Originally posted by jstevewhite That's pretty much what you already said, and I said I more-or-less agree with; but it's not physics, it's technology. Many years ago, before practical digital cameras existed, people had a million reasons digital photography could never exist; Then they had a million reasons why it could never compete with film; then they had a million reasons why it could never displace film... and so on. Technology marches on. There's no way of predicting whether or not tomorrow Ricoh (or anyone) is going to announce a new phase system that can work twice as accurately with half the photons. That was my only point there.
No, it is physics!
Phase detect AF systems require at minimum 30% of the light TTL to get through to their array. This has been the case for 30 years now. We cannot increase the amount of light transmitted due to the First Law of Thermodynamics. Nor can we increase the sensitivity of the CCD on the phase system too much or it freaks out due to noise, highlights, etc. There's lots and lots of literature on this point
Understanding Camera Autofocus
70% is reflected through the penta. This all flips out when shutter pulls.
The only way to increase the amount of light is to reduce minimum aperture to less than f/8 or something like that (not practical), or to decrease the 70% going to the VF, but that creates brightness issues with our own eyes as well as metering. In low-light our all systems currently need a strobe flash and/or an AF assist lamp. If you're shooting a Nikon D700 and are amazed it can "see in the dark", that's because a hefty amount of real estate has gone into the AF system at the expense of size.
To get around this Sony incorporated an EVF into the A33/55 so the pellicle still transmitted enough light to the sensor after the AF system gets its share. Since the amount of light is fixed in total, the only way to take a lesser share of the pie and make it functional is to capture it through an enhanced EVF. The EVF of the Sony A55 blows away the OVF of the Nikon D300, but you have all the compromises of an EVF, especially when seeing DR and in low-light.
In fact, the recent trend has been to actually make phase detect AF systems bigger so they can 3D track through cross-points and for greater accuracy overall as digitial sensors (especially FF) can pick up the slightest sharpness loss due to focus error. If we increase the CCD sensitivity it becomes less accurate and less able to handle diverse lighting situations (Pentax K-5). If we increase the light we have less bright and even unusable VF. If we increase the aperture we are stuck at f/8 or larger.
It's fairy dust thinking that Pentax has some engineer in a closet who can do more with less with PD AF when no other company has been able to do so. It's been 3 decades looking for this and the end result has been focus assist, making the array even larger, or switching to contrast detect and new mounts, systems, etc. entirely. Whole companies (Olympus, and maybe Sony) have switched their entire corporate philosophy when the tech hits an unreachable limit.
My hope is that Pentax still produces DSLR's with pentaprisms throughout the line, in a small an APS-C form factor as possible. I like a meaty grip about the size of the K-x.
Has it occurred to anyone that perhaps some of Pentax's QC issues, as with AF on the K-5 , has something to do with trying to compact the design too much? I suspect so. It took 4 generations of Pentax DSLR's to get an AF assist lamp in the whole line.