Noise Levels

[Gerry]

I get a whole lot of noise at ISO 100 in low light (1 hour after sunrise) with my new canon 7d. Is this usual with this camera? I get a lot less noise under same conditions with my old rebel. Wazzup with that? I am very disappointed.

[Jim B.]

Hi Gerry,
I don't have the 7D,but I came from a Rebel to a 50D.What I have noticed with the higher megapixels you need to 'nail" the exposure.Not as forgiving as the Rebels.All the reviews I have read say the 7D handles noise very well.
I'm no expert by a long shot.Maybe post an example with EXIF info intact?There are others with more experience than I and I'm sure you will get some input from them.
You might also want to take a look at some noise reduction programs.DPP noise reduction is okay,but it doesn't give you much control.Neat image,Noise Ninja and Noiseware are nice programs.

[Gerry]

Hi Gerry,

You might also want to take a look at some noise reduction programs.DPP noise reduction is okay,but it doesn't give you much control.Neat image,Noise Ninja and Noiseware are nice programs.

Thx, Jim; I have Dfine as a noise reduction plug in, but I shoot for stock so can't use it for those submissions. I will post a sample of what I'm getting in my images. Thx.

[Gerry]

Is this the best sharpness I can expect with this camera? The front towers look soft and in other files with lower light, shot a few minutes earlier, they are softer yet. The cables nearest overhead cables are about 60' from camera and the nearest towers are about 110'. Does low light affect sharpness?




In the boat pic, the closest boats are about 80' and I focused at about 20'. I shot in raw, edited in ACR & CS3. I had to post crops since the files were too big. Sorry.



Also, noise is a problem, especially in transmission tower pic sky & in lower portion with lower light. Thx. for help.

[Jim B.]

Hi Gerry,

I took a look at the 1st photo.Histogram looked good.I think the softness is due to the noise.I ran it through Noiseware at really low settings.The towers sharpened up nicely.I'm a novice so please explain "stock".Are they photos that are to be sold?Why is there restrictions on cleaning up noise?What post processing are you allowed to do?
Nice shots BTW.

[Gerry]

Thx. for looking at this, Jim. Google "Stock Photography" and that whole world will open up for you! Yes, the are images for sale to the advertising & marketing industry, etc. Noise reduction and sharpening are big no-nos along with anything in post to any extreme such as over-saturation, contrast, etc. Noise reduction leaves details blurred at 100% and sharpening leaves halos on edges. It is very subtle and takes some time to develop an eye for this. Do you think I can expect this amount of noise usually with the 7d? I even notice noise in the full light shots when viewed at 100%. Funny that the Rebel never gave me these problems. Same sensor size and, I would presume, same exact sensor construction.

[Jim B.]

Speaking from my point of view having the 50D I would say,yes,that's probably the performance you're going to get with the 7D.If you are doing mostly shots of the type you posted I suggest you take a look at the 5DII.I own one and it handles noise extremely well.Thanks for the explaination on stock photography.
Remember the Rebels are 10MP and the 7D is 18MP? There were complaints when the 50D came out that the added MPs effected the noise performance.I noticed a differnce coming from a 40D to the 50D.40D had cleaner results.I have to really pay attention to exposure with the 50D.

[Dave Humphries]

~ Noise reduction leaves details blurred at 100% and sharpening leaves halos on edges. ~

Hi Gerry,

I'm not suggesting you break the rules, but those sound like the stock shop's quoted rules, I don't believe either statement is true if you know what you are doing and can invest the time to do the NR and sharpening properly.

I see it as their way to stop every Tom, Richard and Harriet with a digital camera (typically a P&S) thinking they are David Bailey (oops, showing my age now) and whacking the pics through say, Picasa's sharpening without bothering to even look at 100% and expecting it to be bought.

However, I fear that anyone that does know what they are doing probably couldn't make stock photography pay for itself if they took the necessary time in PP. That said, the use the images get put to should mean that the PP is taken care of later in production.

Jim is spot on about needing to get the exposure right, but we'd need to see the histogram of the shot before you have done any exposure or levels tweaks in ACR and/or CS3. Obviously any raising of levels will bring up noise, so if you see that an "Auto" in ACR is consistently raisng the exposure, beware and look at a better metering option for the subject or using EC when prudent.

I haven't yet completely got my head around why noise seems to be so visible in shots at even low ISO and usually in the sky, which isn't where you'd expect it to be. I see it in my DSLR shots too.

