Some shooters are willing to trade off FX IQ for DX compactness, crop factor and weight. For others, this isn’t acceptable. FX and DX can coincide for a long time

So far, I have not seen significant advancements in APS-C compared with the jump from D3 -> D3s. (The D3s is at least 3+ stops better than D300s at high-ISOs – this disparity cannot be attributed to just the pixel sizes alone) It’s the APS-C sensors that need the upgrade – they have fallen quite far behind.

Better yet would be some sort of digital interconnect. XLR is crappy technology.

People who want quality are going to go to the trinity instead, or use primes. I would see this as a stepping stone lens for folk moving to FX from DX.

As for ‘different markets’, I would see ‘different uses’. Most professional photographers have a backup. The primary market for the D3x is people who already have a D3. The D3 was a little odd in that respect as the D3 and D300 were both so much better than the D2H/X. If you had DX lenses a D300 backup was going to be needed.

If we give up the expectation that the D4 launches at the same time as the D400, there really is no reason not to bring out a D4 as soon as they can manufacture a 24MP sensor using the new technology that delivers ISO6400 performance in quantity.

The idea of a D900 just seems wrong to me. A D700s would make prefect sense: As soon as the D300s run is done, take the D3s sensor and update the D700. A D900 should be a completely new model, either an 18MP or a 24MP.

If there is a D900 it is because Nikon is protecting their D700 line against the competition from Canon and put a higher priority on that than keeping to their traditional roadmap. You could even argue that the jump from 700 to 900 was meant to indicate that something different is happening. So if the D900 launches I think that any expectations for the D4 launch date have to be put aside. Nikon is going to have to update the flagship line so it is ahead of the D900. So the D4 might come as soon as the D3s production run is complete.

Whatever happens, I cannot see any way that Nikon would make another production run of the D3x sensor. They have better technology, they have to use it.

The technology to make a 50MP FX camera already exists, it is already in some of the Coolpix cameras which have tiny sensors that would correspond to 80, maybe 100MP if made full frame.

Making a larger sensor is a function of process yield. In general doubling the size of a sensor means that you have doubled your chances for failure. If the process yield is 99% then doubling the sensor size means your yield falls to 99% * 99% = 98%

The problem is that the VLSI processes we are talking about are much more difficult than regular process. So the yield is more likely to be 80% so your yield is going to be 64%. The net is that if you start with a low yield then doubling the size of the die means that you get far fewer than half the number of good chips. If the yield is 50% you are only going to get a quarter of the number of good ones.

I would not be surprised if Nikon and Sony have not pushed through a few wafers with some really large sensors on them just so they can see what the effect is. Sony will be doing that sort of thing as part of their process improvement program. To increase yield you make stuff that is likely to fail and then put the wafers under a microscope.

So I would not be surprised if there wasn’t a 50MP body or two floating about round Nikon HQ.

There is an inescapable relationship between ISO rating and pixel pitch. Smaller pixels mean less light and more noise per pixel. The ISO rating is dependent on noise per pixel and so for any given technology there is a tradeoff.

But ISO response is not the same as noise response. If you have a 12MP and a 24MP black and white digital camera built using the same technology and sensor size and take pictures at the same ISO settings and then downsample the 24MP photo to 12MP by averaging across pairs of adjacent cells you should end up with nearly identical results, that is all that is happening with larger pixel area. Colour and the Bayer pattern makes a difference of course. But not much and may actually help the higher resolution camera.

A much better way to measure noise would be to compare the noise levels when the RAW files are converted to a PNG image at a specific resolution. If we use this method of measuring a higher resolution sensor is likely to provide a slightly lower noise rating in a given technology than a lower one which appears to have a higher ISO response.

The reason for that is that more pixels means more information for noise reduction algorithms to work on. [Which in turn means that you need to introduce some measure of contrast or the metric is too easily gamed.]

So as a rough guide to low light noise response in finished photos multiply the Max ISO figure by the sensor resolution.

A better measure would be to have a chart similar to an MTF chart but for sensor response giving noise response for the maximum resolution and down sampled to 12MP. The max ISO rating is interesting but irrelevant since I don’t usually shoot at max ISO (even on a D300 I am more often finding that I want lower ISO response than higher). I want to know the noise response at ISO200/400/800/1600 That is where I take most of my photos.

The D200 already kills it in resolution, SNR, and DR, lol!

I don’t think it’s going to be too hard. . .sure, better video, etc., but as a camera, the D300s stands on its own against the 7D. . .(minus a few mpx)