Canon 7D noise at high ISO

[tonycarlson]

I seem to be experiencing a lot of noise from ISO 800 and up. I have had the camera a little more than 2 years with no

real problems until recently. Any suggestions? Could it be:

1. Firmware issue? (haven't updated, Canon's instructions are confusing, but would doing so fix it?)

2. Damaged sensor?

3. Normal? (I know noise comes with high ISO but certainly not this bad)

Thanks

[wgpinc]

<p>Try exposing a little more to the right side of the histogram. Underexposure is the biggest cause of digital noise. Instead of letting the auto pilot fly the plane learn to use exposure compensation to expose correctly. And stop pixel peeping aka viewing your images at 100%, instead look at them at the magnification that they would normally be printed. Do learn how to update your firmware. n It isn't that hard. Most likely your camera and it's sensor are fine. Better to work on and improve your technique. JMNHO Good luck!</p>

[JDMvW]

<p>I'd personally guess that you started pixel-peeping just lately.</p>

<p>Pixel-peeping is the largest single cause of excessive noise known to photography, much more than even underexposure.</p>

<p>Of course, it is always possible that something really is wrong. Can you show us some 100% clips from the old days and just recently to verify that there is a difference in the images, as opposed to the perception ?</p>

[JDMvW]

<p>e.g., http://www.photo.net/learn/optics/pixel_peep/pixel_peep_part1.html </p>

[paddler4]

<p>Can't say what is normal without seeing the image and ideally the histogram, but if your firmware hasn't changed, it can't be that (you can't explain a variable by reference to a constant), and I have never heard of an increase in overall noise from sensor damage. I'm guessing one of the explanations that Gil and JDM suggested: underexposure or pixel peeping.</p>

[User_1172872]

<p>If you are shooting JPEG, check the camera menu for 'high iso noise reduction' or something similar; you may have changed the setting by accident. If you shoot raw, check the noise reduction settings.</p>

[dcstep]

<p>Welcome to the real world of the 7D. ISO 800 is the practical limit if you're going to crop, like for much bird photography. Follow Gil's advice and ETTR. Exposed properly, the 7D can do pretty darn good up to ISO 6400. Underexpose and the noise gets nasty, even without pixel-peeping. Here's one at 6400:</p>

<p><a title="Michelle checks for text message... by David Stephens, on Flickr" href=" src="https://farm8.staticflickr.com/7170/6582141799_dc9f11e821_b.jpg" alt="Michelle checks for text message..." width="800" height="533" /></a></p>

[robert100]

I've never been supremely impressed by the iso range of my 7d.... my experience has been dark colors show substantial noise at anything above iso 1250 out of my 7d.......if you expose in raw (i always do, and always in "faithful" setting, with no in-camera application of anything) and process in DPP, be careful about applying 'sharpening' in the rgb section of the tool palette..........especially if you do so 'in addition to' applying sharpening using the unsharp mask in the raw tools section...... .....stick to 4 max sharpening in unsharp mask, and keep your threshold in the unsharp mask set at level 6 or 7.......you'll be able to get decently clean edges without blowing apart internal areas......

 

however, having said that, you also may want to consider that part of it may be the specific lens you're using

and what f-stop you're using it at......ie i primarily use, and like very much, a tamron 70-300 on my 7d.....i hike reasonably long distances and often up steep mountain slopes so the light weight is a great boon........but at 300 mm reach at 5.6, i imagine that lens, although still producing sharply focused images, is likely partially responsible for noise in shadows in the images.....

 

David Steven's opinions on it, and re how much may be the body -vs- how much is lens aspect will be the most valuable

because he has both a 7d and 5diii and an array of glass.

 

(if nikon had a decent buffer on one of their mid-affordable bodies which could handle repeated bursts of 5-pause 6-pause, 2-pause- 7-pause, etc like the 7d can instead of bogging down and playing like it's permanently dead after 5 or 6 images, the cleanliness of their 24 meg sensors would wipe the 7d off the planet.......but, they don't have. That is the one thing which makes the 7D the camera to carry, it's not the 8fps "burst rate", I also have a 40D and 6-6.5 fps is liveable. what makes the 7D a 7D is that the buffer can handle the quick bursts of "noisy images" all day long. The noise is annoying, annoying enough that i could very easily switch....but that's why I live with it, the 7D's buffer capacity and empty speed.....)