Cheers,

[PeterL]

Hi Gerry
For what it's worth I have been closely examining the 7d for a while now as I am trying to decide whether to go for that or the Nikon D300s. From what I've seen of some reviews the softness of the images is normal for that camera. There is one review that I have read that compares the 7d to the rebel and concludes that the rebel is sharper (google "darwin canon 7d") but opinions are varied - I have read many very positive reviews of the 7d also and the positive reviews greatly outweigh the negative so you have to take each with a pinch of salt. The general consensus that I have read for the 7d is that the raw is sharper than the jpeg (same is more true for the D300 as it tends to create much softer jpegs) but it does benefit from sharpening as the images can be soft.
There is also some suggestion that the much higher MP count gives a greater diffraction limit problem so the aperture that you use plays a greater role in sharpness - I haven't checked the apertures that you used for the example shots you posted but it might be worth calculating your diffraction limit for that camera (CiC tutorial on diffraction here https://www.cambridgeincolour.com/tu...hotography.htm)

Pete

[Colin Southern]

Is this the best sharpness I can expect with this camera?

Hi Gerry,

What's your sharpening workflow?

[Colin Southern]

sharpening leaves halos on edges.

That's going to depend on your sharpening parameters. The issues I'm seeing in your images are more to do with high-frequency components; it looks to me as though you're not applying effective capture sharpening (something in the region of 300%, 0.3 pixels, 0 threshold). Correct capture sharpening won't leave halos.

[Gerry]

Here are same images pre-ACR editing:


Third image is same shot with Rebel, much less noise, more sharpness.


Perhaps I need to study how to sharpen as you say, "properly". However, in my experience most of my stock colleagues tell me that any sharpening will get you a rejection. They always seem to see it when I try to sneak it in. Can anyone suggest a good source for subtle techniques on sharpening? I have studied the one here. It is a bit of a learning curve for me. I never remember what the three buttons do. I have tried Colin's version and love the effect but can't seem to get it through the inspectors.

Also, I have my camera set to completely neutral with no camera processing. I like to do all in post with ACR & CS3.

Peter: "Calculating my diffraction Limit"? I have to study up on that a bit. I use calculators to set my exposure/apeture, etc. I may have gotten that wrong but I don't think so according to calculators.

I would love to be able to sharpen and NR on my stock shots and get by with it. I use Dfine for NR but don't get it by reviewers either. Maybe I'm being to heavy handed with it. Any good suggestions?

I chose the 7d mostly because I didn't want to move up to a full sensor. I like the longer range with my lenses. I considered the 5d mKII but declined for that very reason. Is that more successful re; noise levels at low light & sharpness?

Any further comments are appreciated. Thx. all.

[PeterL]

Gerry

Regarding the diffraction limit - I'm probably going to get shot down in flames for over simplifying this (mainly because I haven't yet taken the time to fully read up on it myself) but with a very high MP sensor you will have a larger minimum aperture before the image is softened due to diffraction - I'm sure someone else here can correct me if I'm wrong and I'm also sure I've just made a huge generalisation but that's my rough understanding of the issue. It is certainly something you'll see discussed in the comments section of the review I mentioned in my previous post. I guess you could use the calculator on the CiC tutorial and roughly work out your minimum aperture and see if staying above that helps any.

Pete

[Dave Humphries]

Hi Gerry,

Can you tell me what numbers you have on the six sliders on the Details tab of ACR please?

Here's my attempt at #1, obviously rather restricted by working from the jpg, done with Elements 6.

Neat Image then USM Amount = 200%, Radius = 0.3px, Threshold = 5
Could have been a tad more agressive with these if working from RAW.

Cheers,

[Gerry]

Dave: 25, 1.0, 2.5, 0, 0, 25: I usually avoid this tab altogether and leave defaults in place. Any Sharpening I usually do in CS3/USM. It certainly looks better and I can't see any artifacting from the process but that doesn't mean the inspectors won't. Altho better, it just doesn't seem to be a good as I expected from the 7d and, in fact, not a good as the Rebel. I am going to see if I get some files through with this process. If so, problem solved, I guess. I may still be wishing for the 5d.

I guess you could use the calculator on the CiC tutorial and roughly work out your minimum aperture and see if staying above that helps any.

Pete: By "above that" do you mean larger or smaller apeture, bigger or smaller #? I know, that sounds a bit like learning to tell time...."When the big hand is on the ...."

[Colin Southern]

Perhaps I need to study how to sharpen as you say, "properly". However, in my experience most of my stock colleagues tell me that any sharpening will get you a rejection. They always seem to see it when I try to sneak it in. Can anyone suggest a good source for subtle techniques on sharpening? I have studied the one here. It is a bit of a learning curve for me. I never remember what the three buttons do. I have tried Colin's version and love the effect but can't seem to get it through the inspectors.

Hi Gerry,

There's a few potential issues going on here ...

First up - the definitive reference on sharpening (bar none) is ... Real World Image Sharpening 2nd Edition.

