Fuji sensor dump

[ejmartin]

In what way is the data "more RAW" that dcraw document mode?

Because in dcraw document mode (-D -j) we have several steps of
postprocessing:

• Fuji S5 data is stretched two times in long side direction to

2150x1440 pixels
each pixel become two-component (not one as in original bayer). S2
data is
not stretched because real sensor has right aspect ratio (2944x2144)

• Image extracted 45-degree rotated (for easy interpolation on

demosaic stage)

• Black level subtracted
• Either in document mode (-D) values are scaled to fit full range
• If file was stretched on the first step (S3 and S5 files are

stretched), then one half of pixel values is simpy dropped in
'document mode'

• If no -j ( NO fuji rotate) option is given, image is rotated back

with bi-liniear upsize to 3000x2000 pix

• For 8-bit output gamma correction is performed

So, result is not RAW , just postprocessed less (than full color
processing): resized, scaled, one half of values dropped...

--

Thank you for a concise, informative and polite response.

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[astralux]

Take the time to look at the source code of both, or look at LibRAW
changelog, or post your question to LibRAW site. DPR is hardly the
place to discuss raw data acquisition modules.

It would be the simplest matter for the people who wrote the code and
are familiar with both it and dcraw to answer the question directly.
If DPR is not the place to discuss RAW data acquisition modules, then
why are you continually generating threads about them?

--
emil

Hey, knock off the 3rd degree, the guy is sharing some extremely useful information and he answered your question well. He is doing this work because he has a passion for what he does, so lay off him.

A question for the OP, is there some documentation available that shows the RAF file format (from Fuji I would doubt it )or did you guys unravel it yourself?

 

[FujiColy]

Hi Alex,

Thanks for your work on this software, it has given me some new ideas for my own converter.

Colin

In what way is the data "more RAW" that dcraw document mode?

Because in dcraw document mode (-D -j) we have several steps of
postprocessing:

• Fuji S5 data is stretched two times in long side direction to

2150x1440 pixels
each pixel become two-component (not one as in original bayer). S2
data is
not stretched because real sensor has right aspect ratio (2944x2144)

• Image extracted 45-degree rotated (for easy interpolation on

demosaic stage)

• Black level subtracted
• Either in document mode (-D) values are scaled to fit full range
• If file was stretched on the first step (S3 and S5 files are

stretched), then one half of pixel values is simpy dropped in
'document mode'

• If no -j ( NO fuji rotate) option is given, image is rotated back

with bi-liniear upsize to 3000x2000 pix

• For 8-bit output gamma correction is performed

So, result is not RAW , just postprocessed less (than full color
processing): resized, scaled, one half of values dropped...

--
Alex Tutubalin
http://www.libraw.org

 

[ejmartin]

Take the time to look at the source code of both, or look at LibRAW
changelog, or post your question to LibRAW site. DPR is hardly the
place to discuss raw data acquisition modules.

It would be the simplest matter for the people who wrote the code and
are familiar with both it and dcraw to answer the question directly.
If DPR is not the place to discuss RAW data acquisition modules, then
why are you continually generating threads about them?

--
emil

Hey, knock off the 3rd degree, the guy is sharing some extremely
useful information and he answered your question well. He is doing
this work because he has a passion for what he does, so lay off him.

Look, I asked a simple question about what was done. The response I got was basically, go read the source code and figure it out for yourself. The guy is indeed sharing some useful information, and I was just trying to find out what it was; why is that giving him the 3rd degree? If in the first instance Iliah had spent half the energy answering my question that he did telling me to buzz off, we all would have got the information and he would have saved himself some time.

As it stands, Iliah's collaborator in their project answered my question quite nicely, and I do thank him for both his informative response and its civil tone.

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[FujiColy]

This is the work of Alex, and welcome it is to the Fuji community.

Thanks Alex, and we look forward to future posts by you, your blog is great by the way!

Colin

 

[BobsYourUncle]

this discussion is so far over my head it is gratifying to know there are those out there who have use for the OP's work! Ilah, I feel like an intellectual maggot when I read your posts here and the Nikon Forum.... but I'm glad you do them.
I suppose resistance to Ilah Borg's comments is futile-
Keep up the good work even though I have no need for a data dump.

Bob

 

[Iliah Borg]

is there some documentation available that
shows the RAF file format (from Fuji I would doubt it )or did you
guys unravel it yourself?

Not that I know of from Fuji. The idea behind LibRAW 0.7 branch is to do extensive testing and strip off as much as possible of raw processing dcraw code, that is to allow to output "as much raw as possible".

This is especially important not only for those of us who design and code raw converters, but also for those who perform analysis of raw data to determine noise / dynamic range, colour properties, resolution of sensors and cameras. Those are rather complicated matters, and for sure those who are really involved are interested to read the documentation and source code in a most thorough manner. We even dare to hope they can perform some additional testing and provide their suggestions and feedback

--
http://www.libraw.org/

 

[Alex Tutubalin]

thanks alex and welcome. could you please clarify this as well:
http://forums.dpreview.com/forums/read.asp?forum=1020&message=30682684

Replied in that thread.

--
Alex Tutubalin
http://www.libraw.org

 

[MiTaka]

Thanks Iliah, I took some time and played with the precompiled binaries. You are right that it is different than the possible options with dcraw, and I've learned a lot about LibRaw. Keep up the good work!

Is there a list of software that uses LibRaw, I'm mostly interested in analyze programs like Rawanalyze?

--
Dimitar Ivanov

 

[Alex Tutubalin]

Is there a list of software that uses LibRaw, I'm mostly interested
in analyze programs like Rawanalyze?

For the present time, LibRaw is used in KDE libkdcraw library (name is old and reflects time when dcraw binary was called).

So many KDE (Linux, but avaliable for Windows too) RAW processing tools, such as digiKam, Krita and some others uses LibRaw.

Also, FreeImage library authors are working on integration, so many apps which uses FreeImage will read RAW soon(er or later).

AFAIK, no Raw analyzers uses LibRaw yet. We plan to fix this problem
--
Alex Tutubalin
http://www.libraw.org

 

[miso kaco]

n/t