[tonycarlson]

Thanks everyone. Here's some thoughts for me to consider after reading your helpful replies:

1. I am going to try and remember to use exposure compensation (this is one tool I always forget about -

shame on me)

2. I may be over sharpening, so I will experiment with that

3. It may be the lens. I notice it most when I use my 18-55 kit lens. My favorite lens is the 100mm Macro and I

don't notice it as much.

Tony

[dcstep]

<p>I really think it's the body and not the glass. I shoot a lot at 1,000mm and the bokeh is really strong and there's no mistaking noise for OOF bokeh.</p>

<p>Still, I'm much more bothered by chrominance than luminescence. Luminescence look much like film grain and adds no color. I will keep my noise correction for luminescence low, but I'll raise chormainance correction to get of any red and blue pixels.</p>

<p>Rather than Sharpen, I'll often apply Micro Contrast (that's a DxO term, in Lightroom it's called Clarity) to bring up low level details (like feather and fur details) and keep sharpening and NR as low as I can tolerate.</p>

<p>I do all this at 100% and even 200%. Despite the objections of others, an image that looks great at 200% will blow you away when printed or full-screen on any monitor. If I'm not entirely happy at 200%, I take a look at full-screen. If that's okay, then I might share it on Flickr, particularly when content is "special". I only print a small percentage of my images, but I'm seeking results that can stand being blown up to 50" on the long side.</p>

<p>Can't wait to get my pre-ordered 7D MkII. I've seen some high-ISO results that look very encouraging. </p>

[robert100]

you've just given me something to go test, David. I have to go play in dpp to find out what the difference is

between the contrast tool in the rgb section of the palette vs the contrast tool in the raw section. Also, for

information, for any work an image requires beyond the basics dpp can perform, i have to then convert it to a tif

and take it into an old version of elements 6 i use to do things like clean up distractions by removing tree

branches which my lead the viewer's eye out of frame etc. The older version I have, elements 6, is a superb

piece of software, back then it was still a "user's" set of tools almost as complete as the full p'shop of the time.

Thus, within it, there are numerous ways i can then take any small section of an image i find 'over-sharpened' by

dpp and soften it back just within that small area, or vice-versa, ie i can sharpen only a bird's eye while leaving

the rest of the image more soft and natural appearing. Over-sharpened images start to make animals and birds

look like inanimate cardboard cutouts isolated from the landscape. Anyway, I'll have to go play with the contrast

slider in dpp's rgb section to see if that may be an equivalent "micro contrast" or "clarity" tool. Thanks.

[dcstep]

<p>Robert, my approach is similar to yours, doing Raw conversion and global adjustments in DxO Optics Pro 9.5.1 vs. DPP. Are you using the new v. 4.? of DPP? I found it much more user friendly than the older versions. It hasn't displaced DxO in my workflow, but I know that with my early-issue 7D2, that it'll be my only choice for a few weeks, until DxO comes up with their update for the body. </p>

<p>For local adjustments I go into PS 4. Let us know if you find an equivalent to Micro Contrast or Clarity in DPP. I use it on almost every feathery or furry thing that I photograph. ;-) Like all adjustments, you can't get carried away or the results start looking "unreal." </p>

[robert100]

JDM. The link you provided to the old Bob Atkins article is nice, but, it is "part 1", which challenges you, then

promises to inform you of more in "part 2"......................and now not only am i going nuts because i pixel peep,

i'm also going squirrely and itchy because i haven't been able to track down the promised "part 2"......

[robert100]

David, I just pulled the discs out of the storage box to check. I have three eos digital solutions discs, the discs

are id'd as ver 14.2, ver 15.1, and ver 24.1, each of which would have come with one of the three eos rigs i

own, a 40d, the 7d, and a t3i. The T3i disc would be the most recent of the three. If i recall, when i bought the

t3i i didn't bother to load its disc and then in fact discovered that the version of dpp i was using didn't

recognize the camera, so i did load it. I also recall I later went to the Canon USA site and downloaded the

upgrades to that. I think what I have is very up to date. I'll check the canon site. Thanks for the tip.