Second up - if you're shooting RAW and using ACR then it doesn't make any difference what picture style your camera is set to as everything apart from white balance is ignored (the easy proof of that is to shoot a frame with the camera set to "black and white" ... it'll be back to colour in ACR. Picture styles ARE however applied to the JPEG preview/thumbnail that's generated in-camera (even when shooting RAW).

ALL RAW images require capture sharpening to counter the softening of the digitisation / demosaicing / anti-aliasing processes; usually 300% / 0.3 / 0 (for low ISO images, with the threshold increasing up to about 6 or 7 for high ISO images) is all that's required. You shouldn't get an image rejected for capture sharpening as it doesn't produce halos and is a normal and necessary part of EVERY digital capture (NO exceptions). Best way to see the effect it's having is to zoom to 100% - dial in the number into a USM - and then toggle preview on/off to see the effect. Content/creative sharpening and output sharpening are different matters though - and the numbers used there WILL depend on many factors such as noise levels / the frequencies in the shot / the shooters intent / resolution / output device etc etc etc, which is why the stock photos don't want this phase done, as it has to be REDONE by the end-user with associated image damage (eg sharpening halos that compliments a high resolution image that's displayed at a small size may look hideous if the same image is printed or displayed large). But capture sharpening is still OK. The thing with capture sharpening is though, you can only see it at high magnifications of around 100% - or - of course when looking at 100% crops like (I assume) you're posting here. So personally, I think that that's a big part of your problem.

The reason the Stock folks say "no sharpening" instead of "only correct capture sharpening" is that 99 out of every 100 photographers don't understand capture sharpening, and either do it incorrectly, or not at all - with either way producing a sub-optimal image.

The other variable that comes to mind is the fact that you may also have a small camera/lens combination focus issue between your two cameras (with the Rebel perhaps getting slightly better focus); all in all though, the images you posted look more like a capture sharpening issue to me.

Hope this helps

PS: You've probably mentioned this already (sorry), but what lens are you using?

[Dave Humphries]

Hi Gerry,

After I posted the above version, I switched the layer on and off in Elements to compare to the original and the effect is even more subtle than I intended , so I'm not sure you'll see much difference comparing by scrolling up and down the page. The trouble was, everytime I increased sharpening for more effect, it started leaving nasty 'tell-tales' probably due to the jpg original.

Thanks for the info from ACR.

I completely agree that sharpening is best done after ACR, in CS, no discussion needed on that.

You might want to try these; 25, 1.0, 2.5, 0, 50, 75.
What this is doing is using some of the NR built into ACR, now this does work very well and my defaults for the last two are actually 100,100. but I wouldn't suggest you go that far under the circumstances; just add 50 to what you have now.


I think what Peter means is that by gross simplification, you might discover that your camera shouldn't be used at smaller apertures (or higher f-numbers), than say, f11; so no more f16, f22, etc. for you!

That said, the effect is minimal and in 'normal' image use, is easily overcome by effective sharpening, but if you cannot do that, then maybe the alledged 'diffraction limit' is more real for you.

EDIT: I agree with Colin above, and the figures I was trying to use were the ones for capture sharpening, just couldn't go the 'whole hog' due to jpg artefacts.

Cheers,

[Gerry]

PS: You've probably mentioned this already (sorry), but what lens are you using?

f/4.0L 17-40. I need to re-read all of both yours and Dave's posts to absorb & reply. Re the sharpness issue, I am wondering if I calculated the exposure settings correctly. Did you look at the metadata? Thx.

[Dave Humphries]

I am wondering if I calculated the exposure settings correctly. Did you look at the metadata?

Hi Gerry,

I hadn't, but I have now - it looks like you are using ACR 4.6, I would recommend getting 5.6 from Adobe.

If I am interpretting the EXIF correctly, it doesn't look like there was any exposure corrcetion applied in ACR. I don't know what effect that brightness figure will have though.

# crs:RawFileName = "_MG_2980.dng"
# crs:Version = "4.6"
# crs:WhiteBalance = "As Shot"
# crs:Temperature = "6100"
# crs:Tint = "+8"
# crs:Exposure = "0.00"
# crs:Shadows = "5"
# crs:Brightness = "+50"
# crs:Contrast = "+25"
# crs:Saturation = "0"
# crs:Sharpness = "25"
# crs:LuminanceSmoothing = "0"
# crs:ColorNoiseReduction = "25"

Cheers,

[Gerry]

I would recommend getting 5.6 from Adobe.

5.6 is not compatible with CS3. I was waiting to upgrade to CS5. I was wondering if you thought I calculated aperture correctly at f/4.5. Wires above camera were about 60' and first tower was @ about 120'. Focus was on 1st tower at the top. Sometimes with these foreshortened views the calculation can be tricky and knock the whole frame OOF. Thx.