[tr13]

No. Those cameras will replace some P&S and a few entry level DSLR. Just like optical binoculars won't go away, DSLR's won't either. DSLR's can be made small enough. It will still be the enthusiast choice....

I didn't say that they will be vanished in very near future. No.
Technology is not ready yet.
Sensors will be much more sensitive, at least 2-3 stops in 2-3 years. And Sony is near production state with new sensors.
Electronics viewfinder with FullHD/60fps resolution is superior to optical for ordinary user. If you ever used latest camcorders, you know that JVC-like focus assist could be better then all that we saw in DSLR world.
And, finally, look at DSLR construction. They are simply not competitive due much higher cost of components and extreme amount of adjustment required.

[ozlizard]

Electronics viewfinder with FullHD/60fps resolution is superior to optical for ordinary user. If you ever used latest camcorders, you know that JVC-like focus assist could be better then all that we saw in DSLR world.
And, finally, look at DSLR construction. They are simply not competitive due much higher cost of components and extreme amount of adjustment required.

It has been my understanding that the quality of even HD video (I assume that is what you are referring to), is not equall to the IQ of a top end DSLR not matter how many frames per second it is capable of. Everbody is talking about convergence and when that is over we will all be talking about specialised equipment as the holy grail. Still cameras will still be around, photography didn't kill painting, 3d didn't kill sculpture, DVD's didn't kill movies and camcorders or video enabled DLSR's wont kill photography. Art lives on.

[tr13]

HD video have no relation here, as well as video enabled DSLRs (as I don't believe in them as camcorder eaters :-) ).
All I want to say is that you must see FullHD 60fps picture with focus assist.
And you'll be dreaming about it.
Potential is really great here, as even 5D Mk II scan only each 4th line as far as I remember. So, four times more viewfinder sensitivity is possible with small improvements.
It'll be even better then optical focus screen as you'll have electronic zoom to make perfect focus then you need one.
I don't even mention fast WB adjustment.
Something like three wheels with few modifier buttons may become common.

[snogglethorpe]

I don't agree with "crappy pictures" rants.
I love my TZ5 much better then my Canon S60.
And latter have larger sensor and fewer pixels, but average picture just looks much better on TZ5. Better WB, better noise reduction (if GX20 could have such noise reduction quality as TZ5..), OIS. And auto mode coudn't be compared to GX20, as on GX20 it is barely useful while on TZ5 it could find a face, find proper focus and even fire flash if background is very bright, I don't even mention that it automatically slightly blurres skin tone to hide defects.
Remember usability mantra - ordinary users looks stupid at first sight, but in reality they are not, they are just busy doing really important things.
yes, DSLR is very powerful thing, but it is thing from the past. Last dinasaurs.

There are several independent issues. I think on some, you're probably right, on others it's not so obvious.

• (embedded) software -- the reasons why you like your TZ5 sound like they're pretty much entirely due to the software it uses. It's very unlikely the TZ5 is doing anything unusually clever, but clearly, the makers perceived a need for those features, and went to the trouble of implementing them. If there's a demand for similar features in the DSLR market, it's not hard for DSLR-level camera makers to implement them, so this type of thing can happen in the short term. [Canon and Nikon, for instance, already do more NR on the camera than Pentax, and that pleases a lot of people.]
  
• Optical viewfinders -- I agree that eventually electronic viewfinders will probably win out, but it seems like it's going to take a while before their quality is acceptable to the people who currently buy DSLRs. The EVFs on the current panasonic G1/GH1, for instance, are awful; I'd never use anything like that over an optical viewfinder, and I'm not really all that picky!
  
• Sensor size -- there will continue to be advances in sensor technology, but it seems like it's going to be a long time before smaller-than-APSC sensors can deliver the quality demanded by pickier users (who are buying DSLRs). As soon as there are technology improvements, even if they're driven by the cellphone/p&s market, they'll be snapped up in larger form by the larger-sensor market to improve quality, but I think large sensors will continue to be important for a quite a while.

[tr13]

There are several independent issues. I think on some, you're probably right, on others it's not so obvious.

