Upload & Sell: Off
The questions of what the Merrill sensor can deliver touches some more fundamental questions I keep asking myself since a while. I am also attracted by the Sigma cameras, but so far I am not fully convinced.
Luis Cunha wrote:
The usual bayer lack of sharpness is another theme. That I donīt like.
This is - in my opinion - exagerated. I have looked at the question of detail and sharpness through many samples, dp2 merrill and the cameras I have been using during the last years, amongst which the 5d II and the Nex-7. Used with good lenses (in my case a few Zeiss ZF) and good technique all of those can deliver very good detail resolution and sharpness except where the size of the smallest details comes close to pixel size.
In larger objects both techniques (Bayer and Foveon) deliver easily a good image which displays detail far beyond the reach of the human sight. For example in distant trees one can identify single leafs or structures in the bark of trees, or letters on distant traffic signs, billboards etc.
What both technologies have in common is that from a certain point and smaller, a sensor cannot render an object in a way that shows it's real exact shape anymore. A face rendered by let's say 30 or 40 pixels (just a random number) is no longer realistic. Foliage at a certain distance becomes too small to show each single leaf. Through a Bayer sensor with AA filter, this smallest detail tends to get slightly blurred into a veil of undefined texture (one could say "like for the human eye").
With the Sigma/ Foveon sensor, it is not blurred, but remains sharp. Nevertheless it does not show the real shape of the objects because they are too small. Mini details show up as blocks and structures, which seem sharp but nevertheless they suggest a precision that is not there. This sometimes leads to a hyperdetailed look of photos that - in my impression - distracts from the photo as a whole.
The way a photo displays a scene is after all different from looking at the scene in reality. Lets stay with the nature + trees example: When you look at a number of trees in a landscape, noone can simultaniously identify single leafs or even branches, brain would be flooded with information and would be left in total visual confusion, whereas a normal sized print shall display all of them, according to a popular ideal of digital photographers.
I don't pretend to have definitive answers to those questons, but right now it seems to me that further improvement of the Bayer tech which means more pixels with good or even better DR is probably a good direction for further evolution. A regular grid of pixels will never be totally natural, the ideal thing would probably be a random order of very small photo sites, not a regular grid, something like molecules of a film emulsion, but I suppose this is technically not possible in a forseeable future. The best sensor might very well be fine grained large format... eh - film.
Cheers and happy new year to everyone.