UFRaw with experimental support for all non-Quattro Foveon cameras.

[nkbj]

Hi.

I have just added experimental support for X3F files for all Foveon cameras except the Quattro models to the UFRaw developement code which is here: http://ufraw.cvs.sourceforge.net/viewvc/ufraw/?view=tar

Please read the README file in the archive for information about build requirements.

Decoding works but color correction needs som work. The code in UFRaw is compatible with DCRaw, so it can be moved to that program with little effort.

I big thank you must go to Tam Merlant ("tam-ille") and Roland Karlsson for their efforts in finding the secrets of the X3F file format. Without their work I would not have got to this point.

Please test the code and add your comments, suggestions, patches etc. to this thread.

--
Regards,
Niels Kristian Bech Jensen

 

[nkbj]

I have searched the net for X3F files to test the changes, and I have found files from every camera model from the Polaroid x530 to the Sigma DP2 Quattro except the Sigma DP2S.

Can anybody point med to a place where I can download such a file?

--
Regards,
Niels Kristian Bech Jensen

 

[nkbj]

Got a sample file now. Thanks.

 

[motomanDK]

Hej Niels Kristian

i have a QPcard203 and both SD14 and SD15 - i can make RAW's for you if that could help.

I order to generate camera profiles the QP card have an application that make a graphics presentation of the color diviation - maybe that could help to.







--
/Henrik - Silkeborg - Denmark
(SD14 and 2 * SD15 ... and about 5 kg glass)

 

[Astrotripper]

First of all, that you for your work on this.

I took it for a ride on few of my shots. As it is now, I'd say it's good for monochrome conversions. With all the features of UFRaw, there's plenty of wiggle room to get the B&W one likes. At least that's my impression.

Here's an example mono conversion from UFRaw:



And here's one from SPP 6.2

Note that I didn't really try to get the same result. Also, the SPP image is a bit sharper in the corners I think, maybe that's because SPP does some lens corrections (chromatic aberration maybe?).

Anyway, for B&W, I will be using UFRaw from now on. That will save me the hassle of rebooting to Windows. And it is so much faster than SPP.

Getting the right colors will probably be tricky, but I think Sigma published a chart showing sensitivity of each layer to different wavelengths, but that might have been for Quattro (I wonder if they are the same for all Foveon variants). Maybe that will be helpful.

Also, interesting how the photos at ISOs other than the base look like. I loaded up photo shot at ISO400, and it was very dark, had to crank up exposure comp by around 2 stops. Does that mean that Merrills are isoless?

 

[nkbj]

Hej Henrik.

Thank you for your offer. I might take you up on it at a later date.

Currently it is the SD1 and Merrill cameras that give me most trouble with some ugly color casts.

 

[nkbj]

Also, interesting how the photos at ISOs other than the base look like. I loaded up photo shot at ISO400, and it was very dark, had to crank up exposure comp by around 2 stops. Does that mean that Merrills are isoless?

I have seen the same thing but I have not investigated it further yet.

 

[xpatUSA]

Hi.

I have just added experimental support for X3F files for all Foveon cameras except the Quattro models to the UFRaw developement code which is here: http://ufraw.cvs.sourceforge.net/viewvc/ufraw/?view=tar

I am confused. I already have a version of UFRaw which opens my x530/SD9/10/14 raw files quite well.

Do I need to download this version too?

 

[Scottelly]

That's AWESOME news Niels! Thank you for your work toward making Sigma raw files more accessible to the World.

 

[nkbj]

I am confused. I already have a version of UFRaw which opens my x530/SD9/10/14 raw files quite well.

Do I need to download this version too?

The new version is work in progress to add a similar level of support for cameras newer that the SD14. If you do not use such cameras you do not need to download this version.

 

[xpatUSA]

I am confused. I already have a version of UFRaw which opens my x530/SD9/10/14 raw files quite well.

Do I need to download this version too?

The new version is work in progress to add a similar level of support for cameras newer that the SD14. If you do not use such cameras you do not need to download this version.

Hi Neils, thank you for the clarification.

