P: "There was an error parsing the file" when attempting to convert SRW images fromSamsung models

[boardhead_]

DNG converter says "There was an error parsing the file" when attempting to convert SRW images from some Samsung models after metadata has been edited with ExifTool.  Here are 3 sample images with this problem:

http://130.15.24.88/~phil/tmp/SamsungNX2000.srw
http://130.15.24.88/~phil/tmp/SamsungNX300.srw
http://130.15.24.88/~phil/tmp/SamsungNX30.srw

Also, the following image is not recognized by the converter after a similar edit with ExifTool:

http://130.15.24.88/~phil/tmp/SamsungEK-GN120.srw

[ccastleb]

Hi Phil and Steve, 

I can reproduce this. We will investigate and follow up.

Thanks,

- Chris

[boardhead_]

Also, it is worth noting that this problem does not occur for SRW images from the following models: EX1, EX2F, NX1, NX10, NX100, NX11, NX20, NX200, NX3000, NX500, NXmini and WB2000

- Phil

[boardhead_]

Thanks for your response.  Here are the original files:

http://130.15.24.88/~phil/tmp/SamsungNX2000_orig.srw
http://130.15.24.88/~phil/tmp/SamsungNX300_orig.srw
http://130.15.24.88/~phil/tmp/SamsungNX30_orig.srw
http://130.15.24.88/~phil/tmp/SamsungEK-GN120_orig.srw

And the ExifTool command was:

exiftool -artist=Phil FILE

where FILE is any of the original files.

- Phil

[ssprengel]

These particular raw files also don't work in Lightroom or the Adobe Camera Raw plug-in for Photoshop.

That ExifTool can be told to do things to a raw file that make it unrecognizable to Adobe products isn't necessarily a bug in either set of programs.

For diagnostic purposes, it would be useful to have a pre-ExifTool version of the same raw files, as well as the ExifTool command line that was used to edit the raw files' metadata, so the situation can be replicated by others.

[boardhead_]

Wow.  That's just wrong.  I suggest you ignore what Samsung is telling you and decode the maker notes properly.

[ccastleb]

Hi Phil,

Another bit of information from my colleague.

"That's not what we were told by Samsung.
The private MakerNote really is a binary blob and they gave us specific offsets to use to extract their data.
We've never had any problems with Samsung firmware updates.
External tools should never rearrange private MakerNote data for this reason."

[boardhead_]

Hi Chris,

Thanks for replying.

Yes, your colleague is correct, but the maker notes are stored in a TIFF IFD format.  The information is all still there if the IFD is parsed properly.  You can't just look at the maker notes as a black box of binary data if you expect to extract information consistently.  If you do this, then your code will easily break with camera firmware updates (among other things).

[ccastleb]

Hi Phil,

Following is a note from my colleague who has been investigating this issue.

"...the original sample file shows MakerNote tag length is 268182  bytes.  When I run exiftool on the sample file, the result has MakerNote tag length 268100, which is 82 bytes shorter.  Furthermore, when I diff the contents of the MakerNote tags (before/after), I see very different values."

- Chris

[boardhead_]

Sorry for the delay.  Yes.  ExifTool rewrites the MakerNote tag.  It uses a standard TIFF format so the information may move around when rewritten, but all of the same information is there.

[ccastleb]

Hi Phil,

Apologies for the delay.

As far as we have been able to determine, exiftool appears to be modifying the private MakerNote tag which causes Camera Raw to reject the files.

Can you confirm whether exiftool is inadvertently changing the MakerNote tag?

Thanks,

- Chris

[boardhead_]

Any News?

[ccastleb]

Bug 4158567 for your reference.