 

Disappointingly, my old elements 6 can't even be upgraded to work with cr2 files whic h is why i have to

convert to tif....and it only handles 8 bit tifs at that, not 16 bit.

 

C'est la vie, I couldn't go with the new "cloud" crap of pshop, i spend 90 percent of my time off line and in

locales where i can't even go on line, so i have to have pure stand alone stuff. That is one thing which drives

me nuts about the entire industry, is they think everyone is walking around tethered magically to their belly

button cloud. I'm not. I can go weeks where I don't even have access to a wifi, I have to have stand alone

technology.

 

The thing I really appreciate about dpp incidentally, is that you're not actually changing the initial image with

any of the work you do. The original image always remains on file in fact exactly as photographed, all dpp

does is create a file with a "recipe" in it which it applies when you pull up the image file. Thus you can play

and change till the cows come home and all you have to do is punch "revert to "as shot"" and your

experimental changes are dust in the wind. This is not the case in elements or pshop, in them, you make a

change and "save" it, and once you've gone one step past that you no longer have the "undo" available in

edit........your prior image is poof, gone. The amazing thing for people who work on jpegs, is that by applying

only a separately maintained 'recipe' file, dpp allows them to work on pegs without the dreaded degradation

piled on degradation.

 

The truly foolish thing is I actually ignored dpp for years because i hadn't invested the time to learn it was a

decent tool. For years, I was taking the canon cr2 files, which elements can't use, and going through a

horrendously time consuming process of converting them to "dng" files which it can work with, then working

on them in elements, further converting them to tifs where i could layer my "black background/name" to them,

and then downsizing and converting to jpegs to upload to photo.net for my ImagerPro site to access. Massive

amounts of work per-image. Learning what dpp can do has reduced my workload substantially.

[dcstep]

<p>It sounds like you haven't tried DPP v. 4.0. You can upgrade your existing DPP to 4.0. I think you'll love the new interface. In the last version or so they've added Digital Lens Optimization which makes geometric corrections, CA, vignetting, etc. for almost every Canon lens/body combination at every focal length and every aperture. DLO is one of the original reasons that I moved to DxO, but it's now there for free on DPP, at least for Canon lenses. (DxO covers many non-Canon lenses). </p>

<p>To find v. 4, just go to the Canon USA site and search the software update section.</p>

[robert100]

thanks David, actually i have that on the tool palette i have, but the Tamron SP 70-300 which is pretty much the

permanent fixture on the 7D isn't covered.

 

Tony, while I've primarily been watching a football game this afternoon i've also been toying on the side with an

image from a couple of weeks ago off the 7d and that lens, and i've actually been surprised, using a small

addition of that contrast tool in the rgb section i've been able to produce a pretty acceptable shadow area in it, eliminating most of the chrominance noise in the darks, and it's an image i exposed at 2000 iso specifically to test settings. So this may be the comparable tool to lightrooms 'contrast / clarity' tool David mentioned

 

However, you also mentioned you were having particular problem with the noise from your canon lens which

was the kit with the 7d, so the tool David spoke of within dpp which is the "lens" tool on the palette may be of

direct help for you. I know I'm glad you started this thread because it has led me to do some more playing and I'll

be benefiting from it.

[dcstep]

<p>Tony may be seeing chromatic aberration. It can look a lot like chrominance noise and it's not uncommon with kit lenses with lots of zoom range. Digital Lens Optimization might indeed control it or, at least, reduce it.</p>

[robert100]

Tony, if you're still picking up notification on this thread, thanks to the tip from Gil Pruitt (seconded by David S

on here) and some links I read thanks to tips from Mac Hordam on a different thread, I've used a few moments

free time over the last couple of days to play with my 7D, (which as i stated, i've never been happy with above

iso1250 as a practical limit) and their tip of "under exposure is the main culprit in producing digital noise in

images" definitely holds water. Taking a quick look back through a significant batch of images, it really was

evident that the noise I was experiencing which led me to start confining exposures with the 7D to under

iso1250, was pretty much in images I had exposed on "the darker side" of things. (i really like wildlife/bird/nature images with good strong darks in them, and working a tad on the minus-ev-side is a thing which ingrained itself in my brain as second-nature way back in the days of working with 4x5 chrome)

 

So, on finding that, I dragged that camera along on my walk in the woods yesterday with the iso set to 6400, aperture priority to f8, compensation to plus 1/3 ev, and worked specifically at exposing some images "to the lighter" side.

