Canon 50D noise worse than 400D

[wildlight]

<p>Hi everyone,<br>

<br />I recently purchased a Canon EOS 50D to replace my 400D and it is an improvement over my old body in every way except noise control which is pretty disappointing.<br>

I did a test to show the noise levels from the 50D and 400D at ISO 1600 and the 50D came out worse (I will post the photos soon!).<br>

<br />Am I doing something wrong or does every Canon 50D user have this amount of noise at high ISOs?<br /><br />Thanks,<br>

<br />Joe </p><div></div>

[wildlight]

<p>Here's the 400D photo.</p><div></div>

[derek_hofmann]

<p>Assuming both photos are 100% crops, 700 pixels is 18% of the width of the sensor on the 400D but only 14.7% of the width on the 50D, so basically the first photo magnifies the sensor noise on the 50D compared to the second photo.</p>

[mike_earussi1]

<p>Did you up rez the 400D to match the 50D? If you didn't your comparison will be off.</p>

[craig_meddaugh]

<p>+1 to what Derek says. You're comparing two differently sized images to each other. </p>

<p>I own both cameras and I can assure you that when it comes time to print the images, the 50D is better in terms of both resolution and nosie. </p>

[PuppyDigs]

<p>Print a 13x19 using the same lens, subject and light. The 50D print will be cleaner and sharper because it requires less rezzing up than the 400D. Of course if you enjoy viewing on computer monitors at pixel view, any of the older lower MP models will appear cleaner than the newer massive MP models...not how these images are meant to be viewed.</p>

[hawkman]

<p>Joe,<br />That is correct, 50D has more noise than the 40D and 400D at 100% size i.e. 1:1, you are not doing anything wrong. All the legitimate review sites have indicated this as well, and it is expected due to the small pixel size of the 50D. SW noise reduction is necessary for 50D files and you have to expose to the right, otherwise you end up with too much noise. the 50D sensor was a mistake IMO. Hopefully like the G11 the next camera will have a better sensor.<br /><br /><a href="http://www.dpreview.com/reviews/canoneos50d/page18.asp">http://www.dpreview.com/reviews/canoneos50d/page18.asp</a></p>

[ralph_jensen]

<p>OP, as you can see on that page that Arash linked, dpreview makes the same mistake so many others do, not downsizing the larger-resolution output to match the smaller one (notice the size of the queen's head; it's noticeably larger in the 50D crops than in the 40D's).</p>

<p>See also the posts above by Derek, Mike, Craig, and Puppy.</p>

<p> </p>

[craig_meddaugh]

<p>Arash, did you not read the last several posts? The only way the 50D looks noisier than the 400D and 40D is if you improperly compare samples on a equal-pixel basis instead of on an equal-area basis. That is the same mistake they made in your link above and another good reason to be reasonably skeptical of everything you see on the internet, even the supposedly expert work done by those at dpreview. </p>

[nathangardner]

<p>15MP packed into an APS-C sized sensor = NOISE Sell it and get a 40D. The 10MP on a 40D is plenty and handles noise at high ISO's much better. </p>

[hawkman]

<p>Craig and Ralph you are absolutely wrong, image has to be evaluated at 1:1 size anything else is <strong><em>interpolated</em></strong> and a waste of time and does not indicate sensor performance. also I am a bird photographer, I evaluate noise based on field performance in my own hands, I don't care what people say on forums like this. I buy a higher Mpixel camera to print <strong>LARGER </strong>or crop<strong> MORE</strong> not to down-sample back to the size of the older camera. <br /><br />BTW, the link is not mine it is dpreview.com, Bob Atkins has a similar review here on PN, and my response was to OP, this issues has already been discussed to death here and I am not interested in engaging in yet another wasteful discussion.</p>

[keith reeder]

<p>Yep, I'm with Arash on this one (maybe it's a bird 'tog thing!) - uprezzing or downsampling files to match each other is a complete obfuscation of what's really going on.</p>

