reason why light "pillars" are pillars and not a uniformly random phenomenon across the sky (likewise with rainbows). Besides that, if it's a real visible phenomenon, then you'd expect ALL cameras under a wide range of exposure conditions to display it, but I've only ever seen complaints about it with respect to cameras that are relatively exposure constrained (e.g., high base ISO or otherwise insufficiently exposing the sky).
I don't know how many cameras suffer this effect and how many don't. I postulated that it's worse on ones without an AA filter, because sub-pixel bright spots will be aliased into larger effects. If that's the case, you'd expect some kind of dependency on the effect of the AA filter.
You're just guessing, here, but for whatever it's worth my mFT cameras and my D300 all have AA filters and all produce blue sky noise under the right conditions.
Sure, it's a speculation. But, I'm not inclined to take your observations as a data point, since it's very clear that your mind is closed on this matter. It's also not immediately clear that we're talking about the same phenomenon. In any case, as I remember at the time the D300 had a bad reputation for 'sky noise' compared with the D200. Your position seems to be that the 'sky noise' phenomenon, doesn't exist as anything different from normal noise. You may be right, but it seems to be quite often observed, so finding an explanation seems sensible, even if that explanation is that the people who observe it are delusional.
The theory simply doesn't hold up.
I still think it does. Your refutals certainly don't hold up, because they simply don't address the point being made.
What point?
Any point.
Blue sky noise is always uniformly distributed and changes in amplitude consistent with the amount of light captured by the sensor.
Is that the case? I'm not saying that you're wrong, but I've never, ever seen a rigorous evaluation showing this. Presumably, such a stickler for well founded arguments such as you wouldn't be making such a claim unless it had been rigorously verified, so let's see the source.
The source is my years of personal experience using high base ISO cameras that are prone to generating visible blue sky (and water) image noise. Since I tend to exposure-bracket to ensure optimized raw exposure and to extend my image DR through stacking, I end up with a lot of "underexposed" (for raw) frames that are comparable to the exposures other users frequently obtain when using conventional metering. With years of experience viewing these different exposures of the same scene, it's abundantly obvious to me that the variable that differentiates the frames in which sky noise is visible from those in which it isn't is simply exposure. I've literally done this "test" hundreds of times over the years.
OK. So, your position is essentially that 'sky noise', as a separate phenomenon, does not exist. Possibly that's the question to get out of the way before possible causes are discussed. If your starting point is that a phenomenon doesn't exist, then you'll reject absolutely any explanation as to what causes it.
It looks like typical image noise and behaves like typical image noise because it it IS typical image noise.
So, no , why is the sky particularly subject to ';typical image noise'; when other flat areas, even blue ones, aren't?
But they ARE, given the right set of lighting, metering and processing conditions. What particularly distinguishes image noise in the sky (which is usually blue, of course) from similar amounts of noise in other areas of an image with a similar EV is that we
know a priori that the sky should be uniform and textureless. Sky noise sticks out like a sore thumb but noise in other lower midtonal areas of images tends to get hidden by real detail. It's relatively rare to come across those other "flat areas" that aren't in the sky.
So, purely an expectation effect? That merits discussion.
And why do we get it at base ISO, where many will tell us that noise simply is not a problem? Why does it occur even in cameras where noise measurements tell us it shouldn't be a problem?
Because an ISO of 200 (which is the base ISO of all of my cameras in which I've been troubled by sky noise) is just not low enough when utilizing conventional (JPEG optimized) metering to ensure you'll get enough exposure in the deep blue sky portions, especially in the vulnerable red channel. If you don't/can't "ETTR" you can experience blue sky noise.
That should be testable.
I could post many examples, but below is a particularly telling one (it's one of the shots taken during the trip I was referring to earlier in my response). This shot was taken at Crater Lake, Oregon on an early October morning. It was unseasonably warm when I was there - short sleeve weather in the morning and downright hot in the afternoon. There is absolutely no way there were ice crystals in the atmosphere between the camera and the water, it was too warm for that. Yet, if you compare the water noise and the sky noise you will see that it's both uniform and consistent with the level of illumination.
[ATTACH alt="Oly EM5, ISO 200, f/4, 1/4000 (cropped frame from an exposure bracket). Processed in ACR to all default settings except noise has been zero'd out. Be sure to view in "Original" size."]1721304[/ATTACH]
Oly EM5, ISO 200, f/4, 1/4000 (cropped frame from an exposure bracket). Processed in ACR to all default settings except noise has been zero'd out. Be sure to view in "Original" size.
I wouldn't think that shows the phenomenon that people call sky noise. That's just normal image noise.
Hmmmmm...It's blue, it's sky and it's noisy. How silly of me to think it's blue sky noise! Please enlighten me on how one distinguishes normal image noise in deep blue skies from your special but elusive atmospheric phenomenon?
The key point here is whether or not there is a phenomenon of 'sky noise' separate for everyday noise. My comments were made on an assumption (which I'm prepared to admit, could be wrong, but only via a process of reasoned and evidenced discussion) that there was. Yours were made on a certainty that there wasn't. It might be a better approach for you to discuss the key points rather than making veiled and not-so-veiled attacks on my motives and integrity. Your choice, what could be an interesting discussion continues, or it stops here.
--
263, look deader.
