Bad noise on 40D - can you help?
darkdragon
Registered Users Posts: 1,051 Major grins
So, NeiL pointed it out an it really has been concerning me a lot lately - my 40D has a noise problem. I think. I've seen a LOT of great photos at ISO up to 1600 from the 40D (not mine) that looks great with low noise levels. Of course it is hard to tell from looking on the web at less than 100% crop if there is low or high noise, unless the noise is CRAZY high. Also some people have posted "my camera is awesome check out this picture" but fail to mention that they ran it through noiseninja or something similar first.
My photos straight from the cam sometimes (mostly) have horrible noise - at all ISO levels. Seems like it must be an issue with my sensor because so many people say how great and low-noise their 40D is.
I'm going to be taking my camera back to the store on Saturday to have them give me a new one because of the noise issues. I can demonstrate the high noise even at ISO 100 on my cam, but what I need help with is an untouched full rez photo from a 40D with low noise - so that i can show the clerks the difference.
Can anyone help me out here?
I just need you to email me the image or the link where I can download it, and I'd need to get the EXIF from you or better yet have it still embedded so i can pull it up in Lightroom at the store.
I promise I will not print/sell/steal/rent/give/etc your image to anybody, just need to use it on my laptop for the one purpose and then I'll delete it.
Thanks for reading.
My photos straight from the cam sometimes (mostly) have horrible noise - at all ISO levels. Seems like it must be an issue with my sensor because so many people say how great and low-noise their 40D is.
I'm going to be taking my camera back to the store on Saturday to have them give me a new one because of the noise issues. I can demonstrate the high noise even at ISO 100 on my cam, but what I need help with is an untouched full rez photo from a 40D with low noise - so that i can show the clerks the difference.
Can anyone help me out here?
I just need you to email me the image or the link where I can download it, and I'd need to get the EXIF from you or better yet have it still embedded so i can pull it up in Lightroom at the store.
I promise I will not print/sell/steal/rent/give/etc your image to anybody, just need to use it on my laptop for the one purpose and then I'll delete it.
Thanks for reading.
~ Lisa
0
Comments
If you underexpose, and have to boost brightness in post processing, you'll see dramatically worsened noise levels, particularly in shadows, could this be it? Looking at your challenge entry, to my eyes the texture NeilL pointed out could just as easily be your lens picking up light reflecting off the weave of the pants. I think this is more a question of diffusing the flash you used than replacing the camera.
On the subject of that shot, I quite like it (bonus for creativity!) and in no way noticed the textured knee until it was pointed out
If you are noticing noise though (remember that there MUST be noise, reality requires it, it's just a matter of degree), the real question to ask is if it's becoming a problem for your photography. On my 40D I try to stay between ISO 100 and 400, but I'll happily go to 3200 if it means getting the shot. ISO 3200 isn't pretty, but at web resolutions, and even moderate prints, it'll still beat the pants off a lot of cameras, and they wouldn't have a chance of taking similar photos
http://robin-taylor.smugmug.com/gallery/4659823_AoswY#275214413_whnGa
This is a link to a gallery of images taken I believe exclusively at ISO 3200. I've enabled view originals (gasp). These were all shot RAW with high ISO noise reduction and run through digital photo professional (DPP) with both chroma and luminance reduction set at max, also you'll notice that they're mostly shadow. This is a good worst case scenario for you to compare with. For the record the technical quality of the shots is not good at all (focus etc), but I was also the only person to come home with any pictures at all.
I agree about the exposure thing mentioned in the above post.
here are a couple of samples at 100% crop with no post processing at all other than exporting from Lightroom and uploading to SM.
1. ISO 100 EXIF
2. ISO 100 EXIF
3. ISO 500 EXIF
It could just be my imagination or being too picky, but they all seem to be overly noisy. I haven't done anything with brighting or fill light here, so I know it cant be that. These are as "out of the camera" as I can get. For example, the 3rd image @ ISO500 is what I was expecting an image to look like at ISO800, maybe my expectations were too high?
Thank you for the nice comments on my photos and for the link. I checked out the images and it is amazing because at web size they look fantastic but when i look at "original" I can see all the noise. Of course that is to be expected at ISO3200 but definately a great eye-opener as far as web vs. reality.
It is really hard to know what the right thing to do is, as far as if my camera is bad or normal. I don't actually know anyone else with a 40D to compare to.
If so, the first one looks ok to me.
