Conventions shattered: Higher resolution = lower noise levels
Leonard Goh | Nov 28, 2008
Higher resolution = less noise? (Credit: Canon)
However, I have a point to add. The image processor in each camera plays an important role in reducing the noise levels, too. If DxO Labs' theory stands, and when a high-res sensor is coupled with a good imaging engine, the resulting picture will be free of noise. If a sub-standard processor is used, few things could be done to reduce the noise in-camera.
As we've always advocated: Resolution isn't the only thing to look out for when buying a camera. There are other factors such as the lens and image processor to consider, too. For most consumers, a 12-megapixel image is more than sufficient to print an 8R photograph. Unless you need to make A1-sized (59.5cm x 84cm) printout, there are few reasons to lust after a high-res camera.
Talkback
gohleonard says...
yup exactly. and also, when you take a picture, there is much more going on inside the shooter then just the sensor receiving light. all the internal processing and such, especially for compact cameras which cannot capture RAW images, depend heavily on the processors and other mechanics to give a nice-looking image.
Nov 29, 2008 14:56
yup exactly. and also, when you take a picture, there is much more going on inside the shooter then just the sensor receiving light. all the internal processing and such, especially for compact cameras which cannot capture RAW images, depend heavily on the processors and other mechanics to give a nice-looking image.
Nov 29, 2008 14:56
LipWorks says...
The analysis by DxO is very counter-intuitive. To get a high SNR, you either need more signal, or need less noise. Since the amount of light received is signal, and more pixels means per pixel less signal, you should get lesser SNR, meaning poorer pictures.
If there is active steps to reduce noise at the same time when signal is reduced, that would then explain why it could be equal or be better. Where does noise come from in a picture? If you talk about a pixel's sensor, it come from the physical construction of the sensor. There is another type of noise, compression noise and this contributes to digital artifacts - funny pixels that don't exist before jpeg compression but appears after jpeg compression, and this means it has nothing to do with sensors at all (it's the image processor's fault!)
So is this article talking about the sensor, or the image processor? I think the researcher got muddled (else it's the communications guy that's muddled).
Dec 01, 2008 08:56
The analysis by DxO is very counter-intuitive. To get a high SNR, you either need more signal, or need less noise. Since the amount of light received is signal, and more pixels means per pixel less signal, you should get lesser SNR, meaning poorer pictures.
If there is active steps to reduce noise at the same time when signal is reduced, that would then explain why it could be equal or be better. Where does noise come from in a picture? If you talk about a pixel's sensor, it come from the physical construction of the sensor. There is another type of noise, compression noise and this contributes to digital artifacts - funny pixels that don't exist before jpeg compression but appears after jpeg compression, and this means it has nothing to do with sensors at all (it's the image processor's fault!)
So is this article talking about the sensor, or the image processor? I think the researcher got muddled (else it's the communications guy that's muddled).
Dec 01, 2008 08:56
gohleonard says...
Hi LipWorks,
This Crave article is about what DxO Labs has stated, but at the same time provide another side of the theory because, as mentioned, the sensor is just one of the many important elements used when taking a picture.
I agree with you that there are several kinds of digital artifacts, but for RAW images, it was proposed that no form of in-camera processing has been applied. Theoretically then, it would be up to the sensor's performance and i think this is where DxO Lab is coming from.
Dec 01, 2008 14:38
Hi LipWorks,
This Crave article is about what DxO Labs has stated, but at the same time provide another side of the theory because, as mentioned, the sensor is just one of the many important elements used when taking a picture.
I agree with you that there are several kinds of digital artifacts, but for RAW images, it was proposed that no form of in-camera processing has been applied. Theoretically then, it would be up to the sensor's performance and i think this is where DxO Lab is coming from.
Dec 01, 2008 14:38
LipWorks says...
I guess I should have read DxO's article before posting, but after reading the article, it actually makes a fraction of a sense, based on the assumption of 1 scenario: after a photo has been taken at high resolution a user would tend to view the picture at low resolutions, and therefore after applying a reduction in size of the RAW image, you will have less visible noise. In fact I would go and say they have some truth in it! I will agree with Leonard that other parts of the camera is equally important, and hope all consumers don't continue the megapixel chase when it cannot guarantee image quality, it does say in DxO's article the noise reduction of the higher Megapixel camera is manifested when you view the picture at 8Mpixels, I guess we don't buy a 21.5Mpixel 1Ds for this purpose... do we?
Dec 01, 2008 17:20
I guess I should have read DxO's article before posting, but after reading the article, it actually makes a fraction of a sense, based on the assumption of 1 scenario: after a photo has been taken at high resolution a user would tend to view the picture at low resolutions, and therefore after applying a reduction in size of the RAW image, you will have less visible noise. In fact I would go and say they have some truth in it! I will agree with Leonard that other parts of the camera is equally important, and hope all consumers don't continue the megapixel chase when it cannot guarantee image quality, it does say in DxO's article the noise reduction of the higher Megapixel camera is manifested when you view the picture at 8Mpixels, I guess we don't buy a 21.5Mpixel 1Ds for this purpose... do we?
Dec 01, 2008 17:20
To post comments, you need to become a member. It's FREE.
Crave is...
The name says it all. Crave is our blog about cool gadgets and other crushworthy stuff.
Drop us an email if you have tips or suggestions.
Crave for...
» Mobile Phones (2219)
» Digital Cameras (975)
» Notebooks (1228)
» PC & Peripherals (1305)
» Handhelds (584)
» Printers (82)
» Home AV (1139)
» Music & Play (889)
» Gadgets (1476)
» Future Tech (392)
» Green Tech (170)
» Lifestyle (155)
» Luxury (22)
» Home Appliances (57)
» Cars (99)
» Games and Gear (245)
» Software (93)
» Web (107)
Previously...
Hot Topics
What's Hot
apple apple inc. apple iphone apple ipod asia battery camera game google inc. iphone microsoft corp. microsoft windows mobile phone photograph singapore smart phone software sony corp. video
Copyright © 2009 CBS Interactive Inc. All rights reserved. Privacy Policy.
ferdiei
yes its the overall design of the electronics functionaility which dictates the end result of a captured image quality. DxO labs' analysis may have been focused particularly on dSLRs, but in reality the choice of manufacturer's electronic component is what gives them the point of differentiaition in terms of final product pricing & features comparison. so in overall, hi-res doesn't (really) equate to lower-S/N.
Nov 28, 2008 20:06