09-30-2011, 10:43 AM
[quote name='boren' timestamp='1317377368' post='11992']
Brightcolours, if your suggestion is only related to sites/organizations and not private users, and if these sites/organizations don't plan to display the on-screen difference to visitors, (like dpreview or imaging-resource do), then I guess your suggestion may have merit. A few additional questions:
1. Is there any site/organization that successfully carried out such tests, or is this suggestion only based on theory?
2. What print size do you believe would be sufficient to demonstrate the differences between today's 16-24 MP cameras?
3. Aside from resolution and noise, how would you use prints to measure tonal range, dynamic range and color depth? These specific parameters are limited by the printers to a greater extent than the theoretical limit placed by the file bit-depth itself.
[/quote]
1. I have no idea how the different test sites and magazines come to their ratings, an I am sure I do not even know of the existence of a 10th of them. I only said that IF you want to look at images at the same size, to compare noise performance and impact on the image, downsizing the images to then measure them (all still digital, no print) is not a good way, as downsizing messes with pixels and so also with noise, making results very unreliable. If you want to compare images at the same size, then print them at the same size. Of course, then one can only do a crude visual check, and not let computer algorithms compare them with whatever criterias you have come up with.
2. Since we were only talking about comparing noise (and related DR) from different resolution sensors, any size that is not so small that noise disappears would be fine.
3. I wouldn't use prints to measure other aspects at all. I was merely talking about the problems of downsampling an image before performing measurements in name of "normalization".
Brightcolours, if your suggestion is only related to sites/organizations and not private users, and if these sites/organizations don't plan to display the on-screen difference to visitors, (like dpreview or imaging-resource do), then I guess your suggestion may have merit. A few additional questions:
1. Is there any site/organization that successfully carried out such tests, or is this suggestion only based on theory?
2. What print size do you believe would be sufficient to demonstrate the differences between today's 16-24 MP cameras?
3. Aside from resolution and noise, how would you use prints to measure tonal range, dynamic range and color depth? These specific parameters are limited by the printers to a greater extent than the theoretical limit placed by the file bit-depth itself.
[/quote]
1. I have no idea how the different test sites and magazines come to their ratings, an I am sure I do not even know of the existence of a 10th of them. I only said that IF you want to look at images at the same size, to compare noise performance and impact on the image, downsizing the images to then measure them (all still digital, no print) is not a good way, as downsizing messes with pixels and so also with noise, making results very unreliable. If you want to compare images at the same size, then print them at the same size. Of course, then one can only do a crude visual check, and not let computer algorithms compare them with whatever criterias you have come up with.
2. Since we were only talking about comparing noise (and related DR) from different resolution sensors, any size that is not so small that noise disappears would be fine.
3. I wouldn't use prints to measure other aspects at all. I was merely talking about the problems of downsampling an image before performing measurements in name of "normalization".