The shot of the plane at ISO 500 may have a bit much noise than what I'd expect but not if it was underexposed and then corrected.
Try opening it in Canon's DPP and see as DPP has no default corrections.
#3 is a little tougher to call, I would say it's about in line with what I'm getting. It's a bit noisier than ISO 400, which is normal (it's only 2/3rds of a stop down from 800, right?). Also remember that the underside of the plane in #3 is probably the best possible way to find noise if you're looking for it. Shadows, fairly even tonality, and not enough detail to distract from whatever noise is present. I promise though that if you run it through a noise reduction routine with chroma (colour) noise reduction at max, then sharpen it up a tiny bit, it will look better, at least that's what I'd do. Finally, since I doubt you'd be looking to show this size crop under normal circumstances, any reduction in display size would make the noise invisible.
There was a discussion in this forum not long ago about the "tweener" ISO settings, ISO 125, 160, 250, 320, 500, etc. not being necessarily optimal, with the noise not proportionate to sensitivity (e.g. ISO 160 noisier than 200), but the differences are pretty miniscule up to ISO 800, and after that the noise scales up just as you'd expect anyways.
Some useful comparisons can be found that will let you set your expectations for noise:
http://www.dpreview.com/reviews/canoneos40d/page18.asp
http://www.the-digital-picture.com/Reviews/Canon-EOS-40D-Digital-SLR-Camera-Review.aspx
Check out the "noise patches" on both of these reviews, they give a good benchmark.
dak.smugmug.com
The problems with those two copies were stark and indisputable, and the cameras were replaced. I was not the only person shooting images with #2 copy, a photographer friend did too, and wrote a report which I sent to Canon along with sample and comparison images and the camera itself.
I have done a small amount of shooting with the new copy of the 40D which I received very recently. So far it is chalk and cheese to the #2 copy I had (#1 copy would not accept any lenses, as you know).
While it's sensible, as people here have pointed out, to consider all possible causes of noise and to check your expectations against comparable images from other 40Ds, the purpose of this process must be to ascertain the facts, not to indulge in user error masochism! You have used other cameras, you have a not negligible amount of experience, you therefore have pretty realistic expectations and good assessment skills. Trust your judgment!
DPR has a gallery of 40D sample pics at all ISOs, and the 40D user review thread contains, as I have pointed out before, a subset of negative comments about noise from the 40D, including one from a commercial photography group who gave up on the 40D for this precise reason.
I note that only one of the respondents to this thread of yours has posted any samples of their own to show that user error free images from their cameras do not have a noise problem.
You can find a similar thread to this one of yours opened by myself at the time I was also in the midst of the same process you are now in, and one of the images I posted there was a crop of a dark pair of trousers with very similar noise to what is in your challenge pic. And that pic is unique so far among all the entries for the noise it contains. That says something about your copy of the 40D, I suggest. If your camera is typical, the 40D fails to match the standards of all the other cameras used there, for there are various subjects, levels of exposure, shooting conditions and post processing represented. Not likely I reckon!
Best of luck!
Neil
http://www.behance.net/brosepix
I am going to check that this morning here in a few. normally the only default corrections I see being done on imports is a small 20% sharpening. I don't think that could account for the noise, do you?
Yes, they are a bit underexposed. For some reason my flash doesn't reach that height ;-) Anyhow, even though they are underexposed I didn't mess with them at all and I think that #3 especially looks excessively noisy - not just under the wing but throughout the image and even the sky. It was taken minutes after #1 with nearly the same ambient light.
Robinivich: Thanks again for the links, I will be checking those out today as long as the firewall doesn't block them. I don't know why I didn't think to go look at dpreview, duh!
Maybe I should just stop pixel-peeping my images? Of course it makes it tough when people say my images are noisy even after I think they look great, so if nothing else I definitely want to look into this before my warranty expires
Looking at images at 100% is what all photographers who know what they are doing MUST do, so again why be masochistic about it?
Neil
http://www.behance.net/brosepix
These pictures are all under-exposed. That in itself will generate noise. The right hand 5th of your histogram contains 50% of all your picture data. (look at an excellent explaination here at Luminous Landscape) When you shoot a picture not utilizing that much of the data, "or in your case, even less", your signal to noise ratio is very bad. And, to compound the issue, you are looking at 100% crops from a 1.6 crop camera. I have yet to see a good clear image from a 1.6 displayed @ 100%. My FF 5D will, but not a crop camera. (that I've personally seen)
First pic) Histogram shows that the furturest to the right spike is barely over half-way to the right. Its probably only using about 25% of the data that could have been captured.
