maluka
|
2010-01-27
, 00:52
|
|
Posts: 741 |
Thanked: 900 times |
Joined on Nov 2007
@ Auckland NZ
|
#191
|
The Following User Says Thank You to maluka For This Useful Post: | ||
|
2010-01-27
, 01:34
|
|
Posts: 269 |
Thanked: 1,096 times |
Joined on Sep 2009
@ Hampshire, UK
|
#193
|
|
2010-01-27
, 03:26
|
|
Posts: 963 |
Thanked: 626 times |
Joined on Sep 2009
@ Connecticut, USA
|
#195
|
Cool, I will try that in a bit...
Here we are:
K850i set of photos at ISO 100,200,400
N900 set of photos at ISO 100,200,400
|
2010-01-27
, 03:45
|
Posts: 17 |
Thanked: 41 times |
Joined on Jan 2010
@ Davis, CA
|
#196
|
|
2010-01-27
, 04:49
|
|
Posts: 963 |
Thanked: 626 times |
Joined on Sep 2009
@ Connecticut, USA
|
#198
|
|
2010-01-27
, 09:54
|
|
Posts: 1,743 |
Thanked: 1,231 times |
Joined on Jul 2006
@ Twickenham, UK
|
#199
|
The N900 shot has much better white balance, but yes the noise is bad. Even worst, that picture shows banding which means that the sensor is not good in high ISO situations. Try the same shot setting the ISO manually (100, 200, and 400) instead of leaving it in Auto. Also, just for the sake of experiment, try taking the shots with and without a back cover. Let us know what you find.
There is a possiblity to experiment yourself by doing your own IPP (image post-processing element) and hooking it into camera application pipelines through /etc/gdigicam/*.conf -- see gdigicam source code for expected semantics of the element (and there is a need to switch off noise filtering in the omap3camd). There is also some additional work on documentation from our side but generally speaking, if someone wants to experiment with noise filters, it is possible
Interesting comparison. There are a couple of things that are made visible in these two images:
- The biggest difference in the noise levels is caused by the very different autoexposure settings in the two devices: the N900 has used an exposure time of 1/15s and an ISO speed of 800; while the SE device used 1/4s and ISO 250. Higher ISO means a higher analog gain in the camera sensor, which is the main reason for the noise.
The reason N900 uses those settings, is that using 1/4s exposure time when the device is held in the hand is almost certain to cause blur into the image due to either hand shake, or simply that the subject is moving (take an image of people, they won't hold still for 1/4s unless you really ask them to). With the exposure time of 1/15s there are much better chances of getting a sharp image (or at least significantly less blur).
The settings on N900 can be changed so that it will use exposure settings more similar to the SE device: If you can rest N900 e.g. on a table so that you know it will not move during exposure, you can set it to the night scene mode (which allows the camera to use long exposure times), and maybe even also set ISO manually to a low value.
- The horizontal banding is a hardware limitation in the camera sensor in N900. It's known from other products using the same sensor, and as far as I know, there are no hardware updates coming to this sensor. And we don't have a noise filter for filtering out such structured noise.
Tags |
camera, fremantle, gallery, image, maemo, maemo 5, n900, photo, photograph, photography, picture, show and tell |
|