• (embedded) software -- the reasons why you like your TZ5 sound like they're pretty much entirely due to the software it uses. It's very unlikely the TZ5 is doing anything unusually clever, but clearly, the makers perceived a need for those features, and went to the trouble of implementing them. If there's a demand for similar features in the DSLR market, it's not hard for DSLR-level camera makers to implement them, so this type of thing can happen in the short term. [Canon and Nikon, for instance, already do more NR on the camera than Pentax, and that pleases a lot of people.]
  
• Optical viewfinders -- I agree that eventually electronic viewfinders will probably win out, but it seems like it's going to take a while before their quality is acceptable to the people who currently buy DSLRs. The EVFs on the current panasonic G1/GH1, for instance, are awful; I'd never use anything like that over an optical viewfinder, and I'm not really all that picky!
  
• Sensor size -- there will continue to be advances in sensor technology, but it seems like it's going to be a long time before smaller-than-APSC sensors can deliver the quality demanded by pickier users (who are buying DSLRs). As soon as there are technology improvements, even if they're driven by the cellphone/p&s market, they'll be snapped up in larger form by the larger-sensor market to improve quality, but I think large sensors will continue to be important for a quite a while.

OK.

1) This is not software only issue. But, YES, mainly software. And software will be much more important with each day. So good noise reduction is due to very good mathematical algorithms, but they are supported by special sensors and ADC chips by Panasonic. You could compare latest Samsung HD P&S and ones from Panny, both use same sensors and most probably even same processors. But results are not similar.
2) Is it hard to implement same functionality in DSLR? Yes. Most of functionality can't be implemented in current DSLRs in useful way. For example, K-7 uses same face recognition library as TZ5, but usability is much worse.
3) Did you hear about Samsung NX3? As I understand its VF is much better. And technology is ready for FUllHD small screens.
4) Sensor development is most problematic field, I agree. But this picky users are not very important. Much more important is the fact that high ISO numbers will be perfectly acceptable by general user. ISO 1600 will be same as current ISO 200.

[ytterbium]

Talking about sensors, have a look at this:
sCMOS - The online home of Scientific CMOS technology
Scientific CMOS.
5,5 Megapixels (2560(h) x 2160(v)) (And i hope they dont plan to increase much).
Readout noise: < 2 e- rms @ 30 fps; < 3 e- rms @ 100 fps
Maximum fps: 105 fps for rolling shutter and 52.5 fps for global shutter
Pixel pitch 6,5 μm
Dynamic range > 16 000 : 1 (at 30 fps) (approx 84 dB).
Maximum quantum efficiency: 60% (With excellent red and near infra-red response).

To compare, high grade full frame (sensor type not format) CCD:
Kodak KAF-6303 6.3 MP 3088 x 2056 9.0um 33.4mm diagonal 0.6fps:
Quantum efficiency: 40%, 52%, 65% (depending on manufacturing).
Readout noise: 15 electrons rms
Dynamic range: 76 dB (~6310:1) NOTE: Its not directly linked to saturation brightness but min:max value, you can increase theoretical DR by decreasing noise.
Maximum data rate: 10 MHz (0.6 fps).

But i dont think that those are that kind of sensors that are being used in cameras. unfortunately.

[Pål Jensen]

So, four times more viewfinder sensitivity is possible with small improvements.

And what is the viewfinder sensitivity of a piece of glass? An optical finder is the optimum choice for an aiming and composition device. Thats why we still have optical binoculars and not LCD screens. LCD finders are a solution to a non existing problem as far as DSLR's are concerned.
A great LCD screen in addition to an optical viewfinder would be nice; even as an overlay....

[offertonhatter]

Personally I would like a stop to the Pixel race.

I can see the APS-C going to maximum of 20MP, but no more. Even with advance is tech it will probably be too difficult to produce a sensor that will produce great IQ with in excess of 20mp on APS-C. This would also mean that the upper limit for full frame would be 30mp.
However I could be proved wrong.

I would rather see a fix on the biggest problem of a digital sensor over film, namely dynamic range. Hopefully, if a sensor can produce a say 20000/1 contrast range, then sensors would be able to produce a truer image of the world that the naked eye can give.

Ultimately, the best solution would be a total new outlook of the sensor.
By getting ride of the Beyer pattern, and each pixel being able to replicate 16 million colours, not only would you have a truer larger image, but something that neither film nor current sensors can produce. The problem would be to get an individual sensor to be able to record 16 mil colours, rather than red green or blue. But of course this is pipe dream, yet problems like this did'nt stop the great inventors like Turing, Brunel, Edison, Whittle etc etc.

