[Digikam-devel] [Bug 283681] After re-activating several collections, a huge number of raw pics cannot be opened
Axel Krebs
axel.krebs at t-online.de
Wed Nov 2 21:02:29 GMT 2011
https://bugs.kde.org/show_bug.cgi?id=283681
--- Comment #11 from Axel Krebs <axel krebs t-online de> 2011-11-02 21:02:28 ---
(In reply to comment #10)
> I have received two sample pictures off-line from Axel:
>
> Pic 1: DSC_4789.NEF - (original) pic, contains one tag "Tier"
> Pic 2: DSC_4789_plus_new_tag.NEF - immediately after inserting the additional
> new tag "Belege"
>
> Quick observations:
>
> Pic 1 is not straight from the camera, it has been modified by digiKam and
> contains XMP tags (incl. tag Xmp.tiff.Software set to "digiKam-2.0.0-rc"),
> presumably added by exiv2. Thus it seems that exiv2 _can_ in general modify
> your NEFs without destroying them. It has the following tags with the value
> "Tier":
>
> $ exiv2 -pa -u DSC_4789.NEF | grep -e Tier -e Belege
> Iptc.Application2.Keywords String 4 Tier
> Xmp.dc.subject XmpBag 1 Tier
> Xmp.digiKam.TagsList XmpSeq 1 Tier
> Xmp.MicrosoftPhoto.LastKeywordXMP XmpSeq 1 Tier
> Xmp.lr.hierarchicalSubject XmpBag 1 Tier
>
>
> Pic 2 has a totally crippled structure. IFD0 is missing TIFF tags (that's what
> GIMP is complaining about), sub-image2 went missing, the makernote is no longer
> there and almost all Exif tags have disappeared. It has the following tags with the value "Tier" or "Belege":
>
> $ exiv2 -pa -u DSC_4789_plus_new_tag.NEF | grep -e Tier -e Belege
> Iptc.Application2.Keywords String 6 Belege
> Iptc.Application2.Keywords String 4 Tier
> Xmp.digiKam.TagsList XmpSeq 2 Tier, Belege
> Xmp.dc.subject XmpBag 2 Belege, Tier
> Xmp.MicrosoftPhoto.LastKeywordXMP XmpSeq 2 Tier, Belege
> Xmp.lr.hierarchicalSubject XmpBag 2 Tier, Belege
>
>
> To verify if the problem is due to exiv2, it's best to reproduce the problem
> with just the exiv2 tool. However, I can easily add "Belege" as above to Pic 1
> without corrupting it with the following command:
>
> $ cat add-belege.cmd
> add Iptc.Application2.Keywords Belege
> set Xmp.digiKam.TagsList Belege
> set Xmp.dc.subject Belege
> set Xmp.MicrosoftPhoto.LastKeywordXMP Belege
> reg lr http://ns.adobe.com/lightroom/1.0/
> set Xmp.lr.hierarchicalSubject Belege
>
> $ exiv2 -v -m add-belege.cmd DSC_4789.NEF
>
> The result is an intact picture with the "Belege" tags as in Pic 2 (except for
> the order in Xmp.dc.subject):
>
> exiv2 -pa -u DSC_4789.NEF | grep -e Tier -e Belege
> Iptc.Application2.Keywords String 4 Tier
> Iptc.Application2.Keywords String 6 Belege
> Xmp.dc.subject XmpBag 2 Tier, Belege
> Xmp.digiKam.TagsList XmpSeq 2 Tier, Belege
> Xmp.MicrosoftPhoto.LastKeywordXMP XmpSeq 2 Tier, Belege
> Xmp.lr.hierarchicalSubject XmpBag 2 Tier, Belege
>
>
> So this doesn't look like an exiv2 issue. Please provide the steps to reproduce
> it with just exiv2 (a reproducer program or the command line tool), if you
> think it is.
>
> Andreas
---
Hi Andreas:
thank you checking the matter. Sorry for ansewering late- my workday starts at
7 a.m. and end ~9 p.m.
I provided two pics according to Gilles` suggestion.
Pic 1: DSC_4789.NEF - (original) pic, contains one tag "Tier". I took this as
appearantly "unchanged", as I never had problems with tags in NEF in earlier
digiKam version before.
I do not understand, what 4 "tags" should mean: 4 independant tag with the same
content "tier" (maybe not really important?)
Pic 2: DSC_4789_plus_new_tag.NEF - immediately after inserting the _additional_
new tag "Belege". I copied that pic, added the mentioned additional tag- and
got problems. So the _basic_ pic content is identical. You call it "totally
crippled structure".
And this make me wandering: as both pics are definitely the same, I do not
understand, wher this tagf-mixture should come from.
As common sense indicates, certainly not from an additional treatment in my
camera.
Maybe it helps you understand my regular workflow:
- Taking the picture.
- Basic setting NEF plus jpg, usually DIN 100 sensitivity.
- copying the pics from SD-Card to PC
- starting digiKam, sometimes starting "search for new pictures"
- then immediatelly tagging all new pics for some tag-types as "Tiere",
"Freunde", etc.
- there is only a flat hierarchical structure on two levels.
- as I developed the tag-structure over some couple of years, the very tags
stay stable and will not be changed later.
- I showed my recent database statistics
- Before the recent digiKam version, I had
- the same settings- and did _NOT_ experience problems.
- I had the same camera (NIKON D60)- no changes
For me, there seem to appear _two_ changes:
1.) new digiKam version
2.) as described recently, I desized my digiKam database (from 145000 pic to
7800 pics) and included only relevant branches (=directories) I assumed to be
necessary to prepare a selection project.
As I mentioned before, too, digiKams thumbnail database would keep its size
even after several booting procedures over three weeks or so. If self cleaning
feature within databse would have worked prooperly, the thumbnail database
shouöl have decreased from 6,7 GB (for 145000 pic) to 360 kB size at 7800 pic.
This is certainly not correct, or?
You mention:
...
Xmp.digiKam.TagsList XmpSeq 2 Tier, Belege
Xmp.dc.subject XmpBag 2 Belege, Tier
Xmp.MicrosoftPhoto.LastKeywordXMP XmpSeq 2 Tier, Belege
...
Why the sequence of tags changes?
Is there a sequence problem involved?
Why are there three more lines than in pic 1?
You suggest to test exif 2 , if writing tags correctly. Since last digiKam
update I run KUBUNTU, and semantic desktop may run too.
How do you exclude cross-relationships from that side? By checking the "naked
exif2 tool"?
You ask me for the steps to reproduce. It sounds to me a like asking a bird for
how a plane does function... I am npo programmer. And I do not know, which
informations are relevant for a programmer. Thats why I am not really sure, how
I can support you.
Please let me know.
Thank you-
Axel
--
Configure bugmail: https://bugs.kde.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the Digikam-devel
mailing list