The reason given was that each individual pixel sensor was relatively 'huge' and as a simple consequence it gathered more light per pixel. It is the larger pixel size OR in more recent FF cameras, the size and/or the sheer number of pixels, which can produce smooth clean images - although this is MOST noticable when using them at high ISOs (when the lower noise becomes most apparent) and is subject dependent. All this is probably to do with the increased information capacity which FF is able to utilise and is a purely physical attribute of a larger sized sensor.
It's really SNR (Signal to Noise Ratio) that's the concern. Canon did two things well; it controlled noise while increasing signal with large pixel sizes. Today, the 5D has pixel capacity that exceeds the dynamic range of its converter. That means it has identical sensor performance at ISOs higher than base. It also means that Nikon's DX sensors, which also max out the performance of their converter, essentially match the Canon's at base ISO. Current Nikon models have tonality that matches Canon FF at base ISO.
Any talk of increased pixel count improving IQ is just clouding the issue. Increasing pixel count increases resolving power but it does not improve noise or increase tonality. Making such a claim requires comparing images viewed at different dpis which is inherently unfair.
...the other attributes/differences of FF are probably of less relevance and at this stage in the development of digital cameras and they may well be better off sticking to cropped sensor cameras which have other characteristics better suited to such a lens in any case.
The attributes referred to, i.e. better tonality, don't exist for reasons I've stated above and in another thread. It did exist at the time of the 1D, it does exist for higher ISOs, and it may exist again with the upcoming 14 bit converters, but it does not exist with current cameras at ISO 100 where most UW shooting occurs. I've linked to this article before but I'll do it again. It's an in-depth discussion of the issues involved including tests of relevant cameras including the D200 and 5D. The entire article is good but Fig. 5 shows exactly what I'm talking about. Reading the article will clear up a lot of myths and it's written by a Canon guy to boot. ;-) Some feel that the 5D's high ISO performance is its great advantage; I consider it it's achilles heel. How much greater would the 5D have been for us if it had offered 14 bit converters?
For the future who knows what we will see happen? My guess (based on discussions with a lens designer amongst others) is that we'll see 1.5x sensors top out at ~15 MPixels due to lens constraints (Nikon have - well one of their designers has - after all stated that MPixels have matured), but who knows what developments will take place in the future.
I agree, and that means that Nikon and others will inevitably have to increase sensor size or conceed the resolution advantage to those that have/do. It appears that there's at least one more stop, maybe more, in performance to be had at current pixel sizes, but with Canon raising ADC resolution to 14 bits, Nikon is not likely to be able to shrink pixel sizes and keep pace unless they skip 14 bit. That means FF is inevitable for those application where IQ is most important and I'm convinced UW wide angle shooters will want 14 bits. Current rumors have Nikon introducing FF in the D3 model. There are good reasons why that could happen.
I'm certain a fisheye zoom could be made for FF. The real question is how much it would cost and how many would buy it. I want an updated zoom macro with modern focus and 1:1. I'm certain they could make it but I doubt there is a market.