Notices

Originally Posted by ETalvala

The N900's sensor isn't as flexible as we'd like, for something like this.

It has a set of predefined zoom modes, which are various mixes of on-sensor crop and on-sensor binning. (As an example, when you run it in the regular video recording mode, the sensor doesn't crop at all, and averages pixels down to 1296x974. Then the OMAP3 imaging pipe resizes that down to what's actually stored). There's probably a zoom mode for getting 1:1 pixels off the center, but since we targeted FCam more for photography (Where cropping in post is usually not a big deal) than video, we didn't really investigate supporting those modes.

So right now, FCam doesn't let you specify digital zoom modes, and you always get the full frame, optionally reduced in resolution at the sensor itself.

-Eddy, one of the FCam devs

Originally Posted by benny1967

The idea is great (although I don't know if it's technically possible - there's probably a reason why current implementations don't do it this way) and I wouldn't see why one should change the thread title, as it says exactly what the OP asks for. Technical implementations and APIs don't matter at all, the result does.

@TiagoTiago:
You might also consider opening a brainstorm for this. Not that brainstorms are the most useful tool on maemo.org, but you never know - and your proposal just is the typical brainstorm use-case.

Originally Posted by ETalvala

TiagoTiago,

That can work on the image sensor in our F2 prototype (the same sensor the N95 has, incidentally) - it's more flexible in terms of region-of-interest control. The N900 Toshiba sensor isn't as flexible.