Second pic) Same thing
Third pic) Again, same thing
Canon didn't design the camera for images to only use 50% of its image data,"or less", at least not to get optimal results.
Make some test shots that are correctly exposed. Look at those. I think you'll be OK with the results. If there is still an abundance of noise in correctly exposed images, then I would pursue that with Canon.
Hope that helps...
Indeed, among skilled users there is the school of thought that it is sometimes preferable to get maximum detail in underexposure and lift it in PP than to risk loss of detail through overexposure.
What is the point of all those PP application tools related to exposure if the only way to get acceptable images was through correct in camera exposure?
No, I think rather there is a zone of exposure that will produce acceptable noise levels even without PP, and I think the above images are in that zone. I think they should not be so noisy.
Neil
http://www.behance.net/brosepix
Neil,
You sound very defensive
Number one, I was responding to the OP, not to your personal photographic choices.
I don't know what circle of "skilled users" you are in, but from a technical standpoint, if your testing a camera, do so with the design of the camera in mind.
I really can't say I know much about cameras, or their use, although I make my entire living using them.
Of course, one is free to artistically make any exposure decisions that you wish for a given image, but that is completely different than what is "optimal" for the camera to do its best job.
As they say YMMV...
I just want to be clear and say that these are not portfolio images or anything i'd say "yeah, thats a good shot". These images I posted at 100% were taken simply to test autofocus on my lens and make sure the new lens was working right. They were shot on Av mode, so if they are underexposed it is because the camera decided to do that - I did not intentionally over or under expose antyhing just to show bad grain.
I also did not choose the images just to show bad grain, they are the freshest ones I had at the time and they were among very few images I had on the laptop I was using when I posted the thread and the images.
Obviously they are underexposed, obviously if I were to correct that in post I would see a large increase in grain. I'm not a newbie to photography, I do know how to get correct exposure (too bad my camera doesnt do that automatically perfectly every time). Again these were just tests to test a lens, not to test noise or anything like that.
To me, I don't care if my camera has bad grain or normal grain. Either way, I just want the best I can get out of the system. If the problem is not the camera and is user error or even correctable or expected, fine no problem. If the problem is with the camera then I will get it replaced. I just want to know what people think at this point - my problem or a possible camera issue.
Hopefully today after work I can take some more controled images at different ISOs with proper exposure in M and get those up. Then maybe at least we can solve that part of the developing argument.
An interesting fact to note is that they look a lot more grainy in LR on my mac then they do looking at them on the web via PC - I have no idea what that means, just thought I'd mention it.
Thanks again for all the info, everything helps and I am always happy to read links and look at samples from you guys.
I know that more pictures are not shot at the ideal amount to the right than are, yet when you look at all the thousands of pics in this forum, all the others and generally to be viewed anywhere you will not find a great many where noise is the problem it is in the images in this thread. I have seen very acceptable IQ from mobile phone cameras, to take an example, which would be the most laissez faire mode of photography possible. No thought of histograms there!
Your very true point that cameras perform best when they are used in a way that corresponds with how they perform best, supports what I am saying, that is that if a camera ONLY gives acceptable photos in the MOST IDEAL conditions there is something wrong with it.
I think what I meant by skilled (= know what they are doing) photographers is clear and not judgmental or inclusive of myself ( I have always described myself as very nearly skill-less in the practice of photography). So no cause for personal jibes or self-deprecating posturing, I think. It is just a fact that intentional underexposing is practised as a technique to gather the most detail in situations where overexposure is a risk.
And I repeat, there would be no point to common PP tools if underexposed images could not be rendered to an acceptable IQ.
All of the above leads me to believe that there is a zone of exposure in which the typical modern dSLR camera can cope and produce acceptable IQ.
So, please bear with me if I try to find the reality in this issue.
Neil
http://www.behance.net/brosepix
And I direct attention again to a post in the DPR users reviews of the 40D by a commercial photographers group who rejected the 40D outright because of its bad noise performance. So, I think there are more bad copies out there than the two I got, and possibly darkdragon's is one of those?
