Originally posted by rvannatta the next step
being an electronic shake reduction (as opposed to mechanical).
wrt electronic SR ...
This is theoretically feasible but in order to rival the quality of a mechanical SR (be it sensor shift or lens shift) a key innovation is required:
- sensor-embedded pixel registers.
This key innovation comes after sensor-embedded A/D converters like the K-5 has. But it still is rather expensive; e.g., the K-5 would require a hefty 32 MB (64 MB to really allow for new applications) on the sensor chip. E.g., current L1 processor caches tend to be smaller. No CMOS sensor features this or is announced to do so any time soon.
Only then does it become possible to bin different pixels into one virtual pixel during exposure, a method required as long as an entire sensor can't be read out in submilliseconds (which it can't because it would require Terabit/s bandwidth).
Therefore, you're right theoretically. But it will be long before this idea becomes relevant.
btw, SW SR in current P&S etc. use slow readout and software to achieve some effect for very long exposures. But this doesn't compare to mechanical SR.
Originally posted by Gimbal One way of making the camera thinner would be to keep the mirror (but locked down) and put the SR and sensor assembly up where the penta prism is now. That would easily shave 10mm wouldn't it?
It saves more, esp. because half the mirror-box's volume can then be populated with electronics. It would actually be better to flip the mirror and have the sensor moved to the bottom and use the upper empty space to house the EVF and other stuff like flash and electronics. AFAIK, designs like this were discussed.
However, it is difficult to find a mirror which doesn't negatively impact the image. Silver-coated glass or similiar may do it and must be extremely planar. I guess it is doable but maybe not the cheapest way. Pentamirror VFs are darger than pentaprism VFs for a reason ...
But we should keep this kind of design in mind for future mirrorless K mount cameras.
Originally posted by wjjstu What is the point of using K-mount if the new "K" lenses won't work on old SLR bodies and old K-lenses won't be optimal on the new mirrorless bodies?
I think it is perfectly reasonable to do this. The bulk of lenses would be cross compatible but a few special ones would exploit the extra possibilities (esp. UWA and travel lenses).
Anyway, the same kind of argument applied to the first DA lenses when released ...
Originally posted by devorama They explained it as thus:
...
Lighter elements means less inertia, which means they can change position more quickly.
I am aware of this issue. CDAF may require more acceleration to aquire focus in the same time. But there are a number of arguments to consider here:
- Pentax PDAF does some extra focus steps too, very often at least.
- A screw drive motor can be made stronger if in the body.
- But foremost, current CDAF algorithms are poor: they are bad in predicting the distance to real focus, they overshoot to easily and they stop the lens in order to take a new measurement. A sensor which delivers an HD videosignal frame in 17ms can be read out while the lens travels (it takes about 10x the time to travel to focus). A good algorithm would slow down the lens softly until it aquires exact focus right then it eventually stops.
I've written an CDAF algorithm myself and I am not impressed by how current CDAF algorithms behave.
Originally posted by Gimbal A quick mockup of a thinner EVIL, still with SR and fully compatible K-mount
cf. above.