It all started about five years ago. We wanted to do optical zoom
We looked at folded optics for example, we looked at straight optics; various different configurations, and we couldn't make any of those work well enough.
Later on, when we were thinking about this problem again,
This is where, when you are using the default 5 MP setting, we take up to 7 pixels, take the information that you want from those pixels, and process that using some Nokia algorithms.
GSMArena: Why has no one else produced a 41MP sensor? Is it that expensive?
It's not cheap, but when you compare it to optical zoom, you would have to put more cost into the optics.
There are not many people that can manufacture optics at that level of precision, and you need optics, otherwise you just can't make it work.
it is more complicated than binning. We looked at a number of different methods how we could do this, and some methods preserve more detail
We haven't really made that comparison. What I can say is, depending on the interpolation and downsampling you use in Photoshop, it may be possible that you'll get similar performance.
In Photoshop, you typically might get a sharpness that looks higher, but you might get more noise
The companion chip takes the information directly from the sensor and scales it, before that data goes to the GPU.
We do have a digital stabilizer for video, which we are still tuning at the moment.
Carl Zeiss, yes. The [manufacturing] precision is ten times greater than with SLR [lenses]. There are five lenses, quite small at the top, and then get progressively larger as they approach the sensor. There's also a mechanical ND filter which can be controlled in Creative mode, similar to the N8.
In addition to the precision of the optics, one of the things we've been working on for some time is precision alignment. We have a live image sensor setup when we assemble the optics, so that each and every single module we make is built and then assembled with live feedback from the sensor to ensure proper alignment.
We co-developed the sensor.
GSMArena: Any word on who else is involved?
Unfortunately, not at this time.
[ eds: His boss at the press conference revealed it is Toshiba
]
GSMArena: Are you going to share the technology?
No.
The problem you may have experienced using that algorithm, is that it couldn't adapt quickly enough to sudden bursts of sound.
Whereas with this capability, rather than being limited to a certain level, it's always able to resolve this much higher dynamic range, in terms of volume.
Typically you would need an external microphone to get similar performance capability. In addition to that, the frequency response range that we have is almost the same as CD
We're still optimizing the camera; we don't have the final hardware yet. We're still developing the algorithms—particularly low-light. We still have plenty of work still to do in low-light. While the low-light images are significantly better than on other devices, we believe we can make them even better.