11-28-2010, 05:22 PM
[quote name='Brightcolours' timestamp='1290961524' post='4528']
If we look at the high ISO studio (not ideal) shots from dpreview, we do NOT see the K7 being "far better" than the 7D/D7000/60D, they all are very much the same (looked at 100% pixel level, which means we do magnify a bit more with the 7D and 60D).[/quote]
Agreed. Keep in mind that they also all suffer from various DOF issues, which limit the usefulness of the test. I guess the only way to overcome this and still test the sensor to its maximum is to switch to 2D targets (e.g. something like the house poster than imaging-resource.com is using).
In principle NR on RAW is a bad thing, but if the end result is still better than the competition then it can be forgiven. In my opinion a better way to evaluate the "end result" is to use several different RAW conversion applications (including the camera manufacturer own software), while putting some effort to customize the settings of each application to get the best image quality. Afterwards the best result from this group of images can be selected by comparing two images at a time, in a tournament-like process. This process should then be applied to RAW files from all the other cameras in order to get the best possible results that each camera can deliver.
The problem that remains with this approach is that image quality is still mostly a subjective affair, so “best†can be different for different people, but I still think that a comparison of images processed and selected this way would be more representative of how these cameras actually perform. The dpreview method of comparing images converted using ACR default settings is far easy to perform, but the results are of limited value. Small brands specifically (Pentax, Ricoh) suffer from lack of optimization by Adobe compared to the big two (possibly due to the use of DNG which encourages Adobe to be lazy).
My hunch is that even with the in-camera NR (RAW cookery), the K-5 high-ISO results would leave the 7D well behind, providing cleaner images that are also more detailed. However, I'm less certain about the D7000. It could be an interesting (though very time consuming) experiment. Any volunteers?
If we look at the high ISO studio (not ideal) shots from dpreview, we do NOT see the K7 being "far better" than the 7D/D7000/60D, they all are very much the same (looked at 100% pixel level, which means we do magnify a bit more with the 7D and 60D).[/quote]
Agreed. Keep in mind that they also all suffer from various DOF issues, which limit the usefulness of the test. I guess the only way to overcome this and still test the sensor to its maximum is to switch to 2D targets (e.g. something like the house poster than imaging-resource.com is using).
Quote:What we do know about the high ISO from the K5 is that it gets some kind of filtering in the RAW data from ISO 3200 and up.
In principle NR on RAW is a bad thing, but if the end result is still better than the competition then it can be forgiven. In my opinion a better way to evaluate the "end result" is to use several different RAW conversion applications (including the camera manufacturer own software), while putting some effort to customize the settings of each application to get the best image quality. Afterwards the best result from this group of images can be selected by comparing two images at a time, in a tournament-like process. This process should then be applied to RAW files from all the other cameras in order to get the best possible results that each camera can deliver.
The problem that remains with this approach is that image quality is still mostly a subjective affair, so “best†can be different for different people, but I still think that a comparison of images processed and selected this way would be more representative of how these cameras actually perform. The dpreview method of comparing images converted using ACR default settings is far easy to perform, but the results are of limited value. Small brands specifically (Pentax, Ricoh) suffer from lack of optimization by Adobe compared to the big two (possibly due to the use of DNG which encourages Adobe to be lazy).
My hunch is that even with the in-camera NR (RAW cookery), the K-5 high-ISO results would leave the 7D well behind, providing cleaner images that are also more detailed. However, I'm less certain about the D7000. It could be an interesting (though very time consuming) experiment. Any volunteers?