FCAM thing in a nutshell for non-developers...


Okay, well, everyone is talking about how stupid this news is. Lemme see if I can break it down to make this pill a little easier to swallow.

What is it?
It is some developer tools to make things much easier to talk to, and process stuff taken from, the nokia N900 camera.

How does that help?

When I was a young C# programmer in college, My programming teacher used to give us homework. One classic piece of homework, was how to write to a file. here is how in C#

//make file stream object
FileStream fs = new FileStream("c:\\WriteFileStuff.txt", FileMode.OpenOrCreate, FileAccess.Write);

//make stream writer
StreamWriter sw = new StreamWriter(fs);

//write lines
sw.WriteLine("Howdy World.");

//close object
sw.Close();


not being expert programmers yet, our first tries were a disaster at best. We'd leave out a semi colon and get compiler errors, we didnt put the lines in the right order (causing things to be undefined when they really were, just further down), we would forget to close() the object at the end, making the file "currently in use" to the operating system forever, untill we restarted the machine.


eventually, a couple weeks later, the teacher taught us this:

System.IO.File.WriteAllText("C:\\WriteFileStuff.tx t","Howdy world");


... that did everything those 4 lines above did. but this one line could do all: the setup and tear down, for us! this meant syntax errors (like forgotten semicolons) and logic errors (like forgetting to close the stream at the end) would never happen again!


This is what API's do. they let you talk to stuff, in a way where either it wasn't possible before, or it was possible, it just took a sh*tload of code to do. and of course, with more code means more room for bugs.


Okay, well what does this "magical" API do then?

Well, it does 2 things. first it does some extra stuff with the camera that would normally take much more code to do and second it sets up the ability to talk to a gyroscope. I personally believe the gyroscope stuff isn't gonna happen, but the camera stuff is still pretty impressive. Here is the list of what exactly the new API allows. (gyroscope or not)

Rephotography
You take a picture in one place, and it can figure out the exact spot you took it in, and you can return days later and it will still know based off of key points in the image.


IMU based Lucky Imaging
The gyroscope can detect what is your shaky hand, and what is actually moving objects in the photo, and makes proper adjustments for that to prevent blurry images.



Foveal Imaging
The ability to downsample the rest of a video, and make a certain center point in the video sharpest. Bring the quality of everything down except for a certain spot, in that spot, bring the quality up. here's an exerpt:

Downsampling and cropping on the sensor is a capability of the
Aptina sensor in the F2 not exposed by the base API.

As you can see, "cropping" the image sensor is not expose by the normal programming APIs, so this API adds this functionality.

it also allows tracking motion in just one particular area of the camera.



HDR Imaging
This takes 3 pictures, of varying brightness, and merges the 3, to make an image that looks acceptable while leading lots of play for the lighting and movement problems that most cameras suffer from.



Low Light Imagine
for low light images, you can increase the gain (but that makes it grainy), or take a longer exposure (but that makes it blurry from movement). the Fcam API takes 2 pictures: one at high gain, and one at longer exposure. It then merges the two together to make a pretty nice shot



Panorama capture
Panorama. but this is processed as the picture is being taken, and does not require a wait time for the image to process. secondly, with each "shot" it takes a regular exposure, and longer exposure shot, then merges the two, then after it is all finished they are "stiched" together.






as you can see, there is some potential for some kool stuff if people put their minds to it. hope this helps some people figure out how kool this thing can really be.

silly question thats is great but.. step by step install?

http://fcam.garage.maemo.org/

Follow the instructions to install fcamera, hdrcapture and lowlight applications (fcamera from extras-testing and hdrcapture and lowlight from extras-devel).

Is this built on top of the camera driver (i.e. read processed data) or does it read raw data like blessn900?

sony123,

You can actually get raw data through the standard camera driver, if you ask it nicely enough.

In any case, FCam certainly lets you get at the raw sensor data - otherwise there'd be no way we could save DNGs.

how to use Rephotography? I already installed fcamera, hdr and lowlight.

HDR seems to work pretty nicely!

This is great news an thanks that you broke this down for the ney sayers. This will make a good camera even better. Just hope they make it possible in konjunction with Titans Enhanced kernel.

thanks for either not reading or not understanding.

This is working great, I'll show some sample images in a bit. But I'm having trouble with the DNG images that are saved by FCamera.

The DNGs have a really heavy cyan tint on them when I view them in Adobe Bridge, which is not removable no matter what settings I use in Camera Raw. This is Adobe CS5 btw.

When I try to view them in Irfan View they display as a red garbled mess.

When I check the option to save as JPG as well, these images save absolutely fine.

Anyone got any ideas why the DNGs don't display properly on the PC?

Here's an HDR image I made earlier, merged in Photoshop CS5. http://the-buccaneer.co.uk/image_edi...1279809203.jpg

EDIT:

Here's a photo captured in FCamera - This is the automatic JPG exported image http://the-buccaneer.co.uk/image_edi...1279809705.jpg

And this is the same image but the DNG generated by FCamera as seen in Adobe Photoshop CS5 http://the-buccaneer.co.uk/image_edi...1279809766.jpg