[pingflood]

Personally I would like a stop to the Pixel race.

I can see the APS-C going to maximum of 20MP, but no more. Even with advance is tech it will probably be too difficult to produce a sensor that will produce great IQ with in excess of 20mp on APS-C. This would also mean that the upper limit for full frame would be 30mp.

I think you're confusing the crop factor with the sensor area. A 1.5 crop sensor has 1/1.5^2 = 1/2.25 times the area of a FF sensor. So if APS-C "maxes out" at 20MP, the equivalent pixel density on the FF sensor would be 45MP.

[Das Boot]

What good would a high MP sensor do if it's diffraction limited to say f/4.0? All this wonderful glass that's razor sharp at f/8 would be useless along with getting a picture with any sort of depth of field. It's time for the manufacturers to stop the MP cock fight and focus on more important things like dynamic range, noise reduction, a built-in Zippo, beer dispenser, and X-Ray vision to see through the clothes of hot women (and no my ir camera can't do it - I've tried)

[pingflood]

What good would a high MP sensor do if it's diffraction limited to say f/4.0? All this wonderful glass that's razor sharp at f/8 would be useless along with getting a picture with any sort of depth of field. It's time for the manufacturers to stop the MP cock fight and focus on more important things like dynamic range, noise reduction, a built-in Zippo, beer dispenser, and X-Ray vision to see through the clothes of hot women (and no my ir camera can't do it - I've tried)

And it's time for photographers/camera owners to quit blaming the lack of dynamic range, noise reduction, or built in beer dispensers for their lack of ability to produce a good picture. I wonder how many of the pixel peepers out there are convinced that once THE camera arrives it'll somehow be all they need to produce great shots.

[Das Boot]

And it's time for photographers/camera owners to quit blaming the lack of dynamic range, noise reduction, or built in beer dispensers for their lack of ability to produce a good picture. I wonder how many of the pixel peepers out there are convinced that once THE camera arrives it'll somehow be all they need to produce great shots.

Amen brother. Amen....

Hey Ping, are you a garnet & black man or an orange & white one? (please don't say the latter since your credibility as a whole rests on your answer ;-) )

[ytterbium]

Ultimately, the best solution would be a total new outlook of the sensor.
By getting ride of the Beyer pattern, and each pixel being able to replicate 16 million colours, not only would you have a truer larger image, but something that neither film nor current sensors can produce. The problem would be to get an individual sensor to be able to record 16 mil colours, rather than red green or blue. But of course this is pipe dream, yet problems like this did'nt stop the great inventors like Turing, Brunel, Edison, Whittle etc etc.

I apologize to be rude, but have you red anything about human colour perception, colour spaces, standards and where does those 16 millions would come from?

A good start might be here: CIE 1931 color space - Wikipedia, the free encyclopedia

As for Bayer pattern. Well it actually does have one big downside. Its a filter design - this means that all other useful light (approx 2/3) is absorbed by the filter. This means that more than two times lower light is collected than falling on sensor - more than one stop is "eaten" by bayer filter.
I've heard of prism designs (similar as microlens), where small prism covers few pixels and redirects each colour component to separate pixel. So the other colours are not simply absorbed by filter but redirected to appropriate pixel.
(Something like 3CCD but primitive separate micro prism for each 3 pixels).

[tr13]

In fact 3 CMOS 1" sensors are possible and are quite cheap (I guess, cheaper then one APS-C) :-)
In reality this color information is not very useful, just our eyes have worst color resolution :-)
This is the reason why all video have much lower resolution in color channel.
Next rece will be towards sensitivity.
Compacts with usable ISO 1600 and prosumers with perfectly usable ISO 12600 :-)

[ytterbium]

The thing about rgb prism is not about colour resolution or fidelity. Its about availible light ands sensitivity, because with simple Bayer filter you just discard 2/3 of light (Red and blue at green pixel site for example).
Only i dont know how much light one would loose in prism. Such approach would require redesigned pixel structure which could lead to other inefficiencies as well.

One could design a complex crossed prism pattern that can serve as precision low pass filter, micro lens and colour splitter at the same time.