[Brightcolours]

My initial reaction was the same, soLong. Why waste so many pixels?


Well, I just read on the Canon site that each pixel consists of two diodes, which are read out seperately for AF, but combined for the image. So, there you have it, all "40 mps" fully available


Smart approach, IMO. Will be interesting to see how it performs in real life. That's certainly a nice attempt to increase tracking performance in Live View mode. An at the same time cover a lot more of the image with AF sensors.


If this works reliably, the logical next step would be SLT...


-- Markus

What is logical about SLT? It loses you the optical viewfinder, brings heat problems to the camera, and loses 1/2 stop of light. The last point may not be a big issue, but the other two are rather nasty. 

ALso, SLT was needed to bring PD-AF to live view/video (according to Sony).

Now live view has its own PD-AF, so exactly why is SLT a next logical step?

[Brightcolours]

....and i know much much less than you2, and am clueless about keeping to a thread.....but the 7D2 should make those 40mps available as separate little things for lo-iso shots and then pair them for those dark places - no?

There are not really 40 available on any level... Two green (or red, or blue) elements next to eachother are not any different from 1 element.. Will not bring more resolution/detail. And if that were different, it would mean rectangular pixels...

 

So no, it is a 20.x mp sensor, which just happens to divide each pixel into two halves for AF purposes.

[Klaus]

What is logical about SLT? It loses you the optical viewfinder, brings heat problems to the camera, and loses 1/2 stop of light. The last point may not be a big issue, but the other two are rather nasty. 

ALso, SLT was needed to bring PD-AF to live view/video (according to Sony).

Now live view has its own PD-AF, so exactly why is SLT a next logical step?

 

I think Markus tried to point towards a totally mirrorless "DSLR". Which is what Sony is doing next. 

 

Klaus

[Guest]

But will the dang thing actually be accurate ?

[Brightcolours]

I think Markus tried to point towards a totally mirrorless "DSLR". Which is what Sony is doing next. 

 

Klaus

Ah right. I am with you now. Well, for me personally I do not long for a through-the-lens camera without optical viewfinder. 

[Klaus]

But will the dang thing actually be accurate ?

 

It is more accurate than conventional PD-AF.

It is not as good in low light though. At least to date.

[soLong]

....Well, each pixel consists of two diodes, which are read out seperately for AF, but combined for the image. So, there you have it, all "40 mps" fully available  ....If this works reliably, the logical next step would be SLT...


-- Markus

 

<p style="font-size:12px;font-family:Helvetica;">….SLT? - oh, isn't that the hot new religion that's out there on a crusade trawling the world for converts right now….offering a bet each way: 20k to be good good good and 20k for the fun bits

[popo]

You can imagine an different SLT arrangement using this Canon sensor or similar. The fixed mirror can provide an OVF, leaving the main sensor for AF. With a bit of trickery you could possibly even have a hybrid VF like Fuji and switch between OVF/EVF if you like.

As for the potential resolution if the sub-pixels were available, it could and would lead to an increase in resolution potential in whatever direction the pixels are split in. This was a question I saw raised, are the pixels all split in the same direction? Implicitly it may only be sensitive in one direction. Or could/would they have split the pixels in a mix of two axis? The biggest problem would be any debayer routine will be seriously difficult to deal with. Probably worse than the Fuji pattern...

Edit: I just had pointed out elsewhere the split sensors are under the microlens, thus further muddling up what possible info can be extracted from it.