 

I didn't come across any wildlife or birds in the right situations to give me anything which will ever be a

publishable image, but I did get a half dozen of a squirrel in lighting conditions I would have previously just not

bothered "wasting a frame on" at iso1250 and spot-metered to be "centered" on a ttl-exposure reading

because my experience has been that any crop out of it would have too much chrominance noise in the fur to be used. (animal fur and bird feathers are the supreme noise-challenge subject, because they're always light-dark intermingled patches and soft-intermingled-edges)

 

And Voila......playing around with a couple of those exposed experimentally at iso6400 and even only 1/3 plus

ev, then bringing the shadow depth back in during post, gave me cleaner images than what I would have had if

I'd exposed at iso1250 and the needle "centered" on the TTL spot-meter reading. Images which substantiated beyond doubt to

me, that the 7D will in fact give me usable images up to iso6400 just by adhering to that advice Gil gave in his,

the first response posted, to your question: (paraphrased)......open things up and let the sensor acquire enough photon data to climb above its inherent noise level.

 

I'm gald you asked the question initially, and glad Gil and David were here to post an answer. My thanks to all

of you.

 

(the key to photography is the same as the key to everything else in life. No matter how old ya are, no matter

how long you've been doin' somethin', and no matter what level ya been doin' it at.......get out of bed every

single morning with an open mind, and the firm belief you'll learn something new that day.......and, you will.)

[dcstep]

<p>That's the idea Robert, but you can go even stronger with +EV in the situation that you describe. I'd be at +1EV and maybe even more. It seems crazy to raise the ISO so that you can get the higher EV, but it works.</p>

[robert100]

ya, i actually hammered off a couple of iso6400 exposures of one of those cardboard grease-tubes for a

grease gun which was sitting on a wall about 35 feet away, and boosted the exposure to a full plus 1, and then

a full plus 2. I've used it as a 'test sample' other times, specifically because if has a lot of type-setting on it in

various colors and font sizes and wraps around the tube, so i could play with dof tests and sharpness of

lenses.....and the ones at a full plus 2 amazed me with what they give me to play with.....

 

I also fired a couple of exposures of strongly back-lit large maple leaves, with the sun beaming right through

em, virtually blowing them right off the hilite end. I was astounded to discover i could even pull any color and

detail back into them from a full plus 2.......and i found that on those, there's more detail available to pull back

down, with less noise, than there would be if you were coming up from 2 "below".

 

It's simple, it's the same as working with negative back in the film days, the opposite of working with chromes,

"Under-exposing" was a major mistake with negative film, because it meant that there had not been enough

light, enough photon-energy, allowed to hit the particles to register any information on them. Thus, there was

no information there for you to "pull up", it simply never got there. Whereas if you made an "over" error, some

films had a lot more latitude available at the "washed out" looking end of the neg than you'd think. The photon

energy had "been there", it had written the data on the film, and you at least had a shot at recovering it.

 

And I've realized (i'd never given it any thought before) now, that when i'm looking at red and blue dots of

noise in a shadow, what i' actually looking at, is an image of pixel from the sensor which never had enough

light hit it to register an electrical impulse on it greater than the sensor's inherent noise-level. I hadn't clue'd

into that until i read one of the articles Mac linked to, and when i read it, my brain had it's little "aha" moment

and i realized why you gain data by going "to the right" (versus what would be the camera's TTL meter's

opinion.............which is different than taking an incident reading with a flash meter iii)

 

There's a lot of other little specifics which will affect 'when/where/how much", but the key is understanding the

concept of having to get-above-the-floor-noise level. IE,other specifics, such as "just because one camera

model says "this is 100iso", and another camera make/model says "this is 100iso". doesn't mean they are

giving you the "same" "100iso".". An iso "number" given by a camera make/model is a "word term", not a

engineering-specific. This is no different than it was back in film days either incidentally, just because the mfr

rated a film at ASA100 didn't mean it truly was, and in fact, we always bought 120 film in 500 roll quantities

making sure they all came with the same mfr run number (ie dye lot) on them, then we had a specific lighting

set up we test fired 2 rolls under, found out what they "actually" were (anywhere from maybe ASA80 up), and

then stored the entire supply in a fridge so it didn't heat-deteriorate in the 2 or 3 months to use it all.

