Forgot Password
Pentax Camera Forums Home
 
Log in or register to remove ads.

Showing results 1 to 6 of 6 Search:
Forum: Pentax K-1 & K-1 II 05-09-2018, 01:01 PM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
That is very interesting, I was not sure how much of that was due to the noise in the sensor vs incoming noise. I do indeed believe that noise in nature, such as poissonian arrival times, are a thing, although I am unsure how large the effect is. To some degree that is of course testable with my own camera, I've simply never thought to do it.

Practically, I (at least partially) agree with Normhead though. Even though the noise is really there, and my camera might be good at observing it, I will tend to get rid of it if it does not resemble my memory of the scene.
Forum: Pentax K-1 & K-1 II 05-09-2018, 11:19 AM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
Maybe I should not be replying anymore, but it is so hard to resist. I did not state anything about what I prefer in the entire detail-vs-noise problem, except that where possible, I'd like to be able to choose what I get. I am not implying there is anything wrong with what you want from your camera.



In no way do I pretend that human perception is in any way different or not bound to the laws of quantum mechanics. I do not see where I wrote anything that implies that. Neither do I pretend that the complex human part of vision does not exist. But thanks for telling me what I think and believe anyway.



That is the million-dollar question though, isnt it? In what way can we produce a camera that perfectly captures the vision part of the human experience. The camera is at a huge disadvantage here: our brains have access to temporal as well as spatial resolution for filtering, which helps a lot, and our brain is good at filling in gaps where details are missing. Within the much more confined limits of a camera, where noise has to be identified purely from its spatial characteristics, how do we best identify the noise, what do we do about it, and at which point do we sacrifice too much detail? How do we recreate 'the human experience' with a lot less information available to the camera? I think these are very interesting questions.
Note that in writing this, I am not at all trying to imply that you do not already know all of this, I am just trying to have an interesting discussion.


I'll be the first to admit that whenever such a question presents itself, I easily let myself get drawn in by it. That is just curiosity though. Why is it that wanting to know how something works in detail means a person forgets they are an ordinary human? Isnt to wonder at how things work a very human thing? Also, such curiosity does not at all mean that I cannot appreciate pictures that have noise or are not perfectly sharp. Appreciating good photography, enjoying making bad photography myself, and wanting to understand all of the things we are discussing here do not exclude each other at all.


Thanks! It is always good to read some more Sacks. I was unaware he was a physicist though.

---------- Post added 05-09-2018 at 08:49 PM ----------



Maybe I am confused between those? I didn't think I was confused, but now I am. How does low rod-cell density factor in here?



At the risk over being overly nitpicky on exact phrasing, the amount of shot noise goes up with photon flux. It just goes up less quickly than the actual signal, resulting in a better signal to noise ratio at higher light levels (the signal goes as N, the fluctations as sqrt(N), so the signal-to-noise ratio N/sqrt(N) grows with sqrt(N)). However, the better our cameras get, and the more dynamic range we have, the easier it is to detect shot noise.

In the meantime, the problem is that when we do low-light photography at higher ISOs in a digital camera, we take a low input signal, containing a higher fraction of contributions due to shot noise, and then amplify the whole including the noise. I am not stating anything about how much of the high-ISO noise is due to shot noise, because I do not know how it compares to, for example, noise added by the amplification process. But shot noise does get more important with more amplification (ie, high ISO) and less input signal.
Forum: Pentax K-1 & K-1 II 05-09-2018, 06:33 AM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
I went to quantum mechanics because those are the most fundamental laws we know, and they clearly state that there is no signal without noise.

In any case, the human eye can and does perceive noise. In low light, you can actually 'see' noise if you pay attention to what you are seeing. The brain is wired to to give noise a super low priority when processing what we see, so we tend not to notice, but we do actually see it.

The rods in our eye are actually sufficiently sensitive to detect single photons with a reasonable probability. However, there is a quite a lot of light lost between the cornea and the retina, and even if the photon reaches the rod, stimulating a single rod does not generally lead to an actual visible even (even though it can be measured that the rod was really stimulated). It is thought that there is a correlation layer in the retina to reduce noise by only propagating signals which have some spatial or temporal correlation, and also that the visual processing centers in the brain reduce noise. After these processes and the aforementioned losses in the eye, it seems the threshold for detection becomes about 70 photons (source for the number 70 ) .

In other words, it is thought that the human eye does a fair bit of noise reduction, both directly in the eye and in the brain.

As an interesting side note, some animals are more sensitive to light than we are. Frogs have been shown to be pretty good single-photon detectors
Forum: Pentax K-1 & K-1 II 05-08-2018, 09:27 AM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
This is almost the opposite of what is true. There is no nature without noise. It is prohibited by the most fundamental laws of nature (if we assume quantum mechanics to be correct). One of the ways this is clear in photography is by the poissionian arrival times of photons onto the sensor: light is quantized in photons, and the photons do not come into the camera in an entirely smooth continuous stream, rather showing (in most cases) some amount of randomness (according to poisson statistics) in their density. A bit more info here.

That said, obviously, there is added noise in the camera electronics as well.
Forum: Pentax K-1 & K-1 II 04-29-2018, 02:24 PM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
In my opinion this statement is way too black and white. Rather than right or wrong, it seems much more likely to me that the engineers fully knew what they were doing and chose for this particular compromise between NR and sharpness, based on what they perceived was most in demand by the market. Landscapes are often shot below ISO 800, where this is not so important. In astro isos are higher, but I do not think that the in-focus-to-out-of-focus transition is important/existent there. Maybe (and this is pure speculation) it was decided that for things like indoor shots lower noise out-of-camera would be more favourably received by customers than ultimate sharpness.

In either case, no need to tell some people that they simply don't get something which you perceive as obvious truth. Its a design decision. People are free to agree or disagree with it without being right or wrong.

In any case, even though I have neither the K-1 nor K1 mark II (yet), I'd also like this behaviour to be optional.
Forum: Pentax K-1 & K-1 II 04-29-2018, 11:48 AM  
K-1 MK2 - Unhappy with results at moderate ISO settings
Posted By xandos
Replies: 561
Views: 45,892
Sorry for the late reply. As you seem to be quite knowledgeable about these things, I have a question for you. Specifically because I think the reverse is true: as long as the data lives as small voltages inside electronic systems, there is always a noise source. Johnson-Nyquist noise immediately comes to mind. How important that is depends on exactly how the data is digitized. Do you have solid reasons to assume such noise can be completely neglected?

---------- Post added 04-29-2018 at 09:06 PM ----------



This is a very interesting point. However, in practice I don't think such a miraculous algorithm exist. Could you give a concrete example of any algorithm that enhances information? Or even one that reduces noise while not decreasing information? I think it can probably be mathematically proven that any NR algorithm that does not rely on magic can at the very best preserve the information. And with that I actually mean that the amount of detail/information lost by the very best NR algorithm might be nonzero but negligible. I'd be very happy to be wrong though. I'll try to see if I can come up with some kind of simple proof of this statement.

To slightly formalize this, lets take the definition of noise to be the additional random amplitudes added to the signal due to things such as poissonian shot noise. Things that are perfectly reproducible on each shot (such as the different sensitivities of pixels) are not noise according to this definition as they are perfectly predictable, and should obviously be compensated for.
Search took 0.00 seconds | Showing results 1 to 6 of 6

 
Forum Jump


All times are GMT -7. The time now is 06:38 PM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top