OO7MIKE wrote:
Several professional camera review sites such as dpreview.com and popphoto.com compare camera noise by looking at the pixel level.
Noise at a per-pixel level is misleading as few people view images at a per-pixel level. Even the largest sized monitors on the regular market (QSXGA, 2560x2048) have less pixels than photographs from a Digital Rebel, let alone a 5D2 or D3X. Most people will either:
A) Resize the image to a web-friendly image
B) Print the image
Some people will crop, but even then, a pretty heavy crop would be necessary to view the entire image at 100% on all but the largest monitors.
Thus, a per-pixel noise test is inherently flawed and belies the final outcome of someone\'s photographs. It also obfuscates other positive qualities of an image that make image noise irrelevant. The 50D was criticized as having more noise at the pixel level as the 40D at similar ISO sensitivities by the \"professional\" camera review sites, but ignores that the greater resolution of the 50D means that the noise is finer and interferes less with the image as a whole.
In my opinion, measuring noise at the pixel level makes it difficult to compare image noise from image to image. Thus, the best way to represent image noise of a digital sensor is to utilize a ratio of x noise to y area, so that you can scale the amount of noise accordingly.
OO7MIKE wrote:
Several professional camera review sites such as dpreview.com and popphoto.com compare camera noise by looking at the pixel level.
Noise at a per-pixel level is misleading as few people view images at a per-pixel level. Even the largest sized monitors on the regular market (QSXGA, 2560x2048) have less pixels than photographs from a Digital Rebel, let alone a 5D2 or D3X. Most people will either:
A) Resize the image to a web-friendly image
B) Print the image
Some people will crop, but even then, a pretty heavy crop would be necessary to view at 100% on all but the largest monitors.
Thus, a per-pixel noise test is inherently flawed and belies the final outcome of someone\'s photographs. It also obfuscates other positive qualities of an image that make image noise irrelevant. The 50D was criticized as having more noise at the pixel level as the 40D at similar ISO sensitivities by the \"professional\" camera review sites, but ignores that the greater resolution of the 50D means that the noise is finer and interferes less with the image as a whole.
In my opinion, measuring noise at the pixel level makes it difficult to compare image noise from image to image. Thus, the best way to represent image noise of a digital sensor is to utilize a ratio of x noise to y area, so that you can scale the amount of noise accordingly.
Dec 28, 2009 at 05:22 PM
Previous versions of kakomu's message #7937298 « 7d low iso noise vs 40d »