<p>Its why I'm still shooting a 40D instead of a 50D: based on my own testing, and already having a <em>very</em> noise handling efficient workflow, if the only way I can get 40D-level noise performance from a 50D is to downsample to 40D-sized files, <em>what's the point of the 50D's extra pixels? </em> I already get as much fine detail from my 40D files as I could possibly want...<em><br /> </em></p>

[derek_hofmann]

<p>The 50D has more, but smaller, pixels of noise than the 400D. In the end, there's no difference between the two in terms of noise. The difference is that in good lighting conditions where noise isn't an issue, the 50D produces a sharper image.<br /> <br /> Do as Puppy Face suggested: have both images printed at 13x19 and compare the two. That's really the ultimate, apples-and-apples test, the only one that really matters.</p>

[craig_meddaugh]

<p>It's not a bird 'tog thing as I respectfully disagree with both Arash and Keith. Arash... you focus on downsampling the 50D to get the same noise performance of the 40D but what happens when you upres the 40D to get the same resolving power as the 50D... oh yeah... noise goes up considerably doesn't it.</p>

<p>Another way of looking at your false conceptualization is that you are comparing grain size on film by using two loupes of different magnification. Of course the grain (noise) looks bigger when you use a higher magnification. We would never quantify film performance in this manner so why on earth would you want to quantify digital performance. Do as puppyface suggests (i.e. compare actual prints) and you will quickly see that your inappropriate methods or conceptualizations do not square with the actual output.</p>

<p> </p>

[hawkman]

<p>Hi Keith,<br />I agree, I miss my old trusty 40D, it's a shame because 50D has very nice AF IMO for in flight shots and a nice LCD + AF fine tune, it can produce nice shots but noise is there and I am fed up with NR. I have also not seen the level of sharpness that I like to see with lenses like 500 f/4L IS and 400 f/5.6L. Any ways, I just wanted to point out that OP's observation is correct.</p>

<p>For those who are interested in how noise should be measured here is a very good article by R. Clark, this is industry standard , regardless of what surfers say on internet forums. It is not a matter for average Joes to agree or disagree. <a href="http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/">http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/</a><br />You cannot compare interploated images, PERIOD.</p>

<p>Good luck</p><div></div>

[mike_hitchen]

<blockquote>

<p>uprezzing or downsampling files to match each other is a complete obfuscation of what's really going on.</p>

</blockquote>

<p>I don't agree. The object of a photo is to get the best possible quality at a given size so the best comparison is create prints <em>at the same size.</em> Use both cameras to create prints at 12x8 , 20x13 , 30x20 print (or whatever). Compare detail and noise - that is the only comparison that matters.<br>

When you convert output of each each camera to jpeg so you can print, each image will have different levels of uprezzing/downrezzing/whatever you want to call it. So why is uprezzing/downressing a false test?</p>

 

[craig_meddaugh]

<p>Arash, you are still missing the point. Yes the 50D has a smaller full well capacity than the 40D and thus the per-pixel shot noise must be higher. However, nobody is really concerned with how noise looks on a per-pixel basis... or main concern as photographers is the final output. So yes, per-pixel noise is a bit higher on the 50D (I don't think anyone is disputing that) but when you take into context that each 50D pixel is a smaller portion of the final image the overall noise is less. </p>

<p>So I guess the question is: What is more important? The pixels or the image? </p>

[ralph_jensen]

<p>What Mike and Craig said. If I want to make, say, a 20x30" print, resizing the files so that they're the same resolution when comparing noise levels is the only way to tell me what I will get.</p>

<p>Sure, inspecting 100% crops on screen will give me the per-<em>pixel </em> noise level, but for real-world use I only care about a per-<em>inch </em> noise level. <em>Unless I'm going to be printing smaller than the lower-resolution camera's native resolution,</em> obtaining that (noise-per-inch) information requires matching <strong>resolutions </strong> when comparing cameras.<br /></p>

[hawkman]