Neil
http://www.behance.net/brosepix
http://www.digitalgrin.com/showpost.php?p=795244&postcount=57
*http://www.digitalgrin.com/showpost.php?p=795425&postcount=59
http://www.digitalgrin.com/showpost.php?p=795441&postcount=60
another example where users noise issues seemed to go away
http://www.digitalgrin.com/showthread.php?t=90575&highlight=noise
These next two were shot in tough lighting. The first is the original and what I feel was properly exposed. I hurt my eyes looking for unacceptable noise at 100% on my PC.
http://dak.smugmug.com/photos/299687665_M8Z3h-X3.jpg
I bought it home and bumped up the exposure a stop to see some detail under the plane. That introduced some noise/grain.
http://dak.smugmug.com/photos/299683510_5m7CD-X3.jpg
Slight noise difference between them at 100% and more noticable at 200%.
Everything says you don't want blown highlights because you can't rescue them, therefore it is easier to rescue a pic if you under expose, but why expose a shot so you have to rescue it?
Shoot for proper exposure. Get it right in the camera. Of all the seminars I've attended and classes I've taken, I've never heard a Pro photog or Canon Explorer of Light (Fred Marcus, Michel Tcherevkoff, Harry Benson, etc.) say they underexpose on purpose. They shoot for "perfect exposure, everytime" and they bracket if they need to. With digital there is less room for error. They aim to get it right every time.
The fact is that if it were properly exposed, it would have bought out more detail in the shadows as rwells suggested.
As far as the plane shots, #2 for instance, f/13 @ 1/125, 200mm lens, ISO 100
Whatever sharpness you tried to gain by going f/13, you lost by shooting at 1/125 because it's blurry from motion.
An equivalent exposure I would use is f/3.5 @ 1/2000 but since I believe it is underexposed by a stop due to the histogram (I bet that plane is closer to white than that picture shows), I would shoot it closer to f/3.5 @ 1/1000.
After I type this (watching TV and stuff) I see you shot in Av mode and you underestand this probably contributed to the underexposure. I also looked at other images on your site taken with the 40D but since I can't see enlarged versions I can't really comment on noise/grain, although they look fine.
http://darkdragon.smugmug.com/gallery/4815109_V6GvY#286244042_egW6a-L-LB
http://darkdragon.smugmug.com/gallery/3572923_PFZYs#281511796_aS3j2
My main reason for even posting this is not to get you on the defensive, but because I feel that you should probably get a properly exposed image before you go turning in the camera.
I'm far from perfect and I can get noisy images, crap composition, etc. in my own images, plus I don't fully understand my cameras, but I think everyone here is trying to help by correcting what appears to be underexposure causing noise/grain to show on a 40D image.
Yes, I too can underexpose an image and get similar noise to those plane pics.
If I properly expose an image I often get little if any noise.
ISO 1600
http://dak.smugmug.com/photos/299703454_3845N-X3.jpg
Once my 40D batteries finish charging I'll gladly go take some ISO 500, and up images. I stick to 100, 200, 400, 800, 1600 and 3200, but I'll try the other increments out just to see how they turn out.
dak.smugmug.com
Thank you for the links, I'll be checking those out shortly. I definately understand that these images are not great (nor were they inteneded to be as i stated earlier). I'm also trying to not be defensive and I do undertand that a perfectly exposed image will look a lot better - I have many examples of that already.
The only thing I worry about is that if the image is underexposed even slightly, is that necessarily something that should cause a great deal of noise? I have no idea and can only go off what I see other people posting.
As far as the ISO500, that was an accident I wanted ISO400 but finger was movin too fast (I was trying to get some test shots out before the sun went down on tuesday).
I'll definitely get some better test shots tonight, geared to see if there is a noise problem or not. I'll post some crops here. If nothing else maybe this thread can help someone down the road. I did run a couple searches on the forum but didn't turn up much (I'm guessing it was my search terms), before I posted this thread.
One more thing, looking at my open galleries doesn't help a whole lot to judge if my camera has bad noise just because I've run most posted images through LR noise reduction. Especially the underexposed ones ;-) That's why the ones I posted here in this thread i made sure to state that they were untouched as far as adjusting noise and brightness.
Ooh, I am putting a viewpoint only. rwells was pretty strong in the way he put his viewpoint. He is dismissive of anything which doesn't line up with theory as he understands it. I am pointing at what emerges from real people in the real world, including my experience, and responding to darkdragon's doubt that some amount of underexposure could carry such a heavy penalty as seen in these pics. I didn't invent her concerns, she brought them here, and I think her hunch that something is wrong with her camera is not being respected. As I said to her, she is experienced, she knows what to expect. Yet she is being told in effect to go lift her game.