 

It would be interesting (but i'll never take the time) to test each camera you own, at each of it's iso settings,

and find out what they read actually are at each. ie it's possible that what one model says is iso100, is more

like 85 or 90 or 106.....and what it says is iso3600 is maybe only working at 3490, or maybe it's actually 3812.

Those variations are all very technically possible. And "within a manufacturing tolerance". (IE, have you ever

read Bob Atkins test of the Sigma 150-500 lens, in which he found that it's actually only a 150 to about

450.....and that is "within quotable tolerance".........well, see, other "specs" on gear are also subject to

"variable tolerances")

 

David, I've spent months looking at the images you've posted of birds from your 7D where you showed

settings like iso3200, all showing marvelous detail and pretty noise free, and have been frustrated as h with

my own 7D because I haven't been able to use it above 1250 without getting red-blue-dots-junk.....and now I

find out that the key to it, is your images also have always showed "plus" xamount-ev. It had never registered in my tiny brain before, that there was a difference within my own images in noise, in whether they were exposed at a minus-ev, or plus-ev reference. Now that has been mentioned, I flipped back through some files and can spot the difference within my own image files easily.

 

This whole little exercise has totally given my 7d a new lease on life, because seriously, the only reason i've

worked with it at all through the dis-satisfaction with its noise performance, is that it\s the only 'affordable'

camera out here with a buffer which can keep itself emptied at the rate i need it to. I've stated it many times,

the "fps" isn't the key, I love the 8fps and use it, but I could live with 6.5 as long as it has a huge buffer which

can write itself off to keep up, and nothing else out here does that. (believe me, if a d7000 had a decent buffer I would have switched to Nikon a year ago, I stood with one in my hand in the store for fifteen minutes, right on the edge of buying it and ditching canon, but i could not live with the ridiculously small buffer capacity it has.)

 

This changes everything. Just gaining this knowledge, realizing the noise has been hit-or-miss just because of

my own inattention to this one detail of using it, now totally expands the viability of this camera, and will

increase my enjoyment of using it. (and gee, seeing as how I have well over a hundred thousand images off it, it's kinda nice to suddenly wake up one day and find out it's more useable than i knew)

[dcstep]

<p>Nice post Robert. Keep spreading the word. We're not shooting Kodachrome any more!</p>

[tonycarlson]

Yes Robert, I'm still following. After reading all your responses I am going to try an experiment and hopefully

post some comparisons (straight out of the camera). I'll be using my 7D and 18-55 kit lens. On a side note I

have done some long exposures using a tripod at ISO 100-200 and my kit lens and have gotten great

pictures, but if I find myself doing pictures on the move and have to bump up the ISO to get a faster shutter

speed end of story. Hopefully, I'll have some results for you guys soon.

Tony

[robert100]

Tony, based on your comments, you've been doing exactly as I always had, which was 'boost the iso' - but -

still work at an EV essentially "centering" the exposure. And, i was never happy with the noise. The tip given

here, was that the way to reduce the noise problem was to expose to get a +ev reading with the 7D's in-

camera ttl meter, and that is what reduces the noise factor and lets you work with thr higher iso's. In terms of

"long exposures" ie via slow shutter speeds, although I own a good one, I personally may use a tripod once in

five thousand exposures, and I try to work at 1/1000 of a second or faster. 99 point 99 percent of what I

photograph, is high activity, extremely fast-action. And if you grab a look at David Stephens work, i think you'll

notice that to him, 1/2000 is a "slow" shutter......and he gets usable images out of a 7D at iso6400. And what I

talk about which I've done in the past couple of days, is to understand what he does which I had not been

doing consistently, which is expose "brighter". IE, I took a situation which in past I may have exposed to be

"bang on" by using iso6400 and a shutter speed of, say, 1/1000, instead, leaving the aperture alone, I cut the

shutter to maybe 1/500. Which gave me a +EV exposure...........et voila, when i downloaded it and brought the

ev back "down" in post, it came "back down" and gave me an image from an iso set at 6400, with a whole lot

less noise than by working at "normal" ev.

