Why do my photos look like this?

[Senor Hound]

I went and googled a bit to refresh my understanding of this problem.

http://www.photoxels.com/tutorial_noise.html

^^ has a couple loopy comments, but seems overall pretty accurate and complete.

Thank you. This link helped me out a lot, and rereading the posts after reading this article helped out even more. Its just confusing to me how noise lessens when a photo gets smaller. It seems like it should still be just as prevalent. Oh well, I guess it doesn't always matter why as long as you understand it does.

 

[manaheim]

Thank you. This link helped me out a lot, and rereading the posts after reading this article helped out even more. Its just confusing to me how noise lessens when a photo gets smaller. It seems like it should still be just as prevalent. Oh well, I guess it doesn't always matter why as long as you understand it does.

The noise lessening when you shrink an image is another thing. When you shrink an image the algorythm has to look at the pixels around it and average them out to determine what to display in the final. 1 pink pixel and 20 gray ones=1 gray one.

Sorry, I should have said that before, but I was too busy re-reading on why noise happens.

BTW, if you search for it... someone about a month back posted an amazing analysis on noise. Everything that was mentioned in that article only he/she went even deeper. I recall the person was in school studying this. Actually, I think it might have been GarbZ.

 

[sultan]

What is this an image of? What camera did you use? This is definitely far from normal.

LaFotos images look great even at 100% but yours looks like total junk. Uneven lighting and noise that looks like ISO 3200.

 

[astrostu]

I guess it would be color noise. I just don't understand how one pixel can register a certain color and the pixel right next to it seemingly register a different color when the background is the same. The background doesn't look smooth and uniform, it looks like there's lots of variations in it. Some of the pixels look redder than others and whatnot. You know? And even though it is low-key, its at ISO 100! I wouldn't think this would cause issues at such a slow speed (but its there regardless of ISO).

Basically the problem is that pixels aren't perfect. Some are more sensitive than others to different colors of light. And, unless you have a very specific type of detector, the pixels are generally arranged in groups of 4 forming one "macro pixel:"

R G
G B​

So one out of the four records red, one records blue, and two record green. It's the camera's processor that then interpolates and figures out what the "real" color at each location is. This is also part of why if you take a photo of the moon and look at the R, G, and B channels individually, G usually looks much better than the other two (2x as much information).

So that's part of how you get color noise. The other part is the first thing I said, that pixels are not perfect. Some are more sensitive than others, etc.

You also have a degree of uncertainty governed by Poisson statistics. If 10 photons hit a pixel, the uncertainty in that count is SQRT(10) ≈ 3. So two pixels right next to each other that receive the exact same amount of light may record very different intensities - one may record 7 and one may record 13, almost twice as bright. If you record more light like 10000 photons, then the uncertainty is SQRT(10000) = 100, so those two pixels may record 9900 and 10100, which is only a 2% difference.

It doesn't matter what the ISO is for this - you're still going to get these processes occurring ...

Unless you had the EXACT SAME amount of light falling on each of the individual "pixels" of the image, then yes its not going to be "uniform".

... hence why this statement is not correct.

 

[ANDS!]

... hence why this statement is not correct.

WTF. The only way a person is going to have uniformity of color (for whatever reason this is desired) is if each segment of that picture records the exact same light value. How the hell is that not correct.

 

[astrostu]

WTF. The only way a person is going to have uniformity of color (for whatever reason this is desired) is if each segment of that picture records the exact same light value. How the hell is that not correct.

You stated, and I quote:

Unless you had the EXACT SAME amount of light falling on each of the individual "pixels" of the image, then yes its not going to be "uniform".

So your premise is that you can get a smooth, uniform image if you have the exact same amount of light falling across the entire frame.

However, as I explained in my post, this is not correct. Pixels have sensitivity differences. And there is a measured uncertainty when the light is recorded. Hence, even if the same amount of light falls on two pixels, you can still easily NOT get the same amount of light recorded. What goes in does not necessarily come out.


Now what you stated in your somewhat emotive response is something VERY (though subtly) different from the quote I took from you originally. Your statement that I quoted was the "coming in" part. Your yelling at me was the "coming out" part. Yes, you will quite obviously get a uniform picture if the same amount of light is recorded. As stated in your yelling at me. But it is quite difficult to get the same amount of light recorded, even if you have the same amount of light that hits the detector.

That's why. Next time try to be a tad more polite.

 

[ANDS!]

So your premise is that you can get a smooth, uniform image if you have the exact same amount of light falling across the entire frame.

Oh for ****s sake. Yes, thats EXACTLY what I meant. One of the worst sins of the internet is feigning inability to understand what a poster is saying in order to make some point. Why in gods name would ANYONE suggest the above? Seriously.

Politeness goes out the window in the face of semantic jackassery of this kind.

 

[astrostu]

Oh for ****s sake. Yes, thats EXACTLY what I meant. One of the worst sins of the internet is feigning inability to understand what a poster is saying in order to make some point. Why in gods name would ANYONE suggest the above? Seriously.

Politeness goes out the window in the face of semantic jackassery of this kind.

And, as I stated, that premise is NOT correct. I'm not trying to be a jerk, this is actually an incredibly important concept. If astronomers did not take this into account, then we wouldn't be spending hours processing our images ... taking this stuff into account.

In my above post where I quoted you twice, you stated two very different things. One is what the camera records, the second being what actually hits the camera. And again, in your latest post, you say that you meant the latter, which as I have tried to explain is not correct. I'm going to leave it at that and not argue further with you. If you choose to not believe me, you are free to so. If you choose to be angry with me because I am pointing out that you have not been consistent with whether you are referring to what's imaged vs. the image, then again you are free to do so.

 

[Senor Hound]

My camera is a tiny-sensored point and shoot. I know that increases noise, as does low light.

Thanks to everyone for their answers. You have helped out a lot, and helped me realize I don't want to make a 3'x2' photo with my 5mp camera unless I enjoy looking at it like this.

 

[sultan]

My camera is a tiny-sensored point and shoot.

Ah, I see your problem

Get a better camera (How 'bout a Nikon D3 ).

 

[manaheim]

Ah, I see your problem

Get a better camera (How 'bout a Nikon D3 ).
__________________
It's not the camera that's the problem, it's YOU!

You, sir, are being contradicted by your sig!

 

[dadCameraGuy]

See?
It's normal.

Well, i probably should keep my mouth shut, but it doesn't have to be "normal"

Quality lenses and full frame sensors (like in the Canon 5d) provide much more clarity. I started with the digital rebel and the kit lens, wasn't happy pixel peeping (100%)... moved up to the xti and a tamron 28-75... i was a bit happier. Just a few weeks ago, bought the 5d and an L lens... I'm not getting "normal" any more. I'm getting crisp shots at 100% and i'm much happier.

 

[sultan]

You, sir, are being contradicted by your sig!

Lol, just shoot at a lower ISO.

Just a bit of background on my stance on the issue of better equipment:
You need the right kind of equipment (don't try to shoot birds with the 18-55) but a sharper lens or more megapixels won't do a thing towards making a crappy image better. Most people on this forum have an SLR of some sort and nobody needs any more than an entry-level SLR and an appropriate sub $250 lens. However, fancier equipment can be more convenient for the pros and it's fun to show off

 

[manaheim]

^^^ I know, I was just razzin' ya'. Couldn't pass up the opportunity. <chuckle>