 

[Roland Karlsson]

Interesting stuff!

We at the Kalpanika compay have come rather far in image quality for all cameras except the Polaroid and Quattro. It is (unfortunately) not open source - but I think we can show some images soon on the forum.

So far we need to do quite heavy noise reduction - so the conversion is not fast. We still have some color artefacts - but we have found that SPP also have the same - they probably use the same nose reduction. So - we do not worry there.

 

[nkbj]

Interesting stuff!

We at the Kalpanika compay have come rather far in image quality for all cameras except the Polaroid and Quattro. It is (unfortunately) not open source - but I think we can show some images soon on the forum.

One problem with the Quattro models are, that the X3F files does not include a PROP section. To get the WB and color mode settings I had to parse the makernote tags in the embedded JPEG thumbnail. I am currently working on code to implement that in UFRaw.

 

[Roland Karlsson]

One problem with the Quattro models are, that the X3F files does not include a PROP section. To get the WB and color mode settings I had to parse the makernote tags in the embedded JPEG thumbnail. I am currently working on code to implement that in UFRaw.

Yes, that is annoying. All other cameras have it. Why did they remove that?

There is an integer CAMF setting where each integer corresponds to a white balance. I reverse engineered the numbers by testing on my DP2Q. I do not have the data hera at my work, but I think the CAMF setting was called WhiteBalance.

 

[borissimo86]

It works with my SD1 after making a gamma-matrix profile with a colorchecker. I'm not sure yet, but so far my impression is that there might be less color blotching and less green noise when I run files through UFRaw than with Sigma's SPP. More red chroma noise, but it is fine-grained and easier to remove with regular noise removal software.

The biggest problem is exposure. Increasing it creates a magenta cast, decreasing it creates a green cast. This doesn't happen in SPP, only UFRaw. I don't know if exposure is connected to white balance but it seems to be affecting the color channels...

I won't complain about color shading (green/cyan corners) because this is a problem that Sigma solves in software with lens profiles, and it is something that varies from lens to lens. I don't believe that there is one approach that solves the issue with all lenses. The best way, in my opinion, to deal with them in post with flat-field software like CornerFix. Either way, Sigma files from the Merrill cameras will require extra work in processing but it is good to have another choice beside SPP and Iridient Developer (which is MAC-only).

 

[xpatUSA]

It works with my SD1 after making a gamma-matrix profile with a colorchecker. I'm not sure yet, but so far my impression is that there might be less color blotching and less green noise when I run files through UFRaw than with Sigma's SPP. More red chroma noise, but it is fine-grained and easier to remove with regular noise removal software.

That is also my experience with the simple RGB rendering out of RawDigger.

 

[nkbj]

There is an integer CAMF setting where each integer corresponds to a white balance. I reverse engineered the numbers by testing on my DP2Q. I do not have the data hera at my work, but I think the CAMF setting was called WhiteBalance.

I have found it. It is called WhiteBalance and has the value 1 (Auto?) in all the sample files I have.

 

[nkbj]

It works with my SD1 after making a gamma-matrix profile with a colorchecker. I'm not sure yet, but so far my impression is that there might be less color blotching and less green noise when I run files through UFRaw than with Sigma's SPP. More red chroma noise, but it is fine-grained and easier to remove with regular noise removal software.

The biggest problem is exposure. Increasing it creates a magenta cast, decreasing it creates a green cast. This doesn't happen in SPP, only UFRaw. I don't know if exposure is connected to white balance but it seems to be affecting the color channels...

I won't complain about color shading (green/cyan corners) because this is a problem that Sigma solves in software with lens profiles, and it is something that varies from lens to lens. I don't believe that there is one approach that solves the issue with all lenses. The best way, in my opinion, to deal with them in post with flat-field software like CornerFix. Either way, Sigma files from the Merrill cameras will require extra work in processing but it is good to have another choice beside SPP and Iridient Developer (which is MAC-only).

The exposure problem might be due to some of the corrections I have not found and implemented yet. I hope to find some time during the weekend to dig a little deeper.