Nikon D90 or Pentax k20d or Cannon 50d

[Dougjgreen]

That doesn't mean that full frame won't be offered in significantly cheaper DSLRS, of course they will be.

10 years ago, could you have predicted that 4gb flash drives would be in the dollar bin at 7/11? You may have even been inclined to think that Zip drives would catch on and become the standard medium...

Just for clarification: I'm not saying that polymagnesium/relatively high performance full frame slrs are the future, just full frame on its own.


Okay, I'll mark your words and you can mark mine. Time will tell this one ;)
In any event its a moot issue, full frame lens still give a greater sweet spot, which is reason enough on its own and the point of the argument in the first place, whether or not to get full frame lens .

Curious - which semiconductor company? What clients did you sell those CMOS sensors to?

5 years ago I had this same conversation with Nikon fans, I insisted that CMOS was the future and that CCD was a dying technology, 5 years later and while CCD is still commonly used, it is *crystal clear* that CMOS is the sensor technology of the future, undeniable at this point.

The company I worked for was Conexant. Our sensors were pretty much used in phone cameras, and a few point and shoots, but the economics of sensors with respect to size are straightforward to understand. Full Frame is a special problem, because a full frame sensor can only be made using a die with a minimum 43 mm diagonal, which can't ever be reduced. That constrains it to very specialized photolithography which will become ever more of a niche over time. The way electronics are cost reduced is to downsize the die, and SLRs have fixed format sizes that can't get smaller over time. Very few other semiconductors, other than the most costly processors, require a die of around 43mm diagonal. In addition, the process to add microlenses to the finished die is specialized and extremely low volume as far as semiconductors are concerned. A big sensor is extremely difficult to cost reduce, and Moore's law doesn't apply to anything that can't be physically down-sized (unlike Camera-phone and point and shoot sensors, which are still getting smaller). The only real cost reductions that can be gained over time is to improve the process yields, and to improve the efficiency of adding the microlenses. But most of the cost is reflected in sheer silicon real estate, which can never be reduced in a sensor of fixed format size - in fact, that's going UP somewhat over time, although not as fast as yields and adding microlenses costs are going down.

The only GOOD news about this is that both APS and Full Frame sensors will keep getting BETTER over time, because they can't get smaller, each new process generation gives the designers more free real estate to make either more pixels, or more sensitive pixels and other support circuitry around the pixels.

Oh, and anyone who understands semiconductor process technology COULD predict that a given size memory device will continue to get cheaper forever - and that 4 GB of flash memory would soon cost under a dollar. And that Zip drives would NEVER displace solid state memory in the long term because of this. But the cost reduction mechanism which makes flash memory forever get cheaper is by downsizing the memory - that's the Moore's Law mechanism, and that mechanism doesn't apply to sensors that have to have a fixed physical size.

The bottom line is that a full frame sensor will NEVER cost under $200. Which means that a full frame camera body will NEVER cost under $1000 - and more likely, never cost under $1500 if it's made of metal with an optical pentaprism and a rugged shutter. And when they do cost $1500, otherwise similar APS bodies will cost $500 tops.

[Loupey]

...And when they do cost $1500, otherwise similar APS bodies will cost $500 tops.

I would agree with those kinds of figures, but in 5~7 years instead of 10+

I also agree with the others who think that APS, full-frame, and 4/3 formats will all continue to develop their own markets.

But I don't think the price differential will be strictly due to the manufacturing costs of the sensors alone. I can imagine APS cameras taking on a more agressive evolution route of "smaller, lighter, cheaper" and thus diverging from the full-frame group. To me, the Canon 40D/50D line doesn't make marketing sense in light of the falling prices of the 5D (and eventually the 5DmkII and its future replacement). I think that that physical size will eventually become the hallmark of only full-frame cameras.

And when they do become $500/$1500 cameras, who's to say that one of the manufacturers won't pop out a $3000 medium format sensor? Even if the pixel technology remains the same across the formats, there will always be those who want the small, medium, and large stuff. Given the same imaging technology, "bigger" always equalled "better" in terms of image quality (OK not exactly, but I think you get my drift).

Or do we think that, when that time comes, there won't possibly be any need to have imaging sensors larger than 24mm x 36mm?

[Dougjgreen]

And when they do become $500/$1500 cameras, who's to say that one of the manufacturers won't pop out a $3000 medium format sensor? Even if the pixel technology remains the same across the formats, there will always be those who want the small, medium, and large stuff.

The only fly in this ointment is that it's probably always going to be impossible to make a monolithic semiconductor sensor that's that big. In general, Medium format sensors use multiple die (usually 4) packed together on the same substrate. And this introduces other costs needed to align and calibrate the 4 sensors together, as well as more complex packaging. So, at a minimum, the cost of a medium format sensor will always be several times the cost of a full 35mm frame sensor.

BTW, I don't doubt that APS sensor SLRs might some day - even rather soon, diverge from the current form factor - Just look at what Panasonic has done with the G1, eliminating the optical finder light path, for a preview.

[Loupey]

...BTW, I don't doubt that APS sensor SLRs might some day - even rather soon, diverge from the current form factor - Just look at what Panasonic has done with the G1, eliminating the optical finder light path, for a preview.

That's a good point about the OVF. Eliminating that would definitely fit the "smaller, lighter, cheaper" mode and make the transition easier for those coming up from the P&S arena.

So long as there remains a line which has the OVF (perhaps only in the full-frame line in the future) for die-hards like me who can only see the world through a viewfinder

[Loupey]

The only fly in this ointment is that it's probably always going to be impossible to make a monolithic semiconductor sensor that's that big. In general, Medium format sensors use multiple die (usually 4) packed together on the same substrate. And this introduces other costs needed to align and calibrate the 4 sensors together, as well as more complex packaging. So, at a minimum, the cost of a medium format sensor will always be several times the cost of a full 35mm frame sensor.

That's OK. It wasn't too long ago that we were primed into thinking that the lowest price point to have full-frame was $3,500. So comparing $3,500 of then-dollars to perhaps $5,000 of future-dollars for medium format would be a steal.

I didn't sell my medium format gear in hopes of someday getting a digital back for it. But the last time I checked, such a back cost $20,000+