Noise Level

[bill owens]

Could some noise gurus take a look at this shot and tell me if the noise is high, low or about right? Shot with an XSi

at f7.1, 1/1000, ISO 200 (Tamron 17-50 lens at 46mm).

 

Image segment:

<center><img src="http://d6d2h4gfvy8t8.cloudfront.net/7834770-lg.jpg"></center>

 

Original image (compressed).:

<center><img src="http://d6d2h4gfvy8t8.cloudfront.net/7834765-lg.jpg"></center>

[g dan mitchell]

That sure doesn't look like a problem to me at all. You _will_ get some noise on uniform areas like this, and they will be

visible if you inspect them closely at large magnifications. They won't make a bit of difference in a print.

 

In general, noise will be more noticeable in dark areas of the frame, so it goes almost without saying that underexposure

can increase the visibility of noise. It is possible that this shot could have been given a bit more exposure, though I still

think that the "improvement" wouldn't be a significant thing.

 

Dan

[Ken Katz]

Looks OK to me. Given how much of the scene was a high light tone, I probably would have added some exposure to that image, which would also tend to reduce noise.

[martijn_houtman]

<p>

Actually, for iso 200, I think this is actually quite a lot of noise. Is this from a RAW file or an out-of-camera JPEG?

</p>

<p>

Also, what Kenneth says, you might want to try to increase the exposure, just to the point where there are no blown-outs,

that will give you less noise (read up on exposure-to-the-right if you want to know more).

</p>

[andrew_viny]

I've similar concerns about my 30D. Personally I couldn't begin to tell you if this is proper noise or not but I think the go to

rule is to over expose by a 1/3 and bring it back down to avoid noise.

 

~Andrew

[alec_myers]

Can't answer specifically, but the noise you see depends absolutely critically on the raw processor

you're using. Either in-camera to jpg, or in Lightroom, ACR, DPP whatever they all have different noise

reduction algorithms even when NR is dialed back to zero. So it's very very hard to make any kind of

judgement.

[bill owens]

I agree with Martijn and think this is noisy (relative to say, a 20D I had in the past) especially for ISO 200. Enlarging the segment above really shows chroma noise. This was RAW to jpeg with essentially no modification in ACR by me. I have several images of the same subject with better exposure but I picked this one because the whites were already strong (and the noise evident). I took a slew of photos on this Hawaii trip with my just-purchased XSi and the noise seems high in general, but I don't have another camera to compare with objectively.

[g dan mitchell]

If your question is getting at, "Is my camera working correctly?" the answer is almost certainly "yes." (If you are wondering

if there is something about the particular lens that is related to the noise, the answer is, with virtual certainty, "no.")

 

I've looked very closely at many, many photographs of similar subjects (skys, cloudy and cloud free, and similar) and the

noise seen here is not that unusual, especially for the luminosity levels of the area in your crop. The small amount of noise

seen here would be essentially invisible in even a good size print, but if you want to reduce if further you could apply a bit

of chroma and/or luminosity noise reduction. I'd probably do it in the RAW conversion stage, in my case using ACR in CS3

- though you could accomplish more or less the same thing in PS itself.

 

I still tend to think that you could also have diminished the issue a bit with a bit more exposure. The foreground would have

benefited from that as well.

 

Dan

[roger_smith4]

What noise? I can't see a practical purpose where the level of noise in this crop is going to affect your image. Take a deep breath and do a search for a film scan. You'll feel much better after looking at it, I promise.

[bill owens]

G Dan,

 

Well, as a matter of fact, that is my real question.... What level of variance might one see in sensor noise levels in a given DSLR line?

[g dan mitchell]

<p>Are you asking if you would see much noise level variation among different XSi bodies, there really shouldn't be any. If you mean noise

variation between different Rebel series bodies, not much at all - though you might see some differences in side effects of software noise

reduction that would likely be quite subtle.

 

<p>The odds that there is something wrong with your camera are essentially zero, if that is what you are wondering. More likely, what you

see is:

 

<ul>

<li>a completely normal amount of noise. <i>Some</i> noise will be present in all DSLR photographs, and you'll see it if you look closely

enough.

<li>a level of noise that would be invisible in a print even at, say 13 x 19 size and likely larger.

<li>essentially a worst case given that you have probably underexposed a bit and that you are looking at an almost uniform area of slightly

less than average luminosity.

<li>possible to reduce even further by using various post-processing tools, including those that are in the software you likely use with your

photographs.

</ul>

 

<p>We also don't know what noise reduction has or has not been applied to the test image. If you shot it as a jpg, what noise reduction

settings did you use on your camera? If it is a RAW file, what noise reduction did you apply during conversion?

 

<p>By the way, your image is in the wrong color space for posting as a jpg. It is still in AdobeRGB. This creates odd color/brightness shifts

when viewed in some browsers - so what we may think is chroma (color) noise is actually at least partially a result of not placing the file in

sRBG color space before doing the .jpg conversion.

 

<p>Dan

[g dan mitchell]

I wrote: "y the way, your image is in the wrong color space for posting as a jpg. It is still in AdobeRGB. This creates odd

color/brightness shifts when viewed in some browsers - so what we may think is chroma (color) noise is actually at least

partially a result of not placing the file in sRBG color space before doing the .jpg conversion."

 

It wasn't very thoughtful of me to just leave it at that. If you aren't sure how to do the conversion, see if your image

program has something like a "save for the web" setting. That should do it.

 

Take care,

 

Dan

[bill owens]

Thanks to everyone for the feedback.