I can answer that!
Originally Posted by markjwyatt
It is an interesting article, but still does not quite address a key question: Does sensor size mean anything relative to how large of a print you can make- and I mean not just at "appropriate viewing distances". Is a pixel a pixel a pixel, given there is enough light that those other factors come into play (stabilization, hand holding, dof, etc.)? This may portend more to the future of digital cameras then to the present state as the linked article illustrates well.
For instance on a sunny day or maybe in open shade I get a 40 MP image from all with the same angle of view (say a "normal" lens) with no camera shake, no IBIS or OIS needed (even use a tripod if you want):
1. a good quality mobile device
2. a micro-4/3
3. an ASPC
4. a "Full frame" (35mm) sensor
5. A MF sensor
Can I make the same enlargement (say by ink jet or scanning enlarger on silver emulsion), put them all (1-5) up on a wall and and view them from far and near. Would they all look more or less the same?
If the answer is yes then within some limits (all the things in the article and more) a pixel is a pixel is a pixel. If the answer is no, then the actual size of the sensor still matters.
With film it is clear- if the same film and development is used (say FP4 in DDX using the recommended time for format) for any 1 degree of view the larger negative size will have more pixels per degree and will thus render better both in terms of resolution but also achieving full tonal variation. Also due to optical enlarging, optical effects will also come into play both on camera (which will some effect on the digital example) plus at the enlarger. If we scan the negative then ink jet print or use a laser scanning enlarger on silver emulsion, this aspect is in the same boat as the digital case, leaving mainly the grains/degree aspect.
I shot for several years with a Canon 5DmkII, which is a 20mp fullframe camera. For the last year, I have been shooting with 20mp Micro Four Thirds cameras (Olympus Pen-F and Olympus OM-D E-M1 mark II).
So, we have two 20mp systems, one fullframe, the other Micro Four Thirds. Huge sensor-size difference. I think that fullframe has almost 4 times the surface area of m4/3. I have made and sold 16x20 prints from each.
The difference? To be honest, the m4/3 sensors give the finest detail resolution! Why? Part of it is that Canon uses a rather aggressive anti-aliasing filter on the 5DmkII's sensor. This softens fine detail, not all of which is recoverable through image sharpening. Olympus's 20mp sensors have no anti-aliasing filters. Another thing is that the Canon lens I used for the photos I printed large just plain aren't as sharp as the Micro Four Thirds lens I used.
I used the Canon 24-105mm F4L-IS lens for fullframe and the Olympus 12-40mm f2.8 Pro lens for m4/3. The Olympus lens is equivalent to a 24-80mm fullframe lens. I am continually amazed at the image quality of this lens.
So, a smaller sensor with better lenses and no anti-aliasing filter beats a much larger sensor with poorer lenses and an anti-aliasing filter that blurs the finest details.
There is one other difference, and that is noise. The fullframe Canon sensor has less of it than the Olympus Micro Four Thirds sensors. I've found, however, that even using more noise reduction in Lightroom to eliminate the m4/3 noise, the m4/3 images are still better, with more fine detail resolution. Even at high ISO.
More modern fullframe sensors have less noise than the one used on the Canon 5DmkII, so at high-ISO speeds a modern fullframe sensor might be superior to the m4/3 sensors.