THAT is condescending. Maybe you think it's OK because she is a woman?
"Those planes are underexposed. The histo spikes at the middle for #2." This is you, and that is all. Blunt, wouldn't you say? Critical of the OP who asked advice, wouldn't you say? Where are the social graces here?
Or maybe my style of writing which is careful and error free (usually!) you think implies criticism of your and others' styles - and ideas. Well, I hope you can get over it! Fortunately you don't actually have the ability to order people according to your preferences.
For what it's worth, I DO write carefully because I think carefully. Sometimes I am correct. But my motive in participating in forums is to learn, to be helpful and to have fun (and the occasional controversy!).
In any case, while you are certainly right in what you say about underexposure and noise, you have not said why you categorically believe that the camera is not faulty. This is in the face of the fact that some cameras, lenses etc ARE faulty.
Neil
http://www.behance.net/brosepix
So, I checked out all the links you posted. The one with the pasta retraunt was pretty noisy, even at the web resolution. What bothers me about my image is that it seemss noisy as that one but mine is at 500ISO not 1600.
Maybe it's just my monitor but the first image you posted of yours (in the quote above) seems more underexposed then the ISO500 I posted. Even after you bumped up the exposure, there is no where near the amount of noise I see in mine. Examples like that is why I question if it a problem with my camera.
Also of note, it seems to me that the noise issue on my photos is worse now than it was when I got it. This could be my imagaintion, I don't know. I look at some of the first photos I took with my 40D and I don't see nearly the amount of noise that i'm seeing now. This is in over/under/perfectly exposed shots.
I am very grateful for all the help you guys are offering and I hope my new test shots tonight can help even more for us together to figure out if it really might be a camera issue or just me.
Yes, you are doing people a service by thrashing this matter out. Thanks. It's very important to know if equipment which is so valuable in money and more personal ways is faulty. But the exercise of critical assessment of images which is involved in doing that is a profitable one.
Look forward to seeing your batch of tests.
Neil
http://www.behance.net/brosepix
White or gray?
dak.smugmug.com
gray. actually i'd say it was an 18% gray.
My assumption was that the OP's pic was underexposed of a white plane and I got what I deserved; I made an a__ of myself. If I had been right then the plane was 3-5 stops underexposed. I was wrong there.
When I get home I'll put up some full res jpg's at various ISO's and what I consider to be properly exposed for the OP. I will also post the exact exposure settings and use a 60W bulb for all of the shots.
Test it properly is all that is being suggested. If the camera is bad, the camera is bad. Send it in get it fixed, get it replaced, whatever it takes.
Unfortunately, I can't possibly recreate that picture to help DD because I wasn't there at the same time in the same lighting conditions. As is, I would have shot close to f3.5 @ 1/2000 since the plane is gray. I would gladly share the RAW file w/ DD.
Here is why I categorically believe that you should expose properly with plenty of references to back up:
http://hannemyr.com/photo/zonesystem.html
dak.smugmug.com
hey man, no problem - it happens. I still think it is underexposed but probably not by as much as you originally thought.
Sweet, a new link I haven't read. I'll be printing that off and reading it. I thought the Zone System was a diet
Lookin forward to seeing your test photos tonight. Thanks for the help!
Glad you posted this because I forgot to check.
Just FYI - the images I posted (on page 1) have NO mods to them. I just checked all the lightroom settings and everything is on 0.
Having said all that, if you aren't comfortable wtih the noise level in your camera, you can return it or have it serviced by Canon as well.
This is my first time trying to do some controled test shots, so I'm sure it's not up to some of you all's standards.
All images are shot on AV mode. I tried to do things the way a "normal" user would, not a photography expert - my reasoning is that the camera should perform as expected under normal use. A photo expert would be able to work around the flaws of the equiptment if there were any. I didn't touch the WB at all (camera AWB) and the images have been converted from RAW to JPG with no editing at all.
I hope this helps to solves some of the questions we were all having. I have not looked at all these images at 100% yet, I wanted to get them loaded up here first.
Anyhow, here's the goods, click on the thumbnail for the original size images.
1. ISO 100
2: ISO 200
3: ISO 400
4. ISO 800
5: ISO 1600
6: ISO 3200
I also did a similar set with a much shorter zoom. You can see those in my gallery here: http://darkdragon.smugmug.com/gallery/3979820_S8JT6#299862946_PHZ6Z