12 mp resolution with signal processing

[chromis]

Fuji claims that you can achieve a 12 megapixel resolution wit hsignal processing software.

What software are they referring to and how do you use it?

[Tom V]

I may know what you are asking about.

Since the camera makes 12 megapixel images internally, I am sure the software you refer to is what the Fuji camera has built-in, in order to process the data from the sensor. You use it when you use the camera in any mode other than RAW. The RAW files are saved to memory without any processing.

[amazingthailand]

Hi All,

This brings up an important question: RAW files.

Is not the basic concept of the raw file that it is straight from the image sensor and has not been touched by human, or non-human, hands?

The S2 has a 6 megapixel sensor, yet the raw file is a 12 megapixel file, therefor the image HAS been manipulated and is not straight from the sensor.

If we, the S2 owners, want to work with the REAL 6 megapixel data we are out of luck as Fuji seems to have decided that all output will come from their up-processed data.

Not really sure I like this. I would much prefer it if the up-processing was a menu option and not forced on us.

Just thinking out load.
Declan

[Tom V]

Y'all,

The megapixel number indicates how many pixels are in the image.

The megabyte (mb) number indicates the file size.

"Mega" means million

The S2 records 4256 x 2848 pixels, which is 12121088 pixels, which is 12.12 megapixels. This makes a 34.7mb image, when processed as an 8-bit image, or, when processed as a 16-bit image (using the RAW file converter EX) makes a file about 70 mb.

The RAW file size is about 12mb, and it can be processed various ways, up to a maximum pixel size of 12.12 megapixel, and resulting in files up to 70mb.

More later...

[amazingthailand]

Okay Tom,

Maybe I am misinterpreting things here, but according to this web site:

http://www.bythom.com/s2d100.htm

The Maximum REAL Resolution of the S2 is 3024x2016

They also go on to state that 'in camera interpolation to 12 megapixels...'

I think I have seen this kind of statement elsewhere also. The arguement seems to revolve around the octagonol shape of the pixels, which does not allow them to line up in nice neat horizontal and vertical rows, as square (or rectangular) pixels would.

Anyway, no real complaints about the camera, just curious about the whys and the wherefors

Cheers,
Declan

[Topngu]

Tom V,
You are 100% right,you have very good knowleged about S2!

[amazingthailand]

Hi Tom,

I think you might have misunderstood what I was trying to say previously.

I am a programmer, so I fully understaqnd the relationship between image size, format, color depth and compression on the size of the file on the disk.

I was refering to the image size x by y pixels as displayable on the screen.

The S2 manual clearly states that raw format files are available only in the 4256x2848 image size. This is 12 megapixels, interpolated up from the 6 megapixel SCCD.

Everything I have been able to discern about this, seems to point to the octagonal design of the SCCD, which packs the picture elements differently from standard CCD's.

Fuji does the interpolation in the carmera, regardless of file size/format selected. So all images are interpolated up to 12 megapixels, then it is down interpolated to the desired size.

I don't have any complaints about the sharpness, but it is something of an anomoly that even raw is interpolated up to 12 megapixels.

It is something I would like to discuss with an engineer with Fuji. Just to satisfy my curiousity.

Cheers.
Declan

[memobug]

If you check out the schema of the SuperCCD, you'll see that the sensors are still arranged in a grid, it's simply oriented at 45 degrees from the normal row & column arrangement.

http://www.image-acquire.com/news/99/q4/fujiccd.jpg

Now if you look this small arrangement of 32 cells with a "O" marking the position of the sensors, and consider it as if it were a standard 8x8 array with every other row and column missing:

- O - O - O - O
O - O - O - O -
- O - O - O - O
O - O - O - O -
- O - O - O - O
O - O - O - O -
- O - O - O - O
O - O - O - O -

You can see that it would be pretty logical to return the data as 8 rows and 8 columns, using the actual data where you have it "O", and interpolating the data for the missing sites "-" using the surrounding pixels. This may explain why the raw format consists of 12 million points when the measured data is 6 million.

Of course that is just my review of the situation and the Fuji engineer, if you ever corner him may well tell a different story.

Regards,

Matt

[Tom V]

Declan, et al,

I agree on the above conclusions.

I think some of the confusion comes from the count of how many pixels the Fuji sensor contains, and how many pixels are recorded.

The actual number so diodes is not the number of pixels you end up with once the camera or RAW File Converter LE/EX get done hammering on the data.

Fuji says that the RAW file is data taken directly from the sensor, and all processing is done later. To me, that means the the file has not been bumped up to 12mp. Currently, the only methods available for conversion DO create 12mp files, but maybe RAW files can be made into other size files.

I would venture to say that there may be software engineer that could make another converter that could do something else with the RAW data. Maybe it is possible to interpolate the RAW data into more or less than a 12mp file, or into different color spaces (direct to SWOP CMYK?), or to use different algorithms for improved sharpness (no more halos!) or improved conversion speed. Who knows what the limits are for converting the RAW data. I think it is great that Fuji left us the option of doing the conversion later, when more might be possible.

[PaulN]

Just a few thoughts on the way a digital camera processes it's image. When the manufacturers state that a raw file is 'straight off the CCD' they both are and are not correct. They are right in saying that the file is unmanipulated by all the software that's used to impress the user settings on it, but they are wrong to say that it's completely untouched. Each photosite can only record light, not color, so there's a grid (Bayer Pattern) that's placed over the sensors so that each pixel can record only one color. A diode (pixel) is an analog device, so when a picture is taken the signel is fed into an analog/digital converter, then it gets interpolated into a smooth picture. If it didn't, it would look like a mosaic with each pixel recording it's own discrete dot of colored light. So all digital cameras interpolate to get even a raw file. The Fuji Super CCD has the photosites arrayed differently, as was explained in a previous post. This allows them to make 6 mpx. into a 12 mpx. picture. Now, the Fuji CCD only works in the 12 mpx. mode. If you select 3024, the camera downsamples, so if you use the raw mode, naturally you will get a 12 mpx. file. This file is the virgin file that has only been converted to a digital signal and smoothed. The shooting settings you've set in the camera are recorded in the file header for use with the processing software if you instruct it to use some or all of them, or if you use LE. With EX, you can tell it which to use and which to ignore. Some third party systems, like Bibble, and hopefully MacBibble in their latest version also write their settings to the file header so that if you process the file again later you can duplicat your previous efforts.