[Clipart] Sorting images in the open clipart library

Mon Sep 10 04:15:58 PDT 2012

On 10 September 2012 11:01, Jakub Jankiewicz <jcubic at onet.pl> wrote:
> We will have download all clipart from the site soon, the last one, I
> plan to have it back when we get the new site online.
>
> Metadata is stuff that it's in svg file and it's for the developers
> (it's the same info that is on every detail page), we just make it
> simply to use the collection, I can write the code that will sort them
> based on metadata, when we will put it into a file.
>
> and by using find, grep and sed on linux means that it will be very
> easy to create a tool that will do this. Maybe even some graphic
> search/browser for offline use.
>
> I can write something like this but maybe someone else will do it, it
> can have the same functions as a site, and when you double click it
> will open file in Inkscape.
>
> Sorting it by hand is a waste of times, we have computers to do this
> kind of things.

Well yes, if it gets done so it is in a form an ordinary end-user can
use for the specific task they need :-) I haven't found the titles on
images necessarily logical from a search point of view and to me
looking at them and deciding what to use is more than just an
automated process though I agree some automation helps. Depends, I
guess, on what a lot of individuals actually put in the meta data or
the title.

At this point in time I couldn't find a way to use what was on the
OpenClipart.org site in a way that fitted my needs for a particular
task and I don't have time to learn how to automate sorting based on
svg meta data when I'm not even sure it will work out as it depends on
exactly what is in the metadata.  Maybe I'm just unusual ;-)

What I'm looking at for Apache OpenOffice is to define a set of
gallery themes that could be used incrementally. eg just a few of the
best images for the main distribution in a standardised set of
galleries - maybe 50 - and then an extras set that could be entirely
on-line linked to openclipart.org or added to the galleries to suit
the user. That would enable the user to either add to the galleries
and work entirely off-line or the could link to on-line sources. These
are only initial considerations at the moment. Another possibility
would be to completely redesign the gallery part of AOO to integrate
seamlessly with the open clipart.org database but I think that is less
likely to happen at least in the immediate future.

 If OpenClipart.org has a list of standard classifications for
grouping images that will be searchable in the metadata then that
would be useful to know.

Thanks for the info.

>
> On Mon, 10 Sep 2012 10:35:09 +0100
> Ian Lynch <ianrlynch at gmail.com> wrote:
>
>> On 10 September 2012 01:51, S.Kemter
>> <buergermeister at karl-tux-stadt.de> wrote:
>> > Am Mon, 10 Sep 2012 02:40:17 +0200
>> > schrieb Jakub Jankiewicz <jcubic at onet.pl>:
>> >
>> >> It's great but it seems that you use package that it's few years
>> >> old, the last one (which is snapshot on osuosl server
>> >> download.openclipart.org) have more then 700MB and it's 2 year old.
>> >> So I assume that when we get back packages, (We want to pack them
>> >> every month they will be openclipart-svg.<YEAR>.<MONTH>.tar.gz) it
>> >> will have more the 1GB of data, when compress.
>> >>
>> >> We will also sync our database data with svg metadata (like tags
>> >> title and description) and maybe we will be able to fix their
>> >> names. Some of them are digits (I will now talking bad about Aiki
>> >> here) We can recreate names based on titles.
>> >>
>> >> So they will be searchable, using Unix tools like grep and find or
>> >> will allow to create scripts that will parse svg matadata.
>> >>
>> >> We can also sort them by tags and create symlinks (they will not be
>> >> recreated on windows) like
>> >>
>> >> real files:
>> >> /people/...
>> >>
>> >> symlinks:
>> >> /tags/
>> >>
>> >> old packages is just dump of the folder.
>> >
>> > yes thats the point, maybe also for OOo/LibreOffice would be a good
>> > solution to make an import from ocal, like Inkscape has.
>> > But I would make that function better before doing that also for
>> > them. There are some problems with
>> >
>> >
>> > br gnokii
>>
>> Ok, thanks for the info. I just went to the openclipart web site and
>> found what looked like the most comprehensive collection to download.
>> 2.0.
>>
>> Most people do not know anything about metafiles, tags, grep etc.
>> Probably there are better ways of sorting but if you don't know what
>> they are they are not much use to you. I simply needed a collection of
>> folders with related images in each for an EU project.
>>
>> I'm currently downloading the 756 meg file which I see has 3 gig of
>> data in it. So I can up date what I have so far. How easy/difficult
>> would it be to put a prominent link on the home page to enable
>> download of the entire most up to date collection to download all of
>> it? It was not obvious to me how to get the most up to date version of
>> all the images. Maybe I'm just unusual in wanting it ;-) I can be
>> selective in getting my definitive groupings. 3,000 images of an
>> electric guitar can be reduced to a few.
>>
>> I know that some of the Linux distros include the library in the
>> Gallery in themes but looking at LibreO the gallery has themes by user
>> contributor rather than by picture type and for AOO the OpenClipart
>> library isn't included with the standard download. For Windows users
>> there is nothing more than the standard AOO images. So if I have
>> images sorted into types it is not too hard to import a whole group
>> into a gallery theme in AOO. To start with just some instructions and
>> do it manually. There are some problems to overcome like AOO/LibO not
>> fully supporting svg. Also a 1 Gb+  library is not easy to incorporate
>> with an automatic download and linking to the library on-line will not
>> be satisfactory for everyone. Maybe having a subset of the most
>> popular images with the basic distribution with options to add to it
>> or link on-line. Or link on-line and automatically download and sort
>> images into relevant gallery themes. I don't know how feasible this
>> is. So all I'm trying to do is provide some options that might be
>> helpful to some people while better more permanent solutions can be
>> considered and implemented.
>>
>> On my own web site we use a subset of the open clipart library for
>> Drupal user accounts where students can provide evidence for their IT
>> qualifications and have some grouped images directly linked to their
>> accounts. That is now a bit out of date so I'll probably update it
>> once I have sorted out the error of using the old archive ;-)
>>
>> If anyone has any suggestions or ideas in general please say. I'm
>> willing to learn but as is always the case time is always limited ;-)
>>
>
> --
> Jakub Jankiewicz, Web Developer
> http://jcubic.pl
>
> _______________________________________________
> clipart mailing list
> clipart at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/clipart
>

-- 
Ian

Ofqual Accredited IT Qualifications (The Schools ITQ)

www.theINGOTs.org +44 (0)1827 305940

The Learning Machine Limited, Reg Office, 36 Ashby Road, Tamworth,
Staffordshire, B79 8AQ. Reg No: 05560797, Registered in England and
Wales.