 

The slowest shutter speed of any of the images I toyed with in my experiment was and which gave me very usable images, was 1/320 at one image, and four or five of them are images I exposed at iso6400, aperture at f8 (on a 300 mm lens), and a shutter speed of 1/4000 of a second. And that's key, because being able to work at 1/2000 and faster is a huge advantage to me in the kind of work i do.

 

If you click on my name i do have samples in the photonet portfolio space (not from this test) but showing the type of work I'm primarily involved in for industrial/commercial clients. Fast moving subjects, in tough, dirty environments which can tear the living crap out of good camera gear. The fine chaff dust which comes off farm equipment gets in everything, it gets in your eyes, ears, nose....and in your lenses, and in every nook and cranny in a camera. Same thing with the concrete dust off heavy industrial equipment. In an afternoon in the field with equipment spewing blasting dust in the air, it's easy to write off a cameras or a lens in an afternoon. Being able to work with higher iso's so I can use very fast shutter speeds, is a huge boon.

 

I don't have any "fire" samples up right now, but i also do a significant amount of work with fire departments, a fire-investigation department, and firefighter/rescue personnel training in which my most "important" piece of gear isn't my camera, it's my Nomex fire-resistant suit. Same thing in all those situations. Fast moving subjects, dangerous environments. And the increased iso is a huge advantage.

 

And then the advantages of high iso simply carries over to the "birds and the bugs" stuff which I do "primarily" just for myself. Fast moving subjects in rapidly changing lighting situations.

 

This, the tips these guys have passed on to us here and which I'm finding will be useful to me, will help you get away from the need to be "stuck" at slow shutter speeds and low iso's in those situations where they're impractical.

[robert100]

David. Re "we're not shooting Kodachrome anymore". One of my three favorite photography stories of all time,

happened because Kodachrome existed. Galen Rowell, a photographer who did a wonderful body of work

over his career, got his "start" in the world of professional photography by a fluke series of events leading to

him getting an arrangement with National Geographic to photograph, "in the first person", an ascent on Half

Dome.

 

And the first photographs he had published in National Geographic, back in the days when all the newspaper

correspondents were walking around with a pair of F3's hanging off their shoulders, were taken on

Kodachrome, on the only camera he found practical to carry in his pocket on the climb............

 

A Kodak Instamatic. (the original "point and shoot")

 

It was his first of several assignments on behalf of National Geographic.

 

Gear's nice. But the gear don't take the photograph. The photographer takes the photograph.

[robert100]

To the attention of Gil Pruitt and David Stephens. Taken a couple of hours ago on this mornings walk in the

frigid cold wind. This little red squirrel is going to be a perfect test subject for testing with because he/she/it

hangs out in the forest around the studio-trailer because the huge maples are a massive food supply, and

likes popping out to see what i'm up to with the camera.

 

Don't bother paying attention to the composition etc, and don't pick at it at the pixel level, I was firing

exposures purely to test how much workable room i'd get to work with, and if I do the thing and go plus 1 ev, can i

get something usable at iso6400. I really haven't done any of the "detailed" things I would do to make an

image "publishable". all this is showing is two minutes of a first-stage-look at the image in dpp.

 

I am amazed. Plain and simple. Amazed. I've never been happy with this rig even at iso1250. But I have

been working with a 30 year old ingrained habit of, if anything, exposing a little on the "under" side.

 

Check out the exif. Plus one, at 6400. Internal camera high-iso-noise-reduction settings in custom function

set at "zero". Disabled. In DPP post, on the rgb slider which you can't see here, I applied the max 20

chrominance reduction, but zero luminance noise reduction. Leaving the white-balance setting at auto as

exposed, instead of playing with color-temp, I instead just adjusted to warm up the shadows with the tone

slider a touch.

 

Check out the "sharpening" slider on the tool palette, which you can see on the screenshot. Amazing. This

image would not even need any of it applied.

 

The advice they're givin' us works, Tony. This just let me go to f11 and a workable shutter speed instead of trying to dig something out of a hole with f5.6 and trying to hand hold a long slow exposure . Cool.<div></div>