<p>Craig,<br />I have no idea what you are talking about, Image is made of pixels, pixels matter as much as the image does. Extra pixels that contain no information but noise are only a burden on your PC and memory card. 50D has more noise than 40D period, based on physics as well as more than 30,000 field photographs I have taken with this body in all sorts of conditions. Nothing is going to change that. Good luck to you with your photography.<br>

Canon has already realized this and stepped back in Mpixel with their compact cameras, I am sure they will do the same with 60D.</p>

 

[ralph_jensen]

<blockquote>

<p>"50D has more noise than 40D period, based on physics as well as more than 30,000 field photographs I have taken with this body in all sorts of conditions. Nothing is going to change that."</p>

</blockquote>

<p>Maybe you'll believe DxOmark? They say the 40D and 50D are basically identical in low-light ISO:</p>

<p>http://www.dxomark.com/index.php/eng/Image-Quality-Database/Canon</p>

<p>(Maybe you got a bad copy of the 50D?)</p>

<p> </p>

<blockquote>

<p>"Canon has already realized this and stepped back in Mpixel with their compact cameras, I am sure they will do the same with 60D."</p>

</blockquote>

<p>On the contrary: their newest camera looks to be packing 20% more pixels into that APS-C sensor (18mp!):</p>

<p>http://www.photo.net/canon-eos-digital-camera-forum/00UL9W</p>

[hawkman]

<p>Ralph,<br>

Sorry I am not interested in discussing this further, nor am I interested in rumors.</p>

[danielleetaylor]

<p>Arash and Keith - I suppose you two would claim that 4x5 film is just as noisy as 35mm film because, viewed per grain, they look the same.</p>

[keith reeder]

<p>Ralph,</p>

<p>most people with experience of the cameras "tested" by DxOMark are <em>baffled</em> by their utterly bizarre results - they claim for example that the Nikon D200 has better DR than the Canon 40D!</p>

<p><strong>Trust me - that's arrant bloody nonsense, as are many of their results when compared to the Real World experience of people who have used the cameras in question.</strong></p>

<p>Getting back to the original issue. Let's assume that I have a given composition in mind, where the bird is a certain size in the frame and on the piece of paper I'm going to print on: use the cormorant posted above by Arash as the model.</p>

<p>If I can achieve that composition with <em>either 40D or the 50D without cropping </em> (obviously I'd be downrezzing the 50D image) then of course the noise differences between the two cameras will be irrelevant. Nobody's disputing that - although the act of downrezzing the 50D's file to match the 40D's comp would by definition be hiding some of its noise. Nobody would dispute that either, I trust.</p>

<p>But the issue here is <em>noise comparisons - </em> read the OP again if you've forgotten that<em>. </em></p>

<p>It is a given that downrezzing a file inherently reduces noise in it: advantage the bigger sensor.</p>

<p>Likewise, uprezzing a smaller file plays merry hell with it, and (often greatly) exaggerates the noise and other artifacts visible within it: <em>disadvantage</em> the smaller sensor.</p>

<p>So how - <em>exactly</em> - does up/downrezzing deliver a level playing field?</p>

<p>Besides: in the final analysis, my <em>experience,</em> that of Arash, and similar observations from a number of people whose knowledge and photographic skills I respect, all confirm that in the case of the 50D, more isn't better, and all the arguments in the world about <em>how</em> we're supposed to arrive at this final decision are ultimately moot - I can see noise in 50D images that I can't see in equivalent 40D images.</p>

[c._sharon]

<p>50D may or may not be more noisy than 400D/40D in general. But looking at the sample images provided, I don't see more noise in the 50D image.</p>

[danielleetaylor]

<p><em>Let's assume that I have a given composition in mind, where the bird is a certain size in the frame and on the piece of paper I'm going to print on:</em></p>

<p>and</p>

<p><em>It is a given that downrezzing a file inherently reduces noise in it: advantage the bigger sensor.</em></p>

<p>Don't you understand that in print you are doing something similar? If you want a bird of a certain size on a 16x24, the 50D pixels printed to the 16x24 will be smaller than the 40D pixels at the same print size. Noise will be less apparent than in a pixel-to